Sentiment_Classification_Projects-cn-Copy1.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 情感分类 & 建立神经网络时，应如何“搭建问题框架”\n",
    "\n",
    "### 大纲：\n",
    "\n",
    "- 简介： “搭建问题框架”的重要性\n",
    "\n",
    "- [管护数据集](#lesson_1)\n",
    "- [提出“预测理论”](#lesson_2)\n",
    "- [**项目 1**：快速验证理论](#project_1)\n",
    "\n",
    "\n",
    "- [将文本转换成数字](#lesson_3)\n",
    "- [**项目 2**：创建输入/输出数据](#project_2)\n",
    "\n",
    "\n",
    "- 融会贯通，创建神经网络 \n",
    "- [**项目 3**：创建神经网络](#project_3)\n",
    "\n",
    "\n",
    "- [理解神经噪声](#lesson_4)\n",
    "- [**项目 4**：通过降噪来提高网络学习速度](#project_4)\n",
    "\n",
    "\n",
    "- [分析网络的不足之处](#lesson_5)\n",
    "- [**项目 5**：提高网络训练和运行的速度](#project_5)\n",
    "\n",
    "\n",
    "- [进一步降噪](#lesson_6)\n",
    "- [**项目 6**：有策略地删减词汇，进行降噪](#project_6)\n",
    "\n",
    "\n",
    "- [分析：权重怎么了？](#lesson_7)\n",
    "\n",
    "# 课程：管护数据集<a id='lesson_1'></a>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "def pretty_print_review_and_label(i):\n",
    "    print(labels[i] + \"\\t:\\t\" + reviews[i][:80] + \"...\")\n",
    "\n",
    "g = open('reviews.txt','r') # What we know!\n",
    "reviews = list(map(lambda x:x[:-1],g.readlines())) #利用map把每行读出的变为一个列表，lambda 参数：表达式\n",
    "g.close()\n",
    "\n",
    "g = open('labels.txt','r') # What we WANT to know!\n",
    "labels = list(map(lambda x:x[:-1].upper(),g.readlines())) #upper大写\n",
    "g.close()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**注:** 对于正在使用的 `reviews.txt` 中的数据，我们已进行了一些预处理，现在数据只包含小写字符，但如果我们现在使用的是原始数据，因为不知道数据是否全为小写，所以此时应添加一步操作，进行大小写转换。处理同一个单词的不同形态时，如 `The`、 `the` 和 `THE`，也要进行同样的转换。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "25000"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(reviews)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'homelessness  or houselessness as george carlin stated  has been an issue for years but never a plan to help those on the street that were once considered human who did everything from going to school  work  or vote for the matter . most people think of the homeless as just a lost cause while worrying about things such as racism  the war on iraq  pressuring kids to succeed  technology  the elections  inflation  or worrying if they  ll be next to end up on the streets .  br    br   but what if you were given a bet to live on the streets for a month without the luxuries you once had from a home  the entertainment sets  a bathroom  pictures on the wall  a computer  and everything you once treasure to see what it  s like to be homeless  that is goddard bolt  s lesson .  br    br   mel brooks  who directs  who stars as bolt plays a rich man who has everything in the world until deciding to make a bet with a sissy rival  jeffery tambor  to see if he can live in the streets for thirty days without the luxuries if bolt succeeds  he can do what he wants with a future project of making more buildings . the bet  s on where bolt is thrown on the street with a bracelet on his leg to monitor his every move where he can  t step off the sidewalk . he  s given the nickname pepto by a vagrant after it  s written on his forehead where bolt meets other characters including a woman by the name of molly  lesley ann warren  an ex  dancer who got divorce before losing her home  and her pals sailor  howard morris  and fumes  teddy wilson  who are already used to the streets . they  re survivors . bolt isn  t . he  s not used to reaching mutual agreements like he once did when being rich where it  s fight or flight  kill or be killed .  br    br   while the love connection between molly and bolt wasn  t necessary to plot  i found  life stinks  to be one of mel brooks  observant films where prior to being a comedy  it shows a tender side compared to his slapstick work such as blazing saddles  young frankenstein  or spaceballs for the matter  to show what it  s like having something valuable before losing it the next day or on the other hand making a stupid bet like all rich people do when they don  t know what to do with their money . maybe they should give it to the homeless instead of using it like monopoly money .  br    br   or maybe this film will inspire you to help others .  '"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "reviews[2]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'POSITIVE'"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "labels[2]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 课程: 提出预测理论<a id='lesson_2'></a>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "labels.txt \t : \t reviews.txt\n",
      "\n",
      "NEGATIVE\t:\tthis movie is terrible but it has some good effects .  ...\n",
      "POSITIVE\t:\tadrian pasdar is excellent is this film . he makes a fascinating woman .  ...\n",
      "NEGATIVE\t:\tcomment this movie is impossible . is terrible  very improbable  bad interpretat...\n",
      "POSITIVE\t:\texcellent episode movie ala pulp fiction .  days   suicides . it doesnt get more...\n",
      "NEGATIVE\t:\tif you haven  t seen this  it  s terrible . it is pure trash . i saw this about ...\n",
      "POSITIVE\t:\tthis schiffer guy is a real genius  the movie is of excellent quality and both e...\n"
     ]
    }
   ],
   "source": [
    "print(\"labels.txt \\t : \\t reviews.txt\\n\")\n",
    "pretty_print_review_and_label(2137)\n",
    "pretty_print_review_and_label(12816)\n",
    "pretty_print_review_and_label(6267)\n",
    "pretty_print_review_and_label(21934)\n",
    "pretty_print_review_and_label(5297)\n",
    "pretty_print_review_and_label(4998)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 项目 1: 快速验证理论<a id='project_1'></a>\n",
    "\n",
    "\n",
    "请善加利用 [计数器](https://docs.python.org/2/library/collections.html#collections.Counter) 类和 [numpy](https://docs.scipy.org/doc/numpy/reference/) 库。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "from collections import Counter\n",
    "import numpy as np"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "我们要创建三个 `计数器` 对象，其中一个为正面评价的单词计数，一个为负面评价的单词计数，还有一个为所有单词计数。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create three Counter objects to store positive, negative and total counts\n",
    "positive_counts = Counter()\n",
    "negative_counts = Counter()\n",
    "total_counts = Counter()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**待办：** 遍历所有评价，每遇到一个出现在正面评价里的单词，就为该单词的正面评价计数器和总计数器增量；同理，每遇到一个出现在负面评价里的单词，就为该单词的负面评价计数器和总计数器增量。\n",
    "\n",
    "**注:** 在这些项目中，请用 `split(' ')` 来将一段文本（比如一段评价）分割成独立的单词，如果你用的是 `split()`，那结果会和视频以及解决方案里的有所出入。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "# TODO: Loop over all the words in all the reviews and increment the counts in the appropriate counter objects\n",
    "for i in range(len(reviews)):\n",
    "    if labels[i] == 'POSITIVE':\n",
    "        for word in reviews[i].split(\" \"):\n",
    "            positive_counts[word] += 1\n",
    "            total_counts[word] += 1\n",
    "    else:\n",
    "         for word in reviews[i].split(\" \"):\n",
    "            negative_counts[word] += 1\n",
    "            total_counts[word] += 1"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "运行下方两个代码框，分别列出出现在正面评价和负面评价中的单词，并按出现频率由高到低排序。\n",
    "\n",
    "\n",
    "用counter创建为字典类型, 用most_common将数据从大到小排序，每组数据为一个元组"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Examine the counts of the most common words in positive reviews\n",
    "#print(positive_counts)\n",
    "#positive_counts.most_common()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Examine the counts of the most common words in negative reviews\n",
    "#negative_counts.most_common()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "你可以发现，如 “the” 等常见单词在正面评价和负面评价中出现的频率都很高，因此，你要找的其实不是正面评价或负面评价中最常见的单词，而是在正面评价里的出现频率比在负面评价里的高的单词，以及在负面评价里的出现频率比在正面评价里的高的单词，为此，你需要计算单词在正面评价与负面评价中出现的次数的 **比值**。  \n",
    "\n",
    "**待办：** 查看所有你见过的单词，计算各单词在正面评价和负面评价中出现的次数的比值，并将这些比值存入 `pos_neg_ratios`。\n",
    ">提示：计算给定单词的正面评价/负面评价比，可使用 `positive_counts[word] / float(negative_counts[word]+1)`。注意，分母要 `+1`——这样如果遇到只在正面评价中出现的单词，分母不会变为 0。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create Counter object to store positive/negative ratios\n",
    "pos_neg_ratios = Counter()\n",
    "# TODO: Calculate the ratios of positive and negative uses of the most common words\n",
    "#       Consider words to be \"common\" if they've been used at least 100 times \n",
    "for term, cnt in total_counts.most_common():\n",
    "    if cnt > 100:\n",
    "        pos_neg_ratio = positive_counts[term] / float(negative_counts[term] + 1)\n",
    "        pos_neg_ratios[term] = pos_neg_ratio"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "查看若干单词的比值计算结果："
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Pos-to-neg ratio for 'the' = 1.0607993145235326\n",
      "Pos-to-neg ratio for 'amazing' = 4.022813688212928\n",
      "Pos-to-neg ratio for 'terrible' = 0.17744252873563218\n"
     ]
    }
   ],
   "source": [
    "print(\"Pos-to-neg ratio for 'the' = {}\".format(pos_neg_ratios[\"the\"]))\n",
    "print(\"Pos-to-neg ratio for 'amazing' = {}\".format(pos_neg_ratios[\"amazing\"]))\n",
    "print(\"Pos-to-neg ratio for 'terrible' = {}\".format(pos_neg_ratios[\"terrible\"]))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "仔细观察刚才算出的比值，我们可以发现：\n",
    "\n",
    "* 那些你觉得在正面评价里会更常见的单词——如 \"amazing\" ——比值大于 1。单词在正面评价中出现得越频繁，其正面评价/负面评价比就越大于 1。\n",
    "* 那些你觉得在负面评价里会更常见的单词——如 \"terrible\" ——比值小于 1。单词在负面评价里出现得越频繁，其正面评价/负面评价比就越接近 0。\n",
    "* 中性词不含任何感情色彩，毕竟它们在各类评论中都有出现——如 \"the\" —— 比值非常接近 1。 一个完全中性的词，也即在正面评价和负面评价中出现次数完全相同的词，比值会无限接近于 1。虽然因为上文的建议，分母进行了 `+1` 处理，使得单词稍稍偏向负面，但因为这个偏置很小，而我们稍后会忽略太接近于中性的词，所以影响不大。\n",
    "\n",
    "好了，虽然比值能告诉我们哪个单词在正面评价/负面评价中出现得更频繁，但要处理我们算出的这些具体数值还是有点麻烦。褒义色彩十分浓厚的词，如 \"amazing\"，其比值大于4，而贬义色彩十分浓厚的词，如 \"terrible\"，其比值则在 0.18 左右，要比较这些数值并不容易，原因如下：\n",
    "\n",
    "* 现在我们视 1 为中性，但因为褒义色彩十分浓厚的词的正面评价/负面评价比的绝对值比贬义色彩十分浓厚的词的比的绝对值大，所以我们无法直接比较这两者，也就无从确定这两个单词所包含的褒义/贬义程度是否相同，因此，我们应以中性值为所有数值的中心，这样一来，就可以用单词的正面评价/负面评价比与中性值之差的绝对值来衡量单词蕴含了多少感情色彩（褒义/贬义）了。\n",
    "* 在比较绝对值时，以 0 为中心要比以 1 为中心容易。\n",
    "\n",
    "要解决上述问题，我们要对所有比值进行对数变换，从而得出新值。\n",
    "\n",
    "**待办：** 遍历你计算出来的所有比值，将其转化为对数。(也就是说，使用 `np.log(ratio)`)\n",
    "\n",
    "到最后，褒义色彩极为浓厚的单词的正面评价/负面评价比会与贬义色彩极为浓厚的单词的比的大小相似，但符号相反。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "# TODO: Convert ratios to logs、\n",
    "for word, ratio in pos_neg_ratios.most_common():\n",
    "    if ratio > 1 : \n",
    "        pos_neg_ratios[word] = np.log( ratio )\n",
    "    else:\n",
    "        pos_neg_ratios[word] = -np.log( 1 / (ratio + 0.01) )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "查看之前那些单词的新比值计算结果："
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Pos-to-neg ratio for 'the' = 0.05902269426102881\n",
      "Pos-to-neg ratio for 'amazing' = 1.3919815802404802\n",
      "Pos-to-neg ratio for 'terrible' = -1.6742829939664696\n"
     ]
    }
   ],
   "source": [
    "print(\"Pos-to-neg ratio for 'the' = {}\".format(pos_neg_ratios[\"the\"]))\n",
    "print(\"Pos-to-neg ratio for 'amazing' = {}\".format(pos_neg_ratios[\"amazing\"]))\n",
    "print(\"Pos-to-neg ratio for 'terrible' = {}\".format(pos_neg_ratios[\"terrible\"]))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "如果一切顺利，现在中性词的比值应该接近于 0。此时，\"the\" 接近 0 但仍是个十分小的正数，所以这个单词可能在正面评价中出现的次数比在负面评价中的多，但我们来看 \"amazing\"，其比值大于 `1`，意味着这是个明确的褒义词，而 \"terrible\" 的比值与其相似，但符号不同，也就是小于 `-1`，由此我们可以清楚地得知这两个单词具有特定浓度且相反的感情色彩。\n",
    "\n",
    "现在，运行下方代码框，观察更多比值。\n",
    "\n",
    "第一个代码框会列出所有单词，并按单词与正面评价的相关性排序。（你的 notebook 很可能会对输出进行删减，所以你不会真的看到列表里的*所有*单词。）\n",
    "\n",
    "第二个代码框会列出 30 个与负面评价关系最紧密的单词——首先代码会反转第一张列表的顺序，然后查找前 30 个单词。（如果想让第二个代码框列出所有单词并按单词与负面评价的相关性排序，你可以写 `reversed(pos_neg_ratios.most_common())`。）\n",
    "\n",
    "你接下来看到的数值会和我们之前看到的类似——中性词的比值接近 `0`；单词在正面评价里出现得越频繁，其比值就越接近甚至大于 `1`；单词在负面评价里出现得越频繁，其比值就越接近甚至小于 `-1`。这就是我们用对数而非原始比值的原因所在。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[('edie', 4.6913478822291435),\n",
       " ('paulie', 4.07753744390572),\n",
       " ('felix', 3.152736022363656),\n",
       " ('polanski', 2.8233610476132043),\n",
       " ('matthau', 2.80672172860924),\n",
       " ('victoria', 2.681021528714291),\n",
       " ('mildred', 2.6026896854443837),\n",
       " ('gandhi', 2.538973871058276),\n",
       " ('flawless', 2.451005098112319),\n",
       " ('superbly', 2.26002547857525),\n",
       " ('perfection', 2.159484249353372),\n",
       " ('astaire', 2.1400661634962708),\n",
       " ('captures', 2.038619547159581),\n",
       " ('voight', 2.030170492673053),\n",
       " ('wonderfully', 2.0218960560332353),\n",
       " ('powell', 1.978345424808467),\n",
       " ('brosnan', 1.9547990964725592),\n",
       " ('lily', 1.9203768470501485),\n",
       " ('bakshi', 1.9029851043382795),\n",
       " ('lincoln', 1.9014583864844796),\n",
       " ('refreshing', 1.8551812956655511),\n",
       " ('breathtaking', 1.8481124057791867),\n",
       " ('bourne', 1.8478489358790986),\n",
       " ('lemmon', 1.8458266904983307),\n",
       " ('delightful', 1.8002701588959635),\n",
       " ('flynn', 1.7996646487351682),\n",
       " ('andrews', 1.7764919970972666),\n",
       " ('homer', 1.7692866133759964),\n",
       " ('beautifully', 1.7626953362841438),\n",
       " ('soccer', 1.7578579175523736)]"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# words most frequently seen in a review with a \"POSITIVE\" label\n",
    "pos_neg_ratios.most_common(30)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[('boll', -4.07781526027089),\n",
       " ('uwe', -3.921875301871158),\n",
       " ('seagal', -3.320250105858192),\n",
       " ('unwatchable', -3.0269848170580955),\n",
       " ('stinker', -2.9876839403711624),\n",
       " ('mst', -2.775383321170797),\n",
       " ('incoherent', -2.7641396677532537),\n",
       " ('unfunny', -2.5545257844967644),\n",
       " ('waste', -2.4907515123361046),\n",
       " ('blah', -2.4475792789485005),\n",
       " ('horrid', -2.371577964480997),\n",
       " ('pointless', -2.345107387713634),\n",
       " ('atrocious', -2.3187369339642556),\n",
       " ('redeeming', -2.2667790015910296),\n",
       " ('prom', -2.2601040980178784),\n",
       " ('drivel', -2.247602958576693),\n",
       " ('lousy', -2.2118080125207054),\n",
       " ('worst', -2.1930856334332267),\n",
       " ('laughable', -2.172468615469592),\n",
       " ('awful', -2.1385076866397488),\n",
       " ('poorly', -2.132613384420701),\n",
       " ('wasting', -2.117815554561451),\n",
       " ('remotely', -2.111046881095167),\n",
       " ('existent', -2.0024805005437076),\n",
       " ('boredom', -1.9241486572738005),\n",
       " ('miserably', -1.921661093801999),\n",
       " ('sucks', -1.9166645809588516),\n",
       " ('uninspired', -1.9131499212248517),\n",
       " ('lame', -1.9117232884159072),\n",
       " ('insult', -1.908532376937626)]"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# words most frequently seen in a review with a \"NEGATIVE\" label\n",
    "list(reversed(pos_neg_ratios.most_common()))[0:30]\n",
    "\n",
    "# Note：: Above is the code Andrew uses in his solution video, \n",
    "#       so we've included it here to avoid confusion.\n",
    "#       If you explore the documentation for the Counter class, \n",
    "#       you will see you could also find the 30 least common\n",
    "#       words like this: pos_neg_ratios.most_common()[:-31:-1]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 项目 1 到此结束\n",
    "\n",
    "# 将文本转换成数字<a id='lesson_3'></a>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from IPython.display import Image\n",
    "\n",
    "review = \"This was a horrible, terrible movie.\"\n",
    "\n",
    "Image(filename='sentiment_network.png')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "review = \"The movie was excellent\"\n",
    "\n",
    "Image(filename='sentiment_network_pos.png')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 项目 2：创建输入/输出数据<a id='project_2'></a>\n",
    "\n",
    "**待办：** 创建一个名为 `vocab` 的[集合](https://docs.python.org/3/tutorial/datastructures.html#sets)，将词汇表的所有单词放入其中。\n",
    "\n",
    "利用字典keys()返回所有键值"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "# TODO: Create set named \"vocab\" containing all of the words from all of the reviews\n",
    "#print(total_counts.keys())\n",
    "vocab = set(total_counts.keys())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "运行下方代码框，查看词汇表大小，如果一切顺利，代码输出的结果将是 **74074**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "74074\n"
     ]
    }
   ],
   "source": [
    "vocab_size = len(vocab)\n",
    "print(vocab_size)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "观察下图，图中包括了你将在这个 notebook 里创建的神经网络的各层，其中，`layer_0` 是输入层，`layer_1`是隐藏层，`layer_2` 是输出层。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from IPython.display import Image\n",
    "Image(filename='sentiment_network_2.png')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![png](output_38_0.png)\n",
    "\n",
    "\n",
    "\n",
    "**待办：** 创建一个名为 `layer_0` 的 numpy 数组，将其所有元素初始化为 0。此处函数 [zeros](https://docs.scipy.org/doc/numpy/reference/generated/numpy.zeros.html) 可派上大用场。请确保你创建的 `layer_0` 是个二维矩阵，其行数为 1，列数为 `vocab_size` 的值。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "# TODO: Create layer_0 matrix with dimensions 1 by vocab_size, initially filled with zeros\n",
    "layer_0 = np.zeros((1,vocab_size))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "运行下方代码框，结果应为 `(1, 74074)`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1, 74074)"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "layer_0.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from IPython.display import Image\n",
    "Image(filename='sentiment_network.png')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "`layer_0` 包含了词汇表里的所有单词，一项代表词汇表里的一个词，如上图所示。为了确保我们知道每个单词的索引，请运行下方代码框，创建查找表并将各单词的索引存储其中。\n",
    "\n",
    "enumerate() 用于将一个可遍历的数据对象(如列表、元组或字符串)组合为一个索引序列，同时列出数据和数据下标，一般用在for循环里"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "scrolled": false
   },
   "outputs": [],
   "source": [
    "# Create a dictionary of words in the vocabulary mapped to index positions\n",
    "# (to be used in layer_0)\n",
    "word2index = {}\n",
    "for i,word in enumerate(vocab): \n",
    "    word2index[word] = i\n",
    "    \n",
    "# display the map of words to indices\n",
    "#word2index"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**待办:**  实现函数 `update_input_layer`，使该函数统计每个单词出现在给定评论里的次数，并将统计结果存入 `layer_0` 相应的索引下。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "def update_input_layer(review):\n",
    "    \"\"\" Modify the global layer_0 to represent the vector form of review.\n",
    "    The element at a given index of layer_0 should represent\n",
    "    how many times the given word occurs in the review.\n",
    "    Args:\n",
    "        review(string) - the string of the review\n",
    "    Returns:\n",
    "        None\n",
    "    \"\"\"\n",
    "    global layer_0\n",
    "    # clear out previous state by resetting the layer to be all 0s\n",
    "    layer_0 *= 0\n",
    "    for word in review.split(\" \"):\n",
    "        layer_0[0][word2index[word]] += 1\n",
    "    # TODO: count how many times each word is used in the given review and store the results in layer_0 "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "运行下方代码框，测试用第一条评论更新输入层的效果。此处所赋索引可能与解决方案中的不同，但你应该也能看到 `layer_0` 中出现了一些非零数值。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[18.,  0.,  0., ...,  0.,  0.,  0.]])"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "update_input_layer(reviews[0])\n",
    "layer_0"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**待办：** 实现函数 `get_target_for_labels`，使其根据给定标签是 `NEGATIVE` （负面）还是 `POSITIVE` （正面）返回 `0` 或 `1`。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "def get_target_for_label(label):\n",
    "    \"\"\"Convert a label to `0` or `1`.\n",
    "    Args:\n",
    "        label(string) - Either \"POSITIVE\" or \"NEGATIVE\".\n",
    "    Returns:\n",
    "        `0` or `1`.\n",
    "    \"\"\"\n",
    "    # TODO: Your code here\n",
    "    if(label == 'POSITIVE'):\n",
    "        return 1\n",
    "    else:\n",
    "        return 0"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "运行下方两个代码框，结果应分别输出 `'POSITIVE'` 和 `1`。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'POSITIVE'"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "labels[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "get_target_for_label(labels[0])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "运行下方两个代码框，结果应分别输出 `'NEGATIVE'` 和 `0`。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'NEGATIVE'"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "labels[1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "get_target_for_label(labels[1])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 项目 2 到此结束。\n",
    "\n",
    "# 项目 3：创建神经网络<a id='project_3'></a>\n",
    "\n",
    "**待办：** 此处使用了类 `SentimentNetork` 的框架：\n",
    "- 创建一个基本的神经网络，就像你在之前的课程和项目 1 中看到的一样，该网络应具备一个输入层、一个隐藏层和一个输出层。\n",
    "- 请**不要**在隐藏层中添加非线性元素，也就是说，请不要用激活函数计算隐藏层的输出。\n",
    "- 利用本 notebook 前面的代码来创建训练数据（请见代码中的 `TODO` 事项）\n",
    "- 实现函数 `pre_process_data`，为训练数据生成函数创建词汇表\n",
    "- 确保 `train` 会对整个语料库进行训练\n",
    "\n",
    "### 如有需要，请\n",
    "- 重看以前的优达学城课程\n",
    "- 查阅 [《深度学习详解》](https://www.manning.com/books/grokking-deep-learning)的第 3-5 章 - (请进入课室查看优惠码)\n",
    "\n",
    "集合的添加和删除add(), remove()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "import time\n",
    "import sys\n",
    "import numpy as np\n",
    "\n",
    "# Encapsulate our neural network in a class\n",
    "class SentimentNetwork:\n",
    "    def __init__(self, reviews,labels,hidden_nodes = 10, learning_rate = 0.1):\n",
    "        \"\"\"Create a SentimenNetwork with the given settings\n",
    "        Args:\n",
    "            reviews(list) - List of reviews used for training\n",
    "            labels(list) - List of POSITIVE/NEGATIVE labels associated with the given reviews\n",
    "            hidden_nodes(int) - Number of nodes to create in the hidden layer\n",
    "            learning_rate(float) - Learning rate to use while training\n",
    "        \n",
    "        \"\"\"\n",
    "        # Assign a seed to our random number generator to ensure we get\n",
    "        # reproducable results during development \n",
    "        np.random.seed(1)\n",
    "\n",
    "        # process the reviews and their associated labels so that everything\n",
    "        # is ready for training\n",
    "        self.pre_process_data(reviews, labels)\n",
    "        \n",
    "        # Build the network to have the number of hidden nodes and the learning rate that\n",
    "        # were passed into this initializer. Make the same number of input nodes as\n",
    "        # there are vocabulary words and create a single output node.\n",
    "        self.init_network(len(self.review_vocab),hidden_nodes, 1, learning_rate)\n",
    "\n",
    "    def pre_process_data(self, reviews, labels):\n",
    "        \n",
    "        # populate review_vocab with all of the words in the given reviews\n",
    "        review_vocab = set()\n",
    "        for review in reviews:\n",
    "            for word in review.split(\" \"):\n",
    "                review_vocab.add(word)\n",
    "\n",
    "        # Convert the vocabulary set to a list so we can access words via indices\n",
    "        self.review_vocab = list(review_vocab)\n",
    "        \n",
    "        # populate label_vocab with all of the words in the given labels.\n",
    "        label_vocab = set()\n",
    "        for label in labels:\n",
    "            label_vocab.add(label)\n",
    "        \n",
    "        # Convert the label vocabulary set to a list so we can access labels via indices\n",
    "        self.label_vocab = list(label_vocab)\n",
    "        \n",
    "        # Store the sizes of the review and label vocabularies.\n",
    "        self.review_vocab_size = len(self.review_vocab)\n",
    "        self.label_vocab_size = len(self.label_vocab)\n",
    "        \n",
    "        # Create a dictionary of words in the vocabulary mapped to index positions\n",
    "        self.word2index = {}\n",
    "        for i, word in enumerate(self.review_vocab):\n",
    "            self.word2index[word] = i\n",
    "        \n",
    "        # Create a dictionary of labels mapped to index positions\n",
    "        self.label2index = {}\n",
    "        for i, label in enumerate(self.label_vocab):\n",
    "            self.label2index[label] = i\n",
    "        \n",
    "    def init_network(self, input_nodes, hidden_nodes, output_nodes, learning_rate):\n",
    "        # Set number of nodes in input, hidden and output layers.\n",
    "        self.input_nodes = input_nodes\n",
    "        self.hidden_nodes = hidden_nodes\n",
    "        self.output_nodes = output_nodes\n",
    "\n",
    "        # Store the learning rate\n",
    "        self.learning_rate = learning_rate\n",
    "\n",
    "        # Initialize weights\n",
    "\n",
    "        # These are the weights between the input layer and the hidden layer.\n",
    "        self.weights_0_1 = np.zeros((self.input_nodes,self.hidden_nodes))\n",
    "    \n",
    "        # These are the weights between the hidden layer and the output layer.\n",
    "        self.weights_1_2 = np.random.normal(0.0, self.output_nodes**-0.5, \n",
    "                                                (self.hidden_nodes, self.output_nodes))\n",
    "        \n",
    "        # The input layer, a two-dimensional matrix with shape 1 x input_nodes\n",
    "        self.layer_0 = np.zeros((1,input_nodes))\n",
    "    \n",
    "    def update_input_layer(self,review):\n",
    "\n",
    "        # clear out previous state, reset the layer to be all 0s\n",
    "        self.layer_0 *= 0\n",
    "        \n",
    "        for word in review.split(\" \"):\n",
    "            # NOTE: This if-check was not in the version of this method created in Project 2,\n",
    "            #       and it appears in Andrew's Project 3 solution without explanation. \n",
    "            #       It simply ensures the word is actually a key in word2index before\n",
    "            #       accessing it, which is important because accessing an invalid key\n",
    "            #       with raise an exception in Python. This allows us to ignore unknown\n",
    "            #       words encountered in new reviews.\n",
    "            if(word in self.word2index.keys()):\n",
    "                self.layer_0[0][self.word2index[word]] += 1\n",
    "                \n",
    "    def get_target_for_label(self,label):\n",
    "        if(label == 'POSITIVE'):\n",
    "            return 1\n",
    "        else:\n",
    "            return 0\n",
    "        \n",
    "    def sigmoid(self,x):\n",
    "        return 1 / (1 + np.exp(-x))\n",
    "    \n",
    "    def sigmoid_output_2_derivative(self,output):\n",
    "        return output * (1 - output)\n",
    "    \n",
    "    def train(self, training_reviews, training_labels):\n",
    "        \n",
    "        # make sure out we have a matching number of reviews and labels\n",
    "        assert(len(training_reviews) == len(training_labels))\n",
    "        \n",
    "        # Keep track of correct predictions to display accuracy during training \n",
    "        correct_so_far = 0\n",
    "\n",
    "        # Remember when we started for printing time statistics\n",
    "        start = time.time()\n",
    "        \n",
    "        # loop through all the given reviews and run a forward and backward pass,\n",
    "        # updating weights for every item\n",
    "        for i in range(len(training_reviews)):\n",
    "            \n",
    "            # Get the next review and its correct label\n",
    "            review = training_reviews[i]\n",
    "            label = training_labels[i]\n",
    "            \n",
    "            #### Implement the forward pass here ####\n",
    "            ### Forward pass ###\n",
    "\n",
    "            # Input Layer\n",
    "            self.update_input_layer(review)\n",
    "\n",
    "            # Hidden layer\n",
    "            layer_1 = self.layer_0.dot(self.weights_0_1)\n",
    "\n",
    "            # Output layer\n",
    "            layer_2 = self.sigmoid(layer_1.dot(self.weights_1_2))\n",
    "            \n",
    "            #### Implement the backward pass here ####\n",
    "            ### Backward pass ###\n",
    "\n",
    "            # Output error\n",
    "            layer_2_error = layer_2 - self.get_target_for_label(label) # Output layer error is the difference between desired target and actual output.\n",
    "            layer_2_delta = layer_2_error * self.sigmoid_output_2_derivative(layer_2)\n",
    "\n",
    "            # Backpropagated error\n",
    "            layer_1_error = layer_2_delta.dot(self.weights_1_2.T) # errors propagated to the hidden layer\n",
    "            layer_1_delta = layer_1_error # hidden layer gradients - no nonlinearity so it's the same as the error\n",
    "\n",
    "            # Update the weights\n",
    "            self.weights_1_2 -= layer_1.T.dot(layer_2_delta) * self.learning_rate # update hidden-to-output weights with gradient descent step\n",
    "            self.weights_0_1 -= self.layer_0.T.dot(layer_1_delta) * self.learning_rate # update input-to-hidden weights with gradient descent step\n",
    "            \n",
    "            if(i == 1):\n",
    "                print(\"layer_0 shape = {}\\nlayer_1 shape = {}\\nlayer_2 shape = {}\\nlayer_2_delta shape = {}\\nlayer_1_delta shape = {}\\\n",
    "                    \\nweights_1_2 shape = {}\\n weights_0_1 shape = {}\".format(self.layer_0.shape, layer_1.shape, layer_2.shape, layer_2_delta.shape\\\n",
    "                                                                                             ,layer_1_delta.shape,self.weights_1_2.shape, self.weights_0_1.shape))\n",
    "            # Keep track of correct predictions.\n",
    "            if(layer_2 >= 0.5 and label == 'POSITIVE'):\n",
    "                correct_so_far += 1\n",
    "            elif(layer_2 < 0.5 and label == 'NEGATIVE'):\n",
    "                correct_so_far += 1\n",
    "            \n",
    "            # For debug purposes, print out our prediction accuracy and speed \n",
    "            # throughout the training process. \n",
    "            elapsed_time = float(time.time() - start)\n",
    "            reviews_per_second = i / elapsed_time if elapsed_time > 0 else 0\n",
    "            \n",
    "            sys.stdout.write(\"\\rProgress:\" + str(100 * i/float(len(training_reviews)))[:4] \\\n",
    "                             + \"% Speed(reviews/sec):\" + str(reviews_per_second)[0:5] \\\n",
    "                             + \" #Correct:\" + str(correct_so_far) + \" #Trained:\" + str(i+1) \\\n",
    "                             + \" Training Accuracy:\" + str(correct_so_far * 100 / float(i+1))[:4] + \"%\")\n",
    "            if(i % 2500 == 0):\n",
    "                print(\"\")\n",
    "    \n",
    "    def test(self, testing_reviews, testing_labels):\n",
    "        \"\"\"\n",
    "        Attempts to predict the labels for the given testing_reviews,\n",
    "        and uses the test_labels to calculate the accuracy of those predictions.\n",
    "        \"\"\"\n",
    "        \n",
    "        # keep track of how many correct predictions we make\n",
    "        correct = 0\n",
    "\n",
    "        # we'll time how many predictions per second we make\n",
    "        start = time.time()\n",
    "\n",
    "        # Loop through each of the given reviews and call run to predict\n",
    "        # its label. \n",
    "        for i in range(len(testing_reviews)):\n",
    "            pred = self.run(testing_reviews[i])\n",
    "            if(pred == testing_labels[i]):\n",
    "                correct += 1\n",
    "            \n",
    "            # For debug purposes, print out our prediction accuracy and speed \n",
    "            # throughout the prediction process. \n",
    "\n",
    "            elapsed_time = float(time.time() - start)\n",
    "            reviews_per_second = i / elapsed_time if elapsed_time > 0 else 0\n",
    "            \n",
    "            sys.stdout.write(\"\\rProgress:\" + str(100 * i/float(len(testing_reviews)))[:4] \\\n",
    "                             + \"% Speed(reviews/sec):\" + str(reviews_per_second)[0:5] \\\n",
    "                             + \" #Correct:\" + str(correct) + \" #Tested:\" + str(i+1) \\\n",
    "                             + \" Testing Accuracy:\" + str(correct * 100 / float(i+1))[:4] + \"%\")\n",
    "    \n",
    "    def run(self, review):\n",
    "        \"\"\"\n",
    "        Returns a POSITIVE or NEGATIVE prediction for the given review.\n",
    "        \"\"\"\n",
    "        # Run a forward pass through the network, like in the \"train\" function.\n",
    "        \n",
    "        # Input Layer\n",
    "        self.update_input_layer(review.lower())\n",
    "\n",
    "        # Hidden layer\n",
    "        layer_1 = self.layer_0.dot(self.weights_0_1)\n",
    "\n",
    "        # Output layer\n",
    "        layer_2 = self.sigmoid(layer_1.dot(self.weights_1_2))\n",
    "        \n",
    "        # Return POSITIVE for values above greater-than-or-equal-to 0.5 in the output layer;\n",
    "        # return NEGATIVE for other values\n",
    "        if(layer_2[0] >= 0.5):\n",
    "            return \"POSITIVE\"\n",
    "        else:\n",
    "            return \"NEGATIVE\"\n",
    "        "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "运行下方代码框，创建 `SentimentNetwork`，使之训练除最后一千条评论之外的所有评论（那一千条评论将用于测试），此处我们使用的学习率为 `0.1`。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "mlp = SentimentNetwork(reviews[:-1000],labels[:-1000], learning_rate=0.1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "运行下方代码框，用最后一千条评论测试网络的性能（即我们不用作训练集的那部分评论）。\n",
    "\n",
    "**我们还没有训练模型，所以模型只是单纯地在猜测，鉴于可选的值只有两个，因此准确率应该在   50% 左右。**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\r",
      "Progress:0.0% Speed(reviews/sec):0.0 #Correct:1 #Tested:1 Testing Accuracy:100.%\r",
      "Progress:0.1% Speed(reviews/sec):92.48 #Correct:1 #Tested:2 Testing Accuracy:50.0%\r",
      "Progress:0.2% Speed(reviews/sec):164.3 #Correct:2 #Tested:3 Testing Accuracy:66.6%\r",
      "Progress:0.3% Speed(reviews/sec):218.5 #Correct:2 #Tested:4 Testing Accuracy:50.0%\r",
      "Progress:0.4% Speed(reviews/sec):267.9 #Correct:3 #Tested:5 Testing Accuracy:60.0%\r",
      "Progress:0.5% Speed(reviews/sec):302.9 #Correct:3 #Tested:6 Testing Accuracy:50.0%\r",
      "Progress:0.6% Speed(reviews/sec):328.1 #Correct:4 #Tested:7 Testing Accuracy:57.1%\r",
      "Progress:0.7% Speed(reviews/sec):351.9 #Correct:4 #Tested:8 Testing Accuracy:50.0%\r",
      "Progress:0.8% Speed(reviews/sec):374.1 #Correct:5 #Tested:9 Testing Accuracy:55.5%\r",
      "Progress:0.9% Speed(reviews/sec):373.0 #Correct:5 #Tested:10 Testing Accuracy:50.0%\r",
      "Progress:1.0% Speed(reviews/sec):394.4 #Correct:6 #Tested:11 Testing Accuracy:54.5%\r",
      "Progress:1.1% Speed(reviews/sec):412.9 #Correct:6 #Tested:12 Testing Accuracy:50.0%\r",
      "Progress:1.2% Speed(reviews/sec):427.5 #Correct:7 #Tested:13 Testing Accuracy:53.8%\r",
      "Progress:1.3% Speed(reviews/sec):438.3 #Correct:7 #Tested:14 Testing Accuracy:50.0%\r",
      "Progress:1.4% Speed(reviews/sec):451.3 #Correct:8 #Tested:15 Testing Accuracy:53.3%\r",
      "Progress:1.5% Speed(reviews/sec):462.5 #Correct:8 #Tested:16 Testing Accuracy:50.0%\r",
      "Progress:1.6% Speed(reviews/sec):447.5 #Correct:9 #Tested:17 Testing Accuracy:52.9%\r",
      "Progress:1.7% Speed(reviews/sec):454.6 #Correct:9 #Tested:18 Testing Accuracy:50.0%\r",
      "Progress:1.8% Speed(reviews/sec):467.4 #Correct:10 #Tested:19 Testing Accuracy:52.6%\r",
      "Progress:1.9% Speed(reviews/sec):475.7 #Correct:10 #Tested:20 Testing Accuracy:50.0%\r",
      "Progress:2.0% Speed(reviews/sec):479.7 #Correct:11 #Tested:21 Testing Accuracy:52.3%\r",
      "Progress:2.1% Speed(reviews/sec):489.2 #Correct:11 #Tested:22 Testing Accuracy:50.0%\r",
      "Progress:2.2% Speed(reviews/sec):496.8 #Correct:12 #Tested:23 Testing Accuracy:52.1%\r",
      "Progress:2.3% Speed(reviews/sec):502.1 #Correct:12 #Tested:24 Testing Accuracy:50.0%\r",
      "Progress:2.4% Speed(reviews/sec):503.2 #Correct:13 #Tested:25 Testing Accuracy:52.0%\r",
      "Progress:2.5% Speed(reviews/sec):509.4 #Correct:13 #Tested:26 Testing Accuracy:50.0%\r",
      "Progress:2.6% Speed(reviews/sec):514.3 #Correct:14 #Tested:27 Testing Accuracy:51.8%\r",
      "Progress:2.7% Speed(reviews/sec):508.1 #Correct:14 #Tested:28 Testing Accuracy:50.0%\r",
      "Progress:2.8% Speed(reviews/sec):514.1 #Correct:15 #Tested:29 Testing Accuracy:51.7%\r",
      "Progress:2.9% Speed(reviews/sec):518.3 #Correct:15 #Tested:30 Testing Accuracy:50.0%\r",
      "Progress:3.0% Speed(reviews/sec):521.5 #Correct:16 #Tested:31 Testing Accuracy:51.6%\r",
      "Progress:3.1% Speed(reviews/sec):527.5 #Correct:16 #Tested:32 Testing Accuracy:50.0%\r",
      "Progress:3.2% Speed(reviews/sec):531.9 #Correct:17 #Tested:33 Testing Accuracy:51.5%\r",
      "Progress:3.3% Speed(reviews/sec):536.9 #Correct:17 #Tested:34 Testing Accuracy:50.0%\r",
      "Progress:3.4% Speed(reviews/sec):539.5 #Correct:18 #Tested:35 Testing Accuracy:51.4%\r",
      "Progress:3.5% Speed(reviews/sec):544.0 #Correct:18 #Tested:36 Testing Accuracy:50.0%\r",
      "Progress:3.6% Speed(reviews/sec):544.6 #Correct:19 #Tested:37 Testing Accuracy:51.3%\r",
      "Progress:3.7% Speed(reviews/sec):550.2 #Correct:19 #Tested:38 Testing Accuracy:50.0%\r",
      "Progress:3.8% Speed(reviews/sec):551.3 #Correct:20 #Tested:39 Testing Accuracy:51.2%\r",
      "Progress:3.9% Speed(reviews/sec):551.5 #Correct:20 #Tested:40 Testing Accuracy:50.0%\r",
      "Progress:4.0% Speed(reviews/sec):555.0 #Correct:21 #Tested:41 Testing Accuracy:51.2%\r",
      "Progress:4.1% Speed(reviews/sec):559.2 #Correct:21 #Tested:42 Testing Accuracy:50.0%\r",
      "Progress:4.2% Speed(reviews/sec):555.9 #Correct:22 #Tested:43 Testing Accuracy:51.1%\r",
      "Progress:4.3% Speed(reviews/sec):556.9 #Correct:22 #Tested:44 Testing Accuracy:50.0%\r",
      "Progress:4.4% Speed(reviews/sec):555.5 #Correct:23 #Tested:45 Testing Accuracy:51.1%\r",
      "Progress:4.5% Speed(reviews/sec):550.2 #Correct:23 #Tested:46 Testing Accuracy:50.0%\r",
      "Progress:4.6% Speed(reviews/sec):549.4 #Correct:24 #Tested:47 Testing Accuracy:51.0%\r",
      "Progress:4.7% Speed(reviews/sec):554.0 #Correct:24 #Tested:48 Testing Accuracy:50.0%\r",
      "Progress:4.8% Speed(reviews/sec):554.8 #Correct:25 #Tested:49 Testing Accuracy:51.0%\r",
      "Progress:4.9% Speed(reviews/sec):558.1 #Correct:25 #Tested:50 Testing Accuracy:50.0%\r",
      "Progress:5.0% Speed(reviews/sec):561.7 #Correct:26 #Tested:51 Testing Accuracy:50.9%\r",
      "Progress:5.1% Speed(reviews/sec):565.8 #Correct:26 #Tested:52 Testing Accuracy:50.0%\r",
      "Progress:5.2% Speed(reviews/sec):565.4 #Correct:27 #Tested:53 Testing Accuracy:50.9%\r",
      "Progress:5.3% Speed(reviews/sec):568.2 #Correct:27 #Tested:54 Testing Accuracy:50.0%\r",
      "Progress:5.4% Speed(reviews/sec):571.6 #Correct:28 #Tested:55 Testing Accuracy:50.9%\r",
      "Progress:5.5% Speed(reviews/sec):575.9 #Correct:28 #Tested:56 Testing Accuracy:50.0%\r",
      "Progress:5.6% Speed(reviews/sec):571.7 #Correct:29 #Tested:57 Testing Accuracy:50.8%\r",
      "Progress:5.7% Speed(reviews/sec):575.0 #Correct:29 #Tested:58 Testing Accuracy:50.0%\r",
      "Progress:5.8% Speed(reviews/sec):578.0 #Correct:30 #Tested:59 Testing Accuracy:50.8%\r",
      "Progress:5.9% Speed(reviews/sec):581.2 #Correct:30 #Tested:60 Testing Accuracy:50.0%\r",
      "Progress:6.0% Speed(reviews/sec):581.2 #Correct:31 #Tested:61 Testing Accuracy:50.8%\r",
      "Progress:6.1% Speed(reviews/sec):582.6 #Correct:31 #Tested:62 Testing Accuracy:50.0%\r",
      "Progress:6.2% Speed(reviews/sec):586.1 #Correct:32 #Tested:63 Testing Accuracy:50.7%\r",
      "Progress:6.3% Speed(reviews/sec):588.7 #Correct:32 #Tested:64 Testing Accuracy:50.0%\r",
      "Progress:6.4% Speed(reviews/sec):588.2 #Correct:33 #Tested:65 Testing Accuracy:50.7%\r",
      "Progress:6.5% Speed(reviews/sec):590.8 #Correct:33 #Tested:66 Testing Accuracy:50.0%\r",
      "Progress:6.6% Speed(reviews/sec):593.6 #Correct:34 #Tested:67 Testing Accuracy:50.7%\r",
      "Progress:6.7% Speed(reviews/sec):595.9 #Correct:34 #Tested:68 Testing Accuracy:50.0%\r",
      "Progress:6.8% Speed(reviews/sec):598.1 #Correct:35 #Tested:69 Testing Accuracy:50.7%\r",
      "Progress:6.9% Speed(reviews/sec):598.8 #Correct:35 #Tested:70 Testing Accuracy:50.0%\r",
      "Progress:7.0% Speed(reviews/sec):598.7 #Correct:36 #Tested:71 Testing Accuracy:50.7%\r",
      "Progress:7.1% Speed(reviews/sec):601.6 #Correct:36 #Tested:72 Testing Accuracy:50.0%\r",
      "Progress:7.2% Speed(reviews/sec):603.9 #Correct:37 #Tested:73 Testing Accuracy:50.6%\r",
      "Progress:7.3% Speed(reviews/sec):606.8 #Correct:37 #Tested:74 Testing Accuracy:50.0%\r",
      "Progress:7.4% Speed(reviews/sec):608.9 #Correct:38 #Tested:75 Testing Accuracy:50.6%\r",
      "Progress:7.5% Speed(reviews/sec):611.0 #Correct:38 #Tested:76 Testing Accuracy:50.0%\r",
      "Progress:7.6% Speed(reviews/sec):613.1 #Correct:39 #Tested:77 Testing Accuracy:50.6%\r",
      "Progress:7.7% Speed(reviews/sec):615.0 #Correct:39 #Tested:78 Testing Accuracy:50.0%\r",
      "Progress:7.8% Speed(reviews/sec):616.8 #Correct:40 #Tested:79 Testing Accuracy:50.6%\r",
      "Progress:7.9% Speed(reviews/sec):618.3 #Correct:40 #Tested:80 Testing Accuracy:50.0%\r",
      "Progress:8.0% Speed(reviews/sec):620.0 #Correct:41 #Tested:81 Testing Accuracy:50.6%\r",
      "Progress:8.1% Speed(reviews/sec):615.7 #Correct:41 #Tested:82 Testing Accuracy:50.0%\r",
      "Progress:8.2% Speed(reviews/sec):616.1 #Correct:42 #Tested:83 Testing Accuracy:50.6%\r",
      "Progress:8.3% Speed(reviews/sec):617.1 #Correct:42 #Tested:84 Testing Accuracy:50.0%\r",
      "Progress:8.4% Speed(reviews/sec):619.6 #Correct:43 #Tested:85 Testing Accuracy:50.5%\r",
      "Progress:8.5% Speed(reviews/sec):621.0 #Correct:43 #Tested:86 Testing Accuracy:50.0%\r",
      "Progress:8.6% Speed(reviews/sec):620.9 #Correct:44 #Tested:87 Testing Accuracy:50.5%\r",
      "Progress:8.7% Speed(reviews/sec):622.3 #Correct:44 #Tested:88 Testing Accuracy:50.0%\r",
      "Progress:8.8% Speed(reviews/sec):624.0 #Correct:45 #Tested:89 Testing Accuracy:50.5%\r",
      "Progress:8.9% Speed(reviews/sec):621.3 #Correct:45 #Tested:90 Testing Accuracy:50.0%\r",
      "Progress:9.0% Speed(reviews/sec):621.8 #Correct:46 #Tested:91 Testing Accuracy:50.5%\r",
      "Progress:9.1% Speed(reviews/sec):623.6 #Correct:46 #Tested:92 Testing Accuracy:50.0%\r",
      "Progress:9.2% Speed(reviews/sec):625.5 #Correct:47 #Tested:93 Testing Accuracy:50.5%\r",
      "Progress:9.3% Speed(reviews/sec):626.6 #Correct:47 #Tested:94 Testing Accuracy:50.0%\r",
      "Progress:9.4% Speed(reviews/sec):627.6 #Correct:48 #Tested:95 Testing Accuracy:50.5%\r",
      "Progress:9.5% Speed(reviews/sec):628.1 #Correct:48 #Tested:96 Testing Accuracy:50.0%\r",
      "Progress:9.6% Speed(reviews/sec):628.5 #Correct:49 #Tested:97 Testing Accuracy:50.5%\r",
      "Progress:9.7% Speed(reviews/sec):628.4 #Correct:49 #Tested:98 Testing Accuracy:50.0%\r",
      "Progress:9.8% Speed(reviews/sec):629.8 #Correct:50 #Tested:99 Testing Accuracy:50.5%\r",
      "Progress:9.9% Speed(reviews/sec):631.3 #Correct:50 #Tested:100 Testing Accuracy:50.0%\r",
      "Progress:10.0% Speed(reviews/sec):632.6 #Correct:51 #Tested:101 Testing Accuracy:50.4%\r",
      "Progress:10.1% Speed(reviews/sec):633.7 #Correct:51 #Tested:102 Testing Accuracy:50.0%\r",
      "Progress:10.2% Speed(reviews/sec):635.0 #Correct:52 #Tested:103 Testing Accuracy:50.4%\r",
      "Progress:10.3% Speed(reviews/sec):634.2 #Correct:52 #Tested:104 Testing Accuracy:50.0%\r",
      "Progress:10.4% Speed(reviews/sec):635.6 #Correct:53 #Tested:105 Testing Accuracy:50.4%\r",
      "Progress:10.5% Speed(reviews/sec):637.5 #Correct:53 #Tested:106 Testing Accuracy:50.0%\r",
      "Progress:10.6% Speed(reviews/sec):637.7 #Correct:54 #Tested:107 Testing Accuracy:50.4%\r",
      "Progress:10.7% Speed(reviews/sec):638.5 #Correct:54 #Tested:108 Testing Accuracy:50.0%\r",
      "Progress:10.8% Speed(reviews/sec):640.2 #Correct:55 #Tested:109 Testing Accuracy:50.4%\r",
      "Progress:10.9% Speed(reviews/sec):641.4 #Correct:55 #Tested:110 Testing Accuracy:50.0%\r",
      "Progress:11.0% Speed(reviews/sec):642.3 #Correct:56 #Tested:111 Testing Accuracy:50.4%\r",
      "Progress:11.1% Speed(reviews/sec):643.1 #Correct:56 #Tested:112 Testing Accuracy:50.0%\r",
      "Progress:11.2% Speed(reviews/sec):643.7 #Correct:57 #Tested:113 Testing Accuracy:50.4%\r",
      "Progress:11.3% Speed(reviews/sec):643.2 #Correct:57 #Tested:114 Testing Accuracy:50.0%\r",
      "Progress:11.4% Speed(reviews/sec):643.7 #Correct:58 #Tested:115 Testing Accuracy:50.4%\r",
      "Progress:11.5% Speed(reviews/sec):644.9 #Correct:58 #Tested:116 Testing Accuracy:50.0%\r",
      "Progress:11.6% Speed(reviews/sec):646.1 #Correct:59 #Tested:117 Testing Accuracy:50.4%\r",
      "Progress:11.7% Speed(reviews/sec):646.6 #Correct:59 #Tested:118 Testing Accuracy:50.0%\r",
      "Progress:11.8% Speed(reviews/sec):647.6 #Correct:60 #Tested:119 Testing Accuracy:50.4%\r",
      "Progress:11.9% Speed(reviews/sec):647.4 #Correct:60 #Tested:120 Testing Accuracy:50.0%\r",
      "Progress:12.0% Speed(reviews/sec):648.1 #Correct:61 #Tested:121 Testing Accuracy:50.4%\r",
      "Progress:12.1% Speed(reviews/sec):649.0 #Correct:61 #Tested:122 Testing Accuracy:50.0%\r",
      "Progress:12.2% Speed(reviews/sec):649.9 #Correct:62 #Tested:123 Testing Accuracy:50.4%\r",
      "Progress:12.3% Speed(reviews/sec):648.8 #Correct:62 #Tested:124 Testing Accuracy:50.0%\r",
      "Progress:12.4% Speed(reviews/sec):649.0 #Correct:63 #Tested:125 Testing Accuracy:50.4%\r",
      "Progress:12.5% Speed(reviews/sec):649.3 #Correct:63 #Tested:126 Testing Accuracy:50.0%\r",
      "Progress:12.6% Speed(reviews/sec):650.7 #Correct:64 #Tested:127 Testing Accuracy:50.3%\r",
      "Progress:12.7% Speed(reviews/sec):650.6 #Correct:64 #Tested:128 Testing Accuracy:50.0%\r",
      "Progress:12.8% Speed(reviews/sec):651.9 #Correct:65 #Tested:129 Testing Accuracy:50.3%\r",
      "Progress:12.9% Speed(reviews/sec):652.8 #Correct:65 #Tested:130 Testing Accuracy:50.0%\r",
      "Progress:13.0% Speed(reviews/sec):653.5 #Correct:66 #Tested:131 Testing Accuracy:50.3%\r",
      "Progress:13.1% Speed(reviews/sec):654.1 #Correct:66 #Tested:132 Testing Accuracy:50.0%\r",
      "Progress:13.2% Speed(reviews/sec):655.0 #Correct:67 #Tested:133 Testing Accuracy:50.3%\r",
      "Progress:13.3% Speed(reviews/sec):655.9 #Correct:67 #Tested:134 Testing Accuracy:50.0%\r",
      "Progress:13.4% Speed(reviews/sec):655.8 #Correct:68 #Tested:135 Testing Accuracy:50.3%\r",
      "Progress:13.5% Speed(reviews/sec):656.0 #Correct:68 #Tested:136 Testing Accuracy:50.0%\r",
      "Progress:13.6% Speed(reviews/sec):656.9 #Correct:69 #Tested:137 Testing Accuracy:50.3%\r",
      "Progress:13.7% Speed(reviews/sec):657.9 #Correct:69 #Tested:138 Testing Accuracy:50.0%\r",
      "Progress:13.8% Speed(reviews/sec):658.6 #Correct:70 #Tested:139 Testing Accuracy:50.3%"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\r",
      "Progress:13.9% Speed(reviews/sec):657.8 #Correct:70 #Tested:140 Testing Accuracy:50.0%\r",
      "Progress:14.0% Speed(reviews/sec):658.0 #Correct:71 #Tested:141 Testing Accuracy:50.3%\r",
      "Progress:14.1% Speed(reviews/sec):657.4 #Correct:71 #Tested:142 Testing Accuracy:50.0%\r",
      "Progress:14.2% Speed(reviews/sec):657.3 #Correct:72 #Tested:143 Testing Accuracy:50.3%\r",
      "Progress:14.3% Speed(reviews/sec):657.6 #Correct:72 #Tested:144 Testing Accuracy:50.0%\r",
      "Progress:14.4% Speed(reviews/sec):657.5 #Correct:73 #Tested:145 Testing Accuracy:50.3%\r",
      "Progress:14.5% Speed(reviews/sec):658.2 #Correct:73 #Tested:146 Testing Accuracy:50.0%\r",
      "Progress:14.6% Speed(reviews/sec):656.0 #Correct:74 #Tested:147 Testing Accuracy:50.3%\r",
      "Progress:14.7% Speed(reviews/sec):655.6 #Correct:74 #Tested:148 Testing Accuracy:50.0%\r",
      "Progress:14.8% Speed(reviews/sec):655.4 #Correct:75 #Tested:149 Testing Accuracy:50.3%\r",
      "Progress:14.9% Speed(reviews/sec):655.4 #Correct:75 #Tested:150 Testing Accuracy:50.0%\r",
      "Progress:15.0% Speed(reviews/sec):655.6 #Correct:76 #Tested:151 Testing Accuracy:50.3%\r",
      "Progress:15.1% Speed(reviews/sec):655.9 #Correct:76 #Tested:152 Testing Accuracy:50.0%\r",
      "Progress:15.2% Speed(reviews/sec):653.3 #Correct:77 #Tested:153 Testing Accuracy:50.3%\r",
      "Progress:15.3% Speed(reviews/sec):650.8 #Correct:77 #Tested:154 Testing Accuracy:50.0%\r",
      "Progress:15.4% Speed(reviews/sec):650.5 #Correct:78 #Tested:155 Testing Accuracy:50.3%\r",
      "Progress:15.5% Speed(reviews/sec):650.7 #Correct:78 #Tested:156 Testing Accuracy:50.0%\r",
      "Progress:15.6% Speed(reviews/sec):650.0 #Correct:79 #Tested:157 Testing Accuracy:50.3%\r",
      "Progress:15.7% Speed(reviews/sec):650.0 #Correct:79 #Tested:158 Testing Accuracy:50.0%\r",
      "Progress:15.8% Speed(reviews/sec):650.8 #Correct:80 #Tested:159 Testing Accuracy:50.3%\r",
      "Progress:15.9% Speed(reviews/sec):651.1 #Correct:80 #Tested:160 Testing Accuracy:50.0%\r",
      "Progress:16.0% Speed(reviews/sec):651.1 #Correct:81 #Tested:161 Testing Accuracy:50.3%\r",
      "Progress:16.1% Speed(reviews/sec):650.3 #Correct:81 #Tested:162 Testing Accuracy:50.0%\r",
      "Progress:16.2% Speed(reviews/sec):648.2 #Correct:82 #Tested:163 Testing Accuracy:50.3%\r",
      "Progress:16.3% Speed(reviews/sec):645.8 #Correct:82 #Tested:164 Testing Accuracy:50.0%\r",
      "Progress:16.4% Speed(reviews/sec):646.6 #Correct:83 #Tested:165 Testing Accuracy:50.3%\r",
      "Progress:16.5% Speed(reviews/sec):646.4 #Correct:83 #Tested:166 Testing Accuracy:50.0%\r",
      "Progress:16.6% Speed(reviews/sec):646.9 #Correct:84 #Tested:167 Testing Accuracy:50.2%\r",
      "Progress:16.7% Speed(reviews/sec):647.0 #Correct:84 #Tested:168 Testing Accuracy:50.0%\r",
      "Progress:16.8% Speed(reviews/sec):648.0 #Correct:85 #Tested:169 Testing Accuracy:50.2%\r",
      "Progress:16.9% Speed(reviews/sec):648.7 #Correct:85 #Tested:170 Testing Accuracy:50.0%\r",
      "Progress:17.0% Speed(reviews/sec):649.0 #Correct:86 #Tested:171 Testing Accuracy:50.2%\r",
      "Progress:17.1% Speed(reviews/sec):649.8 #Correct:86 #Tested:172 Testing Accuracy:50.0%\r",
      "Progress:17.2% Speed(reviews/sec):649.4 #Correct:87 #Tested:173 Testing Accuracy:50.2%\r",
      "Progress:17.3% Speed(reviews/sec):649.5 #Correct:87 #Tested:174 Testing Accuracy:50.0%\r",
      "Progress:17.4% Speed(reviews/sec):648.1 #Correct:88 #Tested:175 Testing Accuracy:50.2%\r",
      "Progress:17.5% Speed(reviews/sec):648.8 #Correct:88 #Tested:176 Testing Accuracy:50.0%\r",
      "Progress:17.6% Speed(reviews/sec):649.4 #Correct:89 #Tested:177 Testing Accuracy:50.2%\r",
      "Progress:17.7% Speed(reviews/sec):649.5 #Correct:89 #Tested:178 Testing Accuracy:50.0%\r",
      "Progress:17.8% Speed(reviews/sec):650.1 #Correct:90 #Tested:179 Testing Accuracy:50.2%\r",
      "Progress:17.9% Speed(reviews/sec):650.8 #Correct:90 #Tested:180 Testing Accuracy:50.0%\r",
      "Progress:18.0% Speed(reviews/sec):651.3 #Correct:91 #Tested:181 Testing Accuracy:50.2%\r",
      "Progress:18.1% Speed(reviews/sec):652.1 #Correct:91 #Tested:182 Testing Accuracy:50.0%\r",
      "Progress:18.2% Speed(reviews/sec):652.6 #Correct:92 #Tested:183 Testing Accuracy:50.2%\r",
      "Progress:18.3% Speed(reviews/sec):653.1 #Correct:92 #Tested:184 Testing Accuracy:50.0%\r",
      "Progress:18.4% Speed(reviews/sec):652.5 #Correct:93 #Tested:185 Testing Accuracy:50.2%\r",
      "Progress:18.5% Speed(reviews/sec):652.4 #Correct:93 #Tested:186 Testing Accuracy:50.0%\r",
      "Progress:18.6% Speed(reviews/sec):653.3 #Correct:94 #Tested:187 Testing Accuracy:50.2%\r",
      "Progress:18.7% Speed(reviews/sec):653.5 #Correct:94 #Tested:188 Testing Accuracy:50.0%\r",
      "Progress:18.8% Speed(reviews/sec):653.4 #Correct:95 #Tested:189 Testing Accuracy:50.2%\r",
      "Progress:18.9% Speed(reviews/sec):654.0 #Correct:95 #Tested:190 Testing Accuracy:50.0%\r",
      "Progress:19.0% Speed(reviews/sec):654.7 #Correct:96 #Tested:191 Testing Accuracy:50.2%\r",
      "Progress:19.1% Speed(reviews/sec):655.4 #Correct:96 #Tested:192 Testing Accuracy:50.0%\r",
      "Progress:19.2% Speed(reviews/sec):656.0 #Correct:97 #Tested:193 Testing Accuracy:50.2%\r",
      "Progress:19.3% Speed(reviews/sec):655.4 #Correct:97 #Tested:194 Testing Accuracy:50.0%\r",
      "Progress:19.4% Speed(reviews/sec):656.2 #Correct:98 #Tested:195 Testing Accuracy:50.2%\r",
      "Progress:19.5% Speed(reviews/sec):656.5 #Correct:98 #Tested:196 Testing Accuracy:50.0%\r",
      "Progress:19.6% Speed(reviews/sec):656.6 #Correct:99 #Tested:197 Testing Accuracy:50.2%\r",
      "Progress:19.7% Speed(reviews/sec):656.8 #Correct:99 #Tested:198 Testing Accuracy:50.0%\r",
      "Progress:19.8% Speed(reviews/sec):657.5 #Correct:100 #Tested:199 Testing Accuracy:50.2%\r",
      "Progress:19.9% Speed(reviews/sec):657.7 #Correct:100 #Tested:200 Testing Accuracy:50.0%\r",
      "Progress:20.0% Speed(reviews/sec):658.6 #Correct:101 #Tested:201 Testing Accuracy:50.2%\r",
      "Progress:20.1% Speed(reviews/sec):658.9 #Correct:101 #Tested:202 Testing Accuracy:50.0%\r",
      "Progress:20.2% Speed(reviews/sec):659.6 #Correct:102 #Tested:203 Testing Accuracy:50.2%\r",
      "Progress:20.3% Speed(reviews/sec):659.7 #Correct:102 #Tested:204 Testing Accuracy:50.0%\r",
      "Progress:20.4% Speed(reviews/sec):660.3 #Correct:103 #Tested:205 Testing Accuracy:50.2%\r",
      "Progress:20.5% Speed(reviews/sec):660.9 #Correct:103 #Tested:206 Testing Accuracy:50.0%\r",
      "Progress:20.6% Speed(reviews/sec):661.4 #Correct:104 #Tested:207 Testing Accuracy:50.2%\r",
      "Progress:20.7% Speed(reviews/sec):662.1 #Correct:104 #Tested:208 Testing Accuracy:50.0%\r",
      "Progress:20.8% Speed(reviews/sec):662.7 #Correct:105 #Tested:209 Testing Accuracy:50.2%\r",
      "Progress:20.9% Speed(reviews/sec):662.3 #Correct:105 #Tested:210 Testing Accuracy:50.0%\r",
      "Progress:21.0% Speed(reviews/sec):662.6 #Correct:106 #Tested:211 Testing Accuracy:50.2%\r",
      "Progress:21.1% Speed(reviews/sec):663.1 #Correct:106 #Tested:212 Testing Accuracy:50.0%\r",
      "Progress:21.2% Speed(reviews/sec):663.8 #Correct:107 #Tested:213 Testing Accuracy:50.2%\r",
      "Progress:21.3% Speed(reviews/sec):662.6 #Correct:107 #Tested:214 Testing Accuracy:50.0%\r",
      "Progress:21.4% Speed(reviews/sec):663.4 #Correct:108 #Tested:215 Testing Accuracy:50.2%\r",
      "Progress:21.5% Speed(reviews/sec):663.2 #Correct:108 #Tested:216 Testing Accuracy:50.0%\r",
      "Progress:21.6% Speed(reviews/sec):663.7 #Correct:109 #Tested:217 Testing Accuracy:50.2%\r",
      "Progress:21.7% Speed(reviews/sec):663.7 #Correct:109 #Tested:218 Testing Accuracy:50.0%\r",
      "Progress:21.8% Speed(reviews/sec):663.5 #Correct:110 #Tested:219 Testing Accuracy:50.2%\r",
      "Progress:21.9% Speed(reviews/sec):664.3 #Correct:110 #Tested:220 Testing Accuracy:50.0%\r",
      "Progress:22.0% Speed(reviews/sec):664.3 #Correct:111 #Tested:221 Testing Accuracy:50.2%\r",
      "Progress:22.1% Speed(reviews/sec):664.7 #Correct:111 #Tested:222 Testing Accuracy:50.0%\r",
      "Progress:22.2% Speed(reviews/sec):664.7 #Correct:112 #Tested:223 Testing Accuracy:50.2%\r",
      "Progress:22.3% Speed(reviews/sec):665.1 #Correct:112 #Tested:224 Testing Accuracy:50.0%\r",
      "Progress:22.4% Speed(reviews/sec):663.9 #Correct:113 #Tested:225 Testing Accuracy:50.2%\r",
      "Progress:22.5% Speed(reviews/sec):664.2 #Correct:113 #Tested:226 Testing Accuracy:50.0%\r",
      "Progress:22.6% Speed(reviews/sec):663.3 #Correct:114 #Tested:227 Testing Accuracy:50.2%\r",
      "Progress:22.7% Speed(reviews/sec):663.7 #Correct:114 #Tested:228 Testing Accuracy:50.0%\r",
      "Progress:22.8% Speed(reviews/sec):664.1 #Correct:115 #Tested:229 Testing Accuracy:50.2%\r",
      "Progress:22.9% Speed(reviews/sec):664.4 #Correct:115 #Tested:230 Testing Accuracy:50.0%\r",
      "Progress:23.0% Speed(reviews/sec):663.8 #Correct:116 #Tested:231 Testing Accuracy:50.2%\r",
      "Progress:23.1% Speed(reviews/sec):663.9 #Correct:116 #Tested:232 Testing Accuracy:50.0%\r",
      "Progress:23.2% Speed(reviews/sec):664.6 #Correct:117 #Tested:233 Testing Accuracy:50.2%\r",
      "Progress:23.3% Speed(reviews/sec):664.8 #Correct:117 #Tested:234 Testing Accuracy:50.0%\r",
      "Progress:23.4% Speed(reviews/sec):664.0 #Correct:118 #Tested:235 Testing Accuracy:50.2%\r",
      "Progress:23.5% Speed(reviews/sec):664.4 #Correct:118 #Tested:236 Testing Accuracy:50.0%\r",
      "Progress:23.6% Speed(reviews/sec):663.6 #Correct:119 #Tested:237 Testing Accuracy:50.2%\r",
      "Progress:23.7% Speed(reviews/sec):663.8 #Correct:119 #Tested:238 Testing Accuracy:50.0%\r",
      "Progress:23.8% Speed(reviews/sec):664.3 #Correct:120 #Tested:239 Testing Accuracy:50.2%\r",
      "Progress:23.9% Speed(reviews/sec):664.4 #Correct:120 #Tested:240 Testing Accuracy:50.0%\r",
      "Progress:24.0% Speed(reviews/sec):664.3 #Correct:121 #Tested:241 Testing Accuracy:50.2%\r",
      "Progress:24.1% Speed(reviews/sec):664.4 #Correct:121 #Tested:242 Testing Accuracy:50.0%\r",
      "Progress:24.2% Speed(reviews/sec):664.5 #Correct:122 #Tested:243 Testing Accuracy:50.2%\r",
      "Progress:24.3% Speed(reviews/sec):664.0 #Correct:122 #Tested:244 Testing Accuracy:50.0%\r",
      "Progress:24.4% Speed(reviews/sec):663.9 #Correct:123 #Tested:245 Testing Accuracy:50.2%\r",
      "Progress:24.5% Speed(reviews/sec):664.4 #Correct:123 #Tested:246 Testing Accuracy:50.0%\r",
      "Progress:24.6% Speed(reviews/sec):665.1 #Correct:124 #Tested:247 Testing Accuracy:50.2%\r",
      "Progress:24.7% Speed(reviews/sec):665.3 #Correct:124 #Tested:248 Testing Accuracy:50.0%\r",
      "Progress:24.8% Speed(reviews/sec):666.1 #Correct:125 #Tested:249 Testing Accuracy:50.2%\r",
      "Progress:24.9% Speed(reviews/sec):666.5 #Correct:125 #Tested:250 Testing Accuracy:50.0%\r",
      "Progress:25.0% Speed(reviews/sec):667.0 #Correct:126 #Tested:251 Testing Accuracy:50.1%\r",
      "Progress:25.1% Speed(reviews/sec):667.5 #Correct:126 #Tested:252 Testing Accuracy:50.0%\r",
      "Progress:25.2% Speed(reviews/sec):668.0 #Correct:127 #Tested:253 Testing Accuracy:50.1%\r",
      "Progress:25.3% Speed(reviews/sec):668.3 #Correct:127 #Tested:254 Testing Accuracy:50.0%\r",
      "Progress:25.4% Speed(reviews/sec):668.9 #Correct:128 #Tested:255 Testing Accuracy:50.1%\r",
      "Progress:25.5% Speed(reviews/sec):669.5 #Correct:128 #Tested:256 Testing Accuracy:50.0%\r",
      "Progress:25.6% Speed(reviews/sec):670.2 #Correct:129 #Tested:257 Testing Accuracy:50.1%\r",
      "Progress:25.7% Speed(reviews/sec):670.5 #Correct:129 #Tested:258 Testing Accuracy:50.0%\r",
      "Progress:25.8% Speed(reviews/sec):671.0 #Correct:130 #Tested:259 Testing Accuracy:50.1%\r",
      "Progress:25.9% Speed(reviews/sec):671.5 #Correct:130 #Tested:260 Testing Accuracy:50.0%\r",
      "Progress:26.0% Speed(reviews/sec):672.1 #Correct:131 #Tested:261 Testing Accuracy:50.1%\r",
      "Progress:26.1% Speed(reviews/sec):672.8 #Correct:131 #Tested:262 Testing Accuracy:50.0%\r",
      "Progress:26.2% Speed(reviews/sec):672.8 #Correct:132 #Tested:263 Testing Accuracy:50.1%\r",
      "Progress:26.3% Speed(reviews/sec):672.5 #Correct:132 #Tested:264 Testing Accuracy:50.0%\r",
      "Progress:26.4% Speed(reviews/sec):672.8 #Correct:133 #Tested:265 Testing Accuracy:50.1%\r",
      "Progress:26.5% Speed(reviews/sec):672.2 #Correct:133 #Tested:266 Testing Accuracy:50.0%\r",
      "Progress:26.6% Speed(reviews/sec):672.5 #Correct:134 #Tested:267 Testing Accuracy:50.1%\r",
      "Progress:26.7% Speed(reviews/sec):672.5 #Correct:134 #Tested:268 Testing Accuracy:50.0%\r",
      "Progress:26.8% Speed(reviews/sec):672.9 #Correct:135 #Tested:269 Testing Accuracy:50.1%\r",
      "Progress:26.9% Speed(reviews/sec):673.2 #Correct:135 #Tested:270 Testing Accuracy:50.0%\r",
      "Progress:27.0% Speed(reviews/sec):673.2 #Correct:136 #Tested:271 Testing Accuracy:50.1%\r",
      "Progress:27.1% Speed(reviews/sec):673.4 #Correct:136 #Tested:272 Testing Accuracy:50.0%\r",
      "Progress:27.2% Speed(reviews/sec):673.1 #Correct:137 #Tested:273 Testing Accuracy:50.1%\r",
      "Progress:27.3% Speed(reviews/sec):673.6 #Correct:137 #Tested:274 Testing Accuracy:50.0%\r",
      "Progress:27.4% Speed(reviews/sec):673.4 #Correct:138 #Tested:275 Testing Accuracy:50.1%\r",
      "Progress:27.5% Speed(reviews/sec):672.8 #Correct:138 #Tested:276 Testing Accuracy:50.0%\r",
      "Progress:27.6% Speed(reviews/sec):673.3 #Correct:139 #Tested:277 Testing Accuracy:50.1%\r",
      "Progress:27.7% Speed(reviews/sec):673.6 #Correct:139 #Tested:278 Testing Accuracy:50.0%"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\r",
      "Progress:27.8% Speed(reviews/sec):673.7 #Correct:140 #Tested:279 Testing Accuracy:50.1%\r",
      "Progress:27.9% Speed(reviews/sec):674.0 #Correct:140 #Tested:280 Testing Accuracy:50.0%\r",
      "Progress:28.0% Speed(reviews/sec):673.2 #Correct:141 #Tested:281 Testing Accuracy:50.1%\r",
      "Progress:28.1% Speed(reviews/sec):672.9 #Correct:141 #Tested:282 Testing Accuracy:50.0%\r",
      "Progress:28.2% Speed(reviews/sec):673.2 #Correct:142 #Tested:283 Testing Accuracy:50.1%\r",
      "Progress:28.3% Speed(reviews/sec):673.4 #Correct:142 #Tested:284 Testing Accuracy:50.0%\r",
      "Progress:28.4% Speed(reviews/sec):669.9 #Correct:143 #Tested:285 Testing Accuracy:50.1%\r",
      "Progress:28.5% Speed(reviews/sec):669.6 #Correct:143 #Tested:286 Testing Accuracy:50.0%\r",
      "Progress:28.6% Speed(reviews/sec):670.2 #Correct:144 #Tested:287 Testing Accuracy:50.1%\r",
      "Progress:28.7% Speed(reviews/sec):670.7 #Correct:144 #Tested:288 Testing Accuracy:50.0%\r",
      "Progress:28.8% Speed(reviews/sec):671.2 #Correct:145 #Tested:289 Testing Accuracy:50.1%\r",
      "Progress:28.9% Speed(reviews/sec):671.7 #Correct:145 #Tested:290 Testing Accuracy:50.0%\r",
      "Progress:29.0% Speed(reviews/sec):671.5 #Correct:146 #Tested:291 Testing Accuracy:50.1%\r",
      "Progress:29.1% Speed(reviews/sec):671.6 #Correct:146 #Tested:292 Testing Accuracy:50.0%\r",
      "Progress:29.2% Speed(reviews/sec):671.8 #Correct:147 #Tested:293 Testing Accuracy:50.1%\r",
      "Progress:29.3% Speed(reviews/sec):672.4 #Correct:147 #Tested:294 Testing Accuracy:50.0%\r",
      "Progress:29.4% Speed(reviews/sec):672.4 #Correct:148 #Tested:295 Testing Accuracy:50.1%\r",
      "Progress:29.5% Speed(reviews/sec):672.7 #Correct:148 #Tested:296 Testing Accuracy:50.0%\r",
      "Progress:29.6% Speed(reviews/sec):673.2 #Correct:149 #Tested:297 Testing Accuracy:50.1%\r",
      "Progress:29.7% Speed(reviews/sec):672.8 #Correct:149 #Tested:298 Testing Accuracy:50.0%\r",
      "Progress:29.8% Speed(reviews/sec):672.8 #Correct:150 #Tested:299 Testing Accuracy:50.1%\r",
      "Progress:29.9% Speed(reviews/sec):672.9 #Correct:150 #Tested:300 Testing Accuracy:50.0%\r",
      "Progress:30.0% Speed(reviews/sec):673.1 #Correct:151 #Tested:301 Testing Accuracy:50.1%\r",
      "Progress:30.1% Speed(reviews/sec):672.2 #Correct:151 #Tested:302 Testing Accuracy:50.0%\r",
      "Progress:30.2% Speed(reviews/sec):669.8 #Correct:152 #Tested:303 Testing Accuracy:50.1%\r",
      "Progress:30.3% Speed(reviews/sec):668.3 #Correct:152 #Tested:304 Testing Accuracy:50.0%\r",
      "Progress:30.4% Speed(reviews/sec):668.9 #Correct:153 #Tested:305 Testing Accuracy:50.1%\r",
      "Progress:30.5% Speed(reviews/sec):669.3 #Correct:153 #Tested:306 Testing Accuracy:50.0%\r",
      "Progress:30.6% Speed(reviews/sec):667.9 #Correct:154 #Tested:307 Testing Accuracy:50.1%\r",
      "Progress:30.7% Speed(reviews/sec):668.4 #Correct:154 #Tested:308 Testing Accuracy:50.0%\r",
      "Progress:30.8% Speed(reviews/sec):668.6 #Correct:155 #Tested:309 Testing Accuracy:50.1%\r",
      "Progress:30.9% Speed(reviews/sec):667.9 #Correct:155 #Tested:310 Testing Accuracy:50.0%\r",
      "Progress:31.0% Speed(reviews/sec):668.4 #Correct:156 #Tested:311 Testing Accuracy:50.1%\r",
      "Progress:31.1% Speed(reviews/sec):668.5 #Correct:156 #Tested:312 Testing Accuracy:50.0%\r",
      "Progress:31.2% Speed(reviews/sec):668.9 #Correct:157 #Tested:313 Testing Accuracy:50.1%\r",
      "Progress:31.3% Speed(reviews/sec):667.8 #Correct:157 #Tested:314 Testing Accuracy:50.0%\r",
      "Progress:31.4% Speed(reviews/sec):668.2 #Correct:158 #Tested:315 Testing Accuracy:50.1%\r",
      "Progress:31.5% Speed(reviews/sec):668.3 #Correct:158 #Tested:316 Testing Accuracy:50.0%\r",
      "Progress:31.6% Speed(reviews/sec):668.7 #Correct:159 #Tested:317 Testing Accuracy:50.1%\r",
      "Progress:31.7% Speed(reviews/sec):668.8 #Correct:159 #Tested:318 Testing Accuracy:50.0%\r",
      "Progress:31.8% Speed(reviews/sec):669.2 #Correct:160 #Tested:319 Testing Accuracy:50.1%\r",
      "Progress:31.9% Speed(reviews/sec):669.5 #Correct:160 #Tested:320 Testing Accuracy:50.0%\r",
      "Progress:32.0% Speed(reviews/sec):669.9 #Correct:161 #Tested:321 Testing Accuracy:50.1%\r",
      "Progress:32.1% Speed(reviews/sec):670.4 #Correct:161 #Tested:322 Testing Accuracy:50.0%\r",
      "Progress:32.2% Speed(reviews/sec):670.8 #Correct:162 #Tested:323 Testing Accuracy:50.1%\r",
      "Progress:32.3% Speed(reviews/sec):671.0 #Correct:162 #Tested:324 Testing Accuracy:50.0%\r",
      "Progress:32.4% Speed(reviews/sec):670.7 #Correct:163 #Tested:325 Testing Accuracy:50.1%\r",
      "Progress:32.5% Speed(reviews/sec):670.8 #Correct:163 #Tested:326 Testing Accuracy:50.0%\r",
      "Progress:32.6% Speed(reviews/sec):671.1 #Correct:164 #Tested:327 Testing Accuracy:50.1%\r",
      "Progress:32.7% Speed(reviews/sec):670.9 #Correct:164 #Tested:328 Testing Accuracy:50.0%\r",
      "Progress:32.8% Speed(reviews/sec):670.8 #Correct:165 #Tested:329 Testing Accuracy:50.1%\r",
      "Progress:32.9% Speed(reviews/sec):671.2 #Correct:165 #Tested:330 Testing Accuracy:50.0%\r",
      "Progress:33.0% Speed(reviews/sec):671.4 #Correct:166 #Tested:331 Testing Accuracy:50.1%\r",
      "Progress:33.1% Speed(reviews/sec):671.7 #Correct:166 #Tested:332 Testing Accuracy:50.0%\r",
      "Progress:33.2% Speed(reviews/sec):671.7 #Correct:167 #Tested:333 Testing Accuracy:50.1%\r",
      "Progress:33.3% Speed(reviews/sec):672.0 #Correct:167 #Tested:334 Testing Accuracy:50.0%\r",
      "Progress:33.4% Speed(reviews/sec):671.1 #Correct:168 #Tested:335 Testing Accuracy:50.1%\r",
      "Progress:33.5% Speed(reviews/sec):671.4 #Correct:168 #Tested:336 Testing Accuracy:50.0%\r",
      "Progress:33.6% Speed(reviews/sec):671.6 #Correct:169 #Tested:337 Testing Accuracy:50.1%\r",
      "Progress:33.7% Speed(reviews/sec):671.9 #Correct:169 #Tested:338 Testing Accuracy:50.0%\r",
      "Progress:33.8% Speed(reviews/sec):672.3 #Correct:170 #Tested:339 Testing Accuracy:50.1%\r",
      "Progress:33.9% Speed(reviews/sec):672.5 #Correct:170 #Tested:340 Testing Accuracy:50.0%\r",
      "Progress:34.0% Speed(reviews/sec):672.3 #Correct:171 #Tested:341 Testing Accuracy:50.1%\r",
      "Progress:34.1% Speed(reviews/sec):672.6 #Correct:171 #Tested:342 Testing Accuracy:50.0%\r",
      "Progress:34.2% Speed(reviews/sec):671.5 #Correct:172 #Tested:343 Testing Accuracy:50.1%\r",
      "Progress:34.3% Speed(reviews/sec):672.0 #Correct:172 #Tested:344 Testing Accuracy:50.0%\r",
      "Progress:34.4% Speed(reviews/sec):671.8 #Correct:173 #Tested:345 Testing Accuracy:50.1%\r",
      "Progress:34.5% Speed(reviews/sec):672.1 #Correct:173 #Tested:346 Testing Accuracy:50.0%\r",
      "Progress:34.6% Speed(reviews/sec):672.5 #Correct:174 #Tested:347 Testing Accuracy:50.1%\r",
      "Progress:34.7% Speed(reviews/sec):672.6 #Correct:174 #Tested:348 Testing Accuracy:50.0%\r",
      "Progress:34.8% Speed(reviews/sec):672.6 #Correct:175 #Tested:349 Testing Accuracy:50.1%\r",
      "Progress:34.9% Speed(reviews/sec):672.7 #Correct:175 #Tested:350 Testing Accuracy:50.0%\r",
      "Progress:35.0% Speed(reviews/sec):673.0 #Correct:176 #Tested:351 Testing Accuracy:50.1%\r",
      "Progress:35.1% Speed(reviews/sec):673.3 #Correct:176 #Tested:352 Testing Accuracy:50.0%\r",
      "Progress:35.2% Speed(reviews/sec):673.2 #Correct:177 #Tested:353 Testing Accuracy:50.1%\r",
      "Progress:35.3% Speed(reviews/sec):673.7 #Correct:177 #Tested:354 Testing Accuracy:50.0%\r",
      "Progress:35.4% Speed(reviews/sec):673.8 #Correct:178 #Tested:355 Testing Accuracy:50.1%\r",
      "Progress:35.5% Speed(reviews/sec):673.4 #Correct:178 #Tested:356 Testing Accuracy:50.0%\r",
      "Progress:35.6% Speed(reviews/sec):673.5 #Correct:179 #Tested:357 Testing Accuracy:50.1%\r",
      "Progress:35.7% Speed(reviews/sec):673.3 #Correct:179 #Tested:358 Testing Accuracy:50.0%\r",
      "Progress:35.8% Speed(reviews/sec):673.6 #Correct:180 #Tested:359 Testing Accuracy:50.1%\r",
      "Progress:35.9% Speed(reviews/sec):673.8 #Correct:180 #Tested:360 Testing Accuracy:50.0%\r",
      "Progress:36.0% Speed(reviews/sec):673.7 #Correct:181 #Tested:361 Testing Accuracy:50.1%\r",
      "Progress:36.1% Speed(reviews/sec):673.7 #Correct:181 #Tested:362 Testing Accuracy:50.0%\r",
      "Progress:36.2% Speed(reviews/sec):673.0 #Correct:182 #Tested:363 Testing Accuracy:50.1%\r",
      "Progress:36.3% Speed(reviews/sec):673.2 #Correct:182 #Tested:364 Testing Accuracy:50.0%\r",
      "Progress:36.4% Speed(reviews/sec):673.5 #Correct:183 #Tested:365 Testing Accuracy:50.1%\r",
      "Progress:36.5% Speed(reviews/sec):673.6 #Correct:183 #Tested:366 Testing Accuracy:50.0%\r",
      "Progress:36.6% Speed(reviews/sec):673.8 #Correct:184 #Tested:367 Testing Accuracy:50.1%\r",
      "Progress:36.7% Speed(reviews/sec):674.2 #Correct:184 #Tested:368 Testing Accuracy:50.0%\r",
      "Progress:36.8% Speed(reviews/sec):674.2 #Correct:185 #Tested:369 Testing Accuracy:50.1%\r",
      "Progress:36.9% Speed(reviews/sec):674.5 #Correct:185 #Tested:370 Testing Accuracy:50.0%\r",
      "Progress:37.0% Speed(reviews/sec):674.6 #Correct:186 #Tested:371 Testing Accuracy:50.1%\r",
      "Progress:37.1% Speed(reviews/sec):673.9 #Correct:186 #Tested:372 Testing Accuracy:50.0%\r",
      "Progress:37.2% Speed(reviews/sec):674.1 #Correct:187 #Tested:373 Testing Accuracy:50.1%\r",
      "Progress:37.3% Speed(reviews/sec):674.5 #Correct:187 #Tested:374 Testing Accuracy:50.0%\r",
      "Progress:37.4% Speed(reviews/sec):674.2 #Correct:188 #Tested:375 Testing Accuracy:50.1%\r",
      "Progress:37.5% Speed(reviews/sec):674.3 #Correct:188 #Tested:376 Testing Accuracy:50.0%\r",
      "Progress:37.6% Speed(reviews/sec):674.5 #Correct:189 #Tested:377 Testing Accuracy:50.1%\r",
      "Progress:37.7% Speed(reviews/sec):674.7 #Correct:189 #Tested:378 Testing Accuracy:50.0%\r",
      "Progress:37.8% Speed(reviews/sec):675.0 #Correct:190 #Tested:379 Testing Accuracy:50.1%\r",
      "Progress:37.9% Speed(reviews/sec):675.4 #Correct:190 #Tested:380 Testing Accuracy:50.0%\r",
      "Progress:38.0% Speed(reviews/sec):674.7 #Correct:191 #Tested:381 Testing Accuracy:50.1%\r",
      "Progress:38.1% Speed(reviews/sec):675.0 #Correct:191 #Tested:382 Testing Accuracy:50.0%\r",
      "Progress:38.2% Speed(reviews/sec):674.8 #Correct:192 #Tested:383 Testing Accuracy:50.1%\r",
      "Progress:38.3% Speed(reviews/sec):674.9 #Correct:192 #Tested:384 Testing Accuracy:50.0%\r",
      "Progress:38.4% Speed(reviews/sec):675.1 #Correct:193 #Tested:385 Testing Accuracy:50.1%\r",
      "Progress:38.5% Speed(reviews/sec):675.4 #Correct:193 #Tested:386 Testing Accuracy:50.0%\r",
      "Progress:38.6% Speed(reviews/sec):675.4 #Correct:194 #Tested:387 Testing Accuracy:50.1%\r",
      "Progress:38.7% Speed(reviews/sec):675.9 #Correct:194 #Tested:388 Testing Accuracy:50.0%\r",
      "Progress:38.8% Speed(reviews/sec):676.1 #Correct:195 #Tested:389 Testing Accuracy:50.1%\r",
      "Progress:38.9% Speed(reviews/sec):676.4 #Correct:195 #Tested:390 Testing Accuracy:50.0%\r",
      "Progress:39.0% Speed(reviews/sec):676.7 #Correct:196 #Tested:391 Testing Accuracy:50.1%\r",
      "Progress:39.1% Speed(reviews/sec):676.9 #Correct:196 #Tested:392 Testing Accuracy:50.0%\r",
      "Progress:39.2% Speed(reviews/sec):677.2 #Correct:197 #Tested:393 Testing Accuracy:50.1%\r",
      "Progress:39.3% Speed(reviews/sec):677.2 #Correct:197 #Tested:394 Testing Accuracy:50.0%\r",
      "Progress:39.4% Speed(reviews/sec):677.1 #Correct:198 #Tested:395 Testing Accuracy:50.1%\r",
      "Progress:39.5% Speed(reviews/sec):677.2 #Correct:198 #Tested:396 Testing Accuracy:50.0%\r",
      "Progress:39.6% Speed(reviews/sec):677.3 #Correct:199 #Tested:397 Testing Accuracy:50.1%\r",
      "Progress:39.7% Speed(reviews/sec):677.5 #Correct:199 #Tested:398 Testing Accuracy:50.0%\r",
      "Progress:39.8% Speed(reviews/sec):677.4 #Correct:200 #Tested:399 Testing Accuracy:50.1%\r",
      "Progress:39.9% Speed(reviews/sec):677.6 #Correct:200 #Tested:400 Testing Accuracy:50.0%\r",
      "Progress:40.0% Speed(reviews/sec):677.2 #Correct:201 #Tested:401 Testing Accuracy:50.1%\r",
      "Progress:40.1% Speed(reviews/sec):677.7 #Correct:201 #Tested:402 Testing Accuracy:50.0%\r",
      "Progress:40.2% Speed(reviews/sec):677.7 #Correct:202 #Tested:403 Testing Accuracy:50.1%\r",
      "Progress:40.3% Speed(reviews/sec):678.0 #Correct:202 #Tested:404 Testing Accuracy:50.0%\r",
      "Progress:40.4% Speed(reviews/sec):678.3 #Correct:203 #Tested:405 Testing Accuracy:50.1%\r",
      "Progress:40.5% Speed(reviews/sec):678.4 #Correct:203 #Tested:406 Testing Accuracy:50.0%\r",
      "Progress:40.6% Speed(reviews/sec):678.3 #Correct:204 #Tested:407 Testing Accuracy:50.1%\r",
      "Progress:40.7% Speed(reviews/sec):678.4 #Correct:204 #Tested:408 Testing Accuracy:50.0%\r",
      "Progress:40.8% Speed(reviews/sec):678.4 #Correct:205 #Tested:409 Testing Accuracy:50.1%\r",
      "Progress:40.9% Speed(reviews/sec):678.6 #Correct:205 #Tested:410 Testing Accuracy:50.0%\r",
      "Progress:41.0% Speed(reviews/sec):678.8 #Correct:206 #Tested:411 Testing Accuracy:50.1%\r",
      "Progress:41.1% Speed(reviews/sec):678.7 #Correct:206 #Tested:412 Testing Accuracy:50.0%\r",
      "Progress:41.2% Speed(reviews/sec):679.0 #Correct:207 #Tested:413 Testing Accuracy:50.1%\r",
      "Progress:41.3% Speed(reviews/sec):678.7 #Correct:207 #Tested:414 Testing Accuracy:50.0%\r",
      "Progress:41.4% Speed(reviews/sec):679.0 #Correct:208 #Tested:415 Testing Accuracy:50.1%\r",
      "Progress:41.5% Speed(reviews/sec):679.0 #Correct:208 #Tested:416 Testing Accuracy:50.0%\r",
      "Progress:41.6% Speed(reviews/sec):679.2 #Correct:209 #Tested:417 Testing Accuracy:50.1%"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\r",
      "Progress:41.7% Speed(reviews/sec):679.2 #Correct:209 #Tested:418 Testing Accuracy:50.0%\r",
      "Progress:41.8% Speed(reviews/sec):679.2 #Correct:210 #Tested:419 Testing Accuracy:50.1%\r",
      "Progress:41.9% Speed(reviews/sec):679.4 #Correct:210 #Tested:420 Testing Accuracy:50.0%\r",
      "Progress:42.0% Speed(reviews/sec):678.0 #Correct:211 #Tested:421 Testing Accuracy:50.1%\r",
      "Progress:42.1% Speed(reviews/sec):677.8 #Correct:211 #Tested:422 Testing Accuracy:50.0%\r",
      "Progress:42.2% Speed(reviews/sec):678.0 #Correct:212 #Tested:423 Testing Accuracy:50.1%\r",
      "Progress:42.3% Speed(reviews/sec):678.1 #Correct:212 #Tested:424 Testing Accuracy:50.0%\r",
      "Progress:42.4% Speed(reviews/sec):678.2 #Correct:213 #Tested:425 Testing Accuracy:50.1%\r",
      "Progress:42.5% Speed(reviews/sec):678.2 #Correct:213 #Tested:426 Testing Accuracy:50.0%\r",
      "Progress:42.6% Speed(reviews/sec):676.6 #Correct:214 #Tested:427 Testing Accuracy:50.1%\r",
      "Progress:42.7% Speed(reviews/sec):676.6 #Correct:214 #Tested:428 Testing Accuracy:50.0%\r",
      "Progress:42.8% Speed(reviews/sec):676.2 #Correct:215 #Tested:429 Testing Accuracy:50.1%\r",
      "Progress:42.9% Speed(reviews/sec):675.9 #Correct:215 #Tested:430 Testing Accuracy:50.0%\r",
      "Progress:43.0% Speed(reviews/sec):676.1 #Correct:216 #Tested:431 Testing Accuracy:50.1%\r",
      "Progress:43.1% Speed(reviews/sec):676.3 #Correct:216 #Tested:432 Testing Accuracy:50.0%\r",
      "Progress:43.2% Speed(reviews/sec):676.2 #Correct:217 #Tested:433 Testing Accuracy:50.1%\r",
      "Progress:43.3% Speed(reviews/sec):676.4 #Correct:217 #Tested:434 Testing Accuracy:50.0%\r",
      "Progress:43.4% Speed(reviews/sec):676.5 #Correct:218 #Tested:435 Testing Accuracy:50.1%\r",
      "Progress:43.5% Speed(reviews/sec):676.5 #Correct:218 #Tested:436 Testing Accuracy:50.0%\r",
      "Progress:43.6% Speed(reviews/sec):676.7 #Correct:219 #Tested:437 Testing Accuracy:50.1%\r",
      "Progress:43.7% Speed(reviews/sec):676.1 #Correct:219 #Tested:438 Testing Accuracy:50.0%\r",
      "Progress:43.8% Speed(reviews/sec):676.0 #Correct:220 #Tested:439 Testing Accuracy:50.1%\r",
      "Progress:43.9% Speed(reviews/sec):675.0 #Correct:220 #Tested:440 Testing Accuracy:50.0%\r",
      "Progress:44.0% Speed(reviews/sec):674.4 #Correct:221 #Tested:441 Testing Accuracy:50.1%\r",
      "Progress:44.1% Speed(reviews/sec):673.5 #Correct:221 #Tested:442 Testing Accuracy:50.0%\r",
      "Progress:44.2% Speed(reviews/sec):673.8 #Correct:222 #Tested:443 Testing Accuracy:50.1%\r",
      "Progress:44.3% Speed(reviews/sec):673.8 #Correct:222 #Tested:444 Testing Accuracy:50.0%\r",
      "Progress:44.4% Speed(reviews/sec):673.2 #Correct:223 #Tested:445 Testing Accuracy:50.1%\r",
      "Progress:44.5% Speed(reviews/sec):673.3 #Correct:223 #Tested:446 Testing Accuracy:50.0%\r",
      "Progress:44.6% Speed(reviews/sec):673.5 #Correct:224 #Tested:447 Testing Accuracy:50.1%\r",
      "Progress:44.7% Speed(reviews/sec):673.5 #Correct:224 #Tested:448 Testing Accuracy:50.0%\r",
      "Progress:44.8% Speed(reviews/sec):673.7 #Correct:225 #Tested:449 Testing Accuracy:50.1%\r",
      "Progress:44.9% Speed(reviews/sec):673.3 #Correct:225 #Tested:450 Testing Accuracy:50.0%\r",
      "Progress:45.0% Speed(reviews/sec):672.9 #Correct:226 #Tested:451 Testing Accuracy:50.1%\r",
      "Progress:45.1% Speed(reviews/sec):672.2 #Correct:226 #Tested:452 Testing Accuracy:50.0%\r",
      "Progress:45.2% Speed(reviews/sec):672.5 #Correct:227 #Tested:453 Testing Accuracy:50.1%\r",
      "Progress:45.3% Speed(reviews/sec):672.7 #Correct:227 #Tested:454 Testing Accuracy:50.0%\r",
      "Progress:45.4% Speed(reviews/sec):673.1 #Correct:228 #Tested:455 Testing Accuracy:50.1%\r",
      "Progress:45.5% Speed(reviews/sec):673.3 #Correct:228 #Tested:456 Testing Accuracy:50.0%\r",
      "Progress:45.6% Speed(reviews/sec):673.6 #Correct:229 #Tested:457 Testing Accuracy:50.1%\r",
      "Progress:45.7% Speed(reviews/sec):673.9 #Correct:229 #Tested:458 Testing Accuracy:50.0%\r",
      "Progress:45.8% Speed(reviews/sec):674.1 #Correct:230 #Tested:459 Testing Accuracy:50.1%\r",
      "Progress:45.9% Speed(reviews/sec):673.8 #Correct:230 #Tested:460 Testing Accuracy:50.0%\r",
      "Progress:46.0% Speed(reviews/sec):674.1 #Correct:231 #Tested:461 Testing Accuracy:50.1%\r",
      "Progress:46.1% Speed(reviews/sec):673.9 #Correct:231 #Tested:462 Testing Accuracy:50.0%\r",
      "Progress:46.2% Speed(reviews/sec):673.0 #Correct:232 #Tested:463 Testing Accuracy:50.1%\r",
      "Progress:46.3% Speed(reviews/sec):673.3 #Correct:232 #Tested:464 Testing Accuracy:50.0%\r",
      "Progress:46.4% Speed(reviews/sec):673.6 #Correct:233 #Tested:465 Testing Accuracy:50.1%\r",
      "Progress:46.5% Speed(reviews/sec):673.8 #Correct:233 #Tested:466 Testing Accuracy:50.0%\r",
      "Progress:46.6% Speed(reviews/sec):674.0 #Correct:234 #Tested:467 Testing Accuracy:50.1%\r",
      "Progress:46.7% Speed(reviews/sec):673.7 #Correct:234 #Tested:468 Testing Accuracy:50.0%\r",
      "Progress:46.8% Speed(reviews/sec):673.5 #Correct:235 #Tested:469 Testing Accuracy:50.1%\r",
      "Progress:46.9% Speed(reviews/sec):673.8 #Correct:235 #Tested:470 Testing Accuracy:50.0%\r",
      "Progress:47.0% Speed(reviews/sec):674.0 #Correct:236 #Tested:471 Testing Accuracy:50.1%\r",
      "Progress:47.1% Speed(reviews/sec):674.2 #Correct:236 #Tested:472 Testing Accuracy:50.0%\r",
      "Progress:47.2% Speed(reviews/sec):674.5 #Correct:237 #Tested:473 Testing Accuracy:50.1%\r",
      "Progress:47.3% Speed(reviews/sec):674.3 #Correct:237 #Tested:474 Testing Accuracy:50.0%\r",
      "Progress:47.4% Speed(reviews/sec):674.4 #Correct:238 #Tested:475 Testing Accuracy:50.1%\r",
      "Progress:47.5% Speed(reviews/sec):674.3 #Correct:238 #Tested:476 Testing Accuracy:50.0%\r",
      "Progress:47.6% Speed(reviews/sec):673.7 #Correct:239 #Tested:477 Testing Accuracy:50.1%\r",
      "Progress:47.7% Speed(reviews/sec):673.7 #Correct:239 #Tested:478 Testing Accuracy:50.0%\r",
      "Progress:47.8% Speed(reviews/sec):673.3 #Correct:240 #Tested:479 Testing Accuracy:50.1%\r",
      "Progress:47.9% Speed(reviews/sec):673.5 #Correct:240 #Tested:480 Testing Accuracy:50.0%\r",
      "Progress:48.0% Speed(reviews/sec):673.7 #Correct:241 #Tested:481 Testing Accuracy:50.1%\r",
      "Progress:48.1% Speed(reviews/sec):672.7 #Correct:241 #Tested:482 Testing Accuracy:50.0%\r",
      "Progress:48.2% Speed(reviews/sec):672.9 #Correct:242 #Tested:483 Testing Accuracy:50.1%\r",
      "Progress:48.3% Speed(reviews/sec):672.9 #Correct:242 #Tested:484 Testing Accuracy:50.0%\r",
      "Progress:48.4% Speed(reviews/sec):673.1 #Correct:243 #Tested:485 Testing Accuracy:50.1%\r",
      "Progress:48.5% Speed(reviews/sec):673.0 #Correct:243 #Tested:486 Testing Accuracy:50.0%\r",
      "Progress:48.6% Speed(reviews/sec):673.3 #Correct:244 #Tested:487 Testing Accuracy:50.1%\r",
      "Progress:48.7% Speed(reviews/sec):673.6 #Correct:244 #Tested:488 Testing Accuracy:50.0%\r",
      "Progress:48.8% Speed(reviews/sec):673.7 #Correct:245 #Tested:489 Testing Accuracy:50.1%\r",
      "Progress:48.9% Speed(reviews/sec):673.2 #Correct:245 #Tested:490 Testing Accuracy:50.0%\r",
      "Progress:49.0% Speed(reviews/sec):673.3 #Correct:246 #Tested:491 Testing Accuracy:50.1%\r",
      "Progress:49.1% Speed(reviews/sec):673.6 #Correct:246 #Tested:492 Testing Accuracy:50.0%\r",
      "Progress:49.2% Speed(reviews/sec):673.6 #Correct:247 #Tested:493 Testing Accuracy:50.1%\r",
      "Progress:49.3% Speed(reviews/sec):673.6 #Correct:247 #Tested:494 Testing Accuracy:50.0%\r",
      "Progress:49.4% Speed(reviews/sec):673.7 #Correct:248 #Tested:495 Testing Accuracy:50.1%\r",
      "Progress:49.5% Speed(reviews/sec):673.6 #Correct:248 #Tested:496 Testing Accuracy:50.0%\r",
      "Progress:49.6% Speed(reviews/sec):673.8 #Correct:249 #Tested:497 Testing Accuracy:50.1%\r",
      "Progress:49.7% Speed(reviews/sec):674.1 #Correct:249 #Tested:498 Testing Accuracy:50.0%\r",
      "Progress:49.8% Speed(reviews/sec):673.8 #Correct:250 #Tested:499 Testing Accuracy:50.1%\r",
      "Progress:49.9% Speed(reviews/sec):673.7 #Correct:250 #Tested:500 Testing Accuracy:50.0%\r",
      "Progress:50.0% Speed(reviews/sec):674.0 #Correct:251 #Tested:501 Testing Accuracy:50.0%\r",
      "Progress:50.1% Speed(reviews/sec):673.6 #Correct:251 #Tested:502 Testing Accuracy:50.0%\r",
      "Progress:50.2% Speed(reviews/sec):673.5 #Correct:252 #Tested:503 Testing Accuracy:50.0%\r",
      "Progress:50.3% Speed(reviews/sec):673.6 #Correct:252 #Tested:504 Testing Accuracy:50.0%\r",
      "Progress:50.4% Speed(reviews/sec):673.6 #Correct:253 #Tested:505 Testing Accuracy:50.0%\r",
      "Progress:50.5% Speed(reviews/sec):673.9 #Correct:253 #Tested:506 Testing Accuracy:50.0%\r",
      "Progress:50.6% Speed(reviews/sec):673.1 #Correct:254 #Tested:507 Testing Accuracy:50.0%\r",
      "Progress:50.7% Speed(reviews/sec):672.0 #Correct:254 #Tested:508 Testing Accuracy:50.0%\r",
      "Progress:50.8% Speed(reviews/sec):672.3 #Correct:255 #Tested:509 Testing Accuracy:50.0%\r",
      "Progress:50.9% Speed(reviews/sec):672.5 #Correct:255 #Tested:510 Testing Accuracy:50.0%\r",
      "Progress:51.0% Speed(reviews/sec):672.0 #Correct:256 #Tested:511 Testing Accuracy:50.0%\r",
      "Progress:51.1% Speed(reviews/sec):672.2 #Correct:256 #Tested:512 Testing Accuracy:50.0%\r",
      "Progress:51.2% Speed(reviews/sec):672.1 #Correct:257 #Tested:513 Testing Accuracy:50.0%\r",
      "Progress:51.3% Speed(reviews/sec):672.4 #Correct:257 #Tested:514 Testing Accuracy:50.0%\r",
      "Progress:51.4% Speed(reviews/sec):672.4 #Correct:258 #Tested:515 Testing Accuracy:50.0%\r",
      "Progress:51.5% Speed(reviews/sec):672.3 #Correct:258 #Tested:516 Testing Accuracy:50.0%\r",
      "Progress:51.6% Speed(reviews/sec):672.0 #Correct:259 #Tested:517 Testing Accuracy:50.0%\r",
      "Progress:51.7% Speed(reviews/sec):672.1 #Correct:259 #Tested:518 Testing Accuracy:50.0%\r",
      "Progress:51.8% Speed(reviews/sec):672.2 #Correct:260 #Tested:519 Testing Accuracy:50.0%\r",
      "Progress:51.9% Speed(reviews/sec):672.3 #Correct:260 #Tested:520 Testing Accuracy:50.0%\r",
      "Progress:52.0% Speed(reviews/sec):672.6 #Correct:261 #Tested:521 Testing Accuracy:50.0%\r",
      "Progress:52.1% Speed(reviews/sec):672.7 #Correct:261 #Tested:522 Testing Accuracy:50.0%\r",
      "Progress:52.2% Speed(reviews/sec):672.8 #Correct:262 #Tested:523 Testing Accuracy:50.0%\r",
      "Progress:52.3% Speed(reviews/sec):672.7 #Correct:262 #Tested:524 Testing Accuracy:50.0%\r",
      "Progress:52.4% Speed(reviews/sec):672.9 #Correct:263 #Tested:525 Testing Accuracy:50.0%\r",
      "Progress:52.5% Speed(reviews/sec):673.2 #Correct:263 #Tested:526 Testing Accuracy:50.0%\r",
      "Progress:52.6% Speed(reviews/sec):673.5 #Correct:264 #Tested:527 Testing Accuracy:50.0%\r",
      "Progress:52.7% Speed(reviews/sec):673.6 #Correct:264 #Tested:528 Testing Accuracy:50.0%\r",
      "Progress:52.8% Speed(reviews/sec):673.4 #Correct:265 #Tested:529 Testing Accuracy:50.0%\r",
      "Progress:52.9% Speed(reviews/sec):673.4 #Correct:265 #Tested:530 Testing Accuracy:50.0%\r",
      "Progress:53.0% Speed(reviews/sec):673.3 #Correct:266 #Tested:531 Testing Accuracy:50.0%\r",
      "Progress:53.1% Speed(reviews/sec):673.5 #Correct:266 #Tested:532 Testing Accuracy:50.0%\r",
      "Progress:53.2% Speed(reviews/sec):673.7 #Correct:267 #Tested:533 Testing Accuracy:50.0%\r",
      "Progress:53.3% Speed(reviews/sec):673.7 #Correct:267 #Tested:534 Testing Accuracy:50.0%\r",
      "Progress:53.4% Speed(reviews/sec):674.0 #Correct:268 #Tested:535 Testing Accuracy:50.0%\r",
      "Progress:53.5% Speed(reviews/sec):674.2 #Correct:268 #Tested:536 Testing Accuracy:50.0%\r",
      "Progress:53.6% Speed(reviews/sec):674.4 #Correct:269 #Tested:537 Testing Accuracy:50.0%\r",
      "Progress:53.7% Speed(reviews/sec):674.6 #Correct:269 #Tested:538 Testing Accuracy:50.0%\r",
      "Progress:53.8% Speed(reviews/sec):674.7 #Correct:270 #Tested:539 Testing Accuracy:50.0%\r",
      "Progress:53.9% Speed(reviews/sec):675.0 #Correct:270 #Tested:540 Testing Accuracy:50.0%\r",
      "Progress:54.0% Speed(reviews/sec):674.9 #Correct:271 #Tested:541 Testing Accuracy:50.0%\r",
      "Progress:54.1% Speed(reviews/sec):675.0 #Correct:271 #Tested:542 Testing Accuracy:50.0%\r",
      "Progress:54.2% Speed(reviews/sec):674.9 #Correct:272 #Tested:543 Testing Accuracy:50.0%\r",
      "Progress:54.3% Speed(reviews/sec):675.0 #Correct:272 #Tested:544 Testing Accuracy:50.0%\r",
      "Progress:54.4% Speed(reviews/sec):674.9 #Correct:273 #Tested:545 Testing Accuracy:50.0%\r",
      "Progress:54.5% Speed(reviews/sec):674.8 #Correct:273 #Tested:546 Testing Accuracy:50.0%\r",
      "Progress:54.6% Speed(reviews/sec):674.6 #Correct:274 #Tested:547 Testing Accuracy:50.0%\r",
      "Progress:54.7% Speed(reviews/sec):674.5 #Correct:274 #Tested:548 Testing Accuracy:50.0%\r",
      "Progress:54.8% Speed(reviews/sec):674.8 #Correct:275 #Tested:549 Testing Accuracy:50.0%\r",
      "Progress:54.9% Speed(reviews/sec):675.1 #Correct:275 #Tested:550 Testing Accuracy:50.0%\r",
      "Progress:55.0% Speed(reviews/sec):675.3 #Correct:276 #Tested:551 Testing Accuracy:50.0%"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\r",
      "Progress:55.1% Speed(reviews/sec):675.3 #Correct:276 #Tested:552 Testing Accuracy:50.0%\r",
      "Progress:55.2% Speed(reviews/sec):675.4 #Correct:277 #Tested:553 Testing Accuracy:50.0%\r",
      "Progress:55.3% Speed(reviews/sec):675.0 #Correct:277 #Tested:554 Testing Accuracy:50.0%\r",
      "Progress:55.4% Speed(reviews/sec):673.8 #Correct:278 #Tested:555 Testing Accuracy:50.0%\r",
      "Progress:55.5% Speed(reviews/sec):673.8 #Correct:278 #Tested:556 Testing Accuracy:50.0%\r",
      "Progress:55.6% Speed(reviews/sec):673.7 #Correct:279 #Tested:557 Testing Accuracy:50.0%\r",
      "Progress:55.7% Speed(reviews/sec):673.7 #Correct:279 #Tested:558 Testing Accuracy:50.0%\r",
      "Progress:55.8% Speed(reviews/sec):673.4 #Correct:280 #Tested:559 Testing Accuracy:50.0%\r",
      "Progress:55.9% Speed(reviews/sec):673.5 #Correct:280 #Tested:560 Testing Accuracy:50.0%\r",
      "Progress:56.0% Speed(reviews/sec):673.6 #Correct:281 #Tested:561 Testing Accuracy:50.0%\r",
      "Progress:56.1% Speed(reviews/sec):673.8 #Correct:281 #Tested:562 Testing Accuracy:50.0%\r",
      "Progress:56.2% Speed(reviews/sec):673.7 #Correct:282 #Tested:563 Testing Accuracy:50.0%\r",
      "Progress:56.3% Speed(reviews/sec):673.9 #Correct:282 #Tested:564 Testing Accuracy:50.0%\r",
      "Progress:56.4% Speed(reviews/sec):673.6 #Correct:283 #Tested:565 Testing Accuracy:50.0%\r",
      "Progress:56.5% Speed(reviews/sec):673.8 #Correct:283 #Tested:566 Testing Accuracy:50.0%\r",
      "Progress:56.6% Speed(reviews/sec):673.9 #Correct:284 #Tested:567 Testing Accuracy:50.0%\r",
      "Progress:56.7% Speed(reviews/sec):673.9 #Correct:284 #Tested:568 Testing Accuracy:50.0%\r",
      "Progress:56.8% Speed(reviews/sec):674.1 #Correct:285 #Tested:569 Testing Accuracy:50.0%\r",
      "Progress:56.9% Speed(reviews/sec):673.7 #Correct:285 #Tested:570 Testing Accuracy:50.0%\r",
      "Progress:57.0% Speed(reviews/sec):672.4 #Correct:286 #Tested:571 Testing Accuracy:50.0%\r",
      "Progress:57.1% Speed(reviews/sec):672.1 #Correct:286 #Tested:572 Testing Accuracy:50.0%\r",
      "Progress:57.2% Speed(reviews/sec):671.2 #Correct:287 #Tested:573 Testing Accuracy:50.0%\r",
      "Progress:57.3% Speed(reviews/sec):670.4 #Correct:287 #Tested:574 Testing Accuracy:50.0%\r",
      "Progress:57.4% Speed(reviews/sec):670.2 #Correct:288 #Tested:575 Testing Accuracy:50.0%\r",
      "Progress:57.5% Speed(reviews/sec):670.4 #Correct:288 #Tested:576 Testing Accuracy:50.0%\r",
      "Progress:57.6% Speed(reviews/sec):670.6 #Correct:289 #Tested:577 Testing Accuracy:50.0%\r",
      "Progress:57.7% Speed(reviews/sec):670.8 #Correct:289 #Tested:578 Testing Accuracy:50.0%\r",
      "Progress:57.8% Speed(reviews/sec):670.7 #Correct:290 #Tested:579 Testing Accuracy:50.0%\r",
      "Progress:57.9% Speed(reviews/sec):670.9 #Correct:290 #Tested:580 Testing Accuracy:50.0%\r",
      "Progress:58.0% Speed(reviews/sec):670.8 #Correct:291 #Tested:581 Testing Accuracy:50.0%\r",
      "Progress:58.1% Speed(reviews/sec):671.0 #Correct:291 #Tested:582 Testing Accuracy:50.0%\r",
      "Progress:58.2% Speed(reviews/sec):671.2 #Correct:292 #Tested:583 Testing Accuracy:50.0%\r",
      "Progress:58.3% Speed(reviews/sec):670.7 #Correct:292 #Tested:584 Testing Accuracy:50.0%\r",
      "Progress:58.4% Speed(reviews/sec):670.8 #Correct:293 #Tested:585 Testing Accuracy:50.0%\r",
      "Progress:58.5% Speed(reviews/sec):671.0 #Correct:293 #Tested:586 Testing Accuracy:50.0%\r",
      "Progress:58.6% Speed(reviews/sec):671.2 #Correct:294 #Tested:587 Testing Accuracy:50.0%\r",
      "Progress:58.7% Speed(reviews/sec):671.3 #Correct:294 #Tested:588 Testing Accuracy:50.0%\r",
      "Progress:58.8% Speed(reviews/sec):671.6 #Correct:295 #Tested:589 Testing Accuracy:50.0%\r",
      "Progress:58.9% Speed(reviews/sec):671.8 #Correct:295 #Tested:590 Testing Accuracy:50.0%\r",
      "Progress:59.0% Speed(reviews/sec):672.0 #Correct:296 #Tested:591 Testing Accuracy:50.0%\r",
      "Progress:59.1% Speed(reviews/sec):672.1 #Correct:296 #Tested:592 Testing Accuracy:50.0%\r",
      "Progress:59.2% Speed(reviews/sec):672.0 #Correct:297 #Tested:593 Testing Accuracy:50.0%\r",
      "Progress:59.3% Speed(reviews/sec):672.1 #Correct:297 #Tested:594 Testing Accuracy:50.0%\r",
      "Progress:59.4% Speed(reviews/sec):672.2 #Correct:298 #Tested:595 Testing Accuracy:50.0%\r",
      "Progress:59.5% Speed(reviews/sec):672.3 #Correct:298 #Tested:596 Testing Accuracy:50.0%\r",
      "Progress:59.6% Speed(reviews/sec):672.6 #Correct:299 #Tested:597 Testing Accuracy:50.0%\r",
      "Progress:59.7% Speed(reviews/sec):672.7 #Correct:299 #Tested:598 Testing Accuracy:50.0%\r",
      "Progress:59.8% Speed(reviews/sec):672.7 #Correct:300 #Tested:599 Testing Accuracy:50.0%\r",
      "Progress:59.9% Speed(reviews/sec):672.8 #Correct:300 #Tested:600 Testing Accuracy:50.0%\r",
      "Progress:60.0% Speed(reviews/sec):672.8 #Correct:301 #Tested:601 Testing Accuracy:50.0%\r",
      "Progress:60.1% Speed(reviews/sec):673.0 #Correct:301 #Tested:602 Testing Accuracy:50.0%\r",
      "Progress:60.2% Speed(reviews/sec):673.1 #Correct:302 #Tested:603 Testing Accuracy:50.0%\r",
      "Progress:60.3% Speed(reviews/sec):673.3 #Correct:302 #Tested:604 Testing Accuracy:50.0%\r",
      "Progress:60.4% Speed(reviews/sec):673.5 #Correct:303 #Tested:605 Testing Accuracy:50.0%\r",
      "Progress:60.5% Speed(reviews/sec):672.9 #Correct:303 #Tested:606 Testing Accuracy:50.0%\r",
      "Progress:60.6% Speed(reviews/sec):673.0 #Correct:304 #Tested:607 Testing Accuracy:50.0%\r",
      "Progress:60.7% Speed(reviews/sec):673.1 #Correct:304 #Tested:608 Testing Accuracy:50.0%\r",
      "Progress:60.8% Speed(reviews/sec):673.3 #Correct:305 #Tested:609 Testing Accuracy:50.0%\r",
      "Progress:60.9% Speed(reviews/sec):673.6 #Correct:305 #Tested:610 Testing Accuracy:50.0%\r",
      "Progress:61.0% Speed(reviews/sec):673.5 #Correct:306 #Tested:611 Testing Accuracy:50.0%\r",
      "Progress:61.1% Speed(reviews/sec):673.6 #Correct:306 #Tested:612 Testing Accuracy:50.0%\r",
      "Progress:61.2% Speed(reviews/sec):673.9 #Correct:307 #Tested:613 Testing Accuracy:50.0%\r",
      "Progress:61.3% Speed(reviews/sec):674.1 #Correct:307 #Tested:614 Testing Accuracy:50.0%\r",
      "Progress:61.4% Speed(reviews/sec):674.3 #Correct:308 #Tested:615 Testing Accuracy:50.0%\r",
      "Progress:61.5% Speed(reviews/sec):674.4 #Correct:308 #Tested:616 Testing Accuracy:50.0%\r",
      "Progress:61.6% Speed(reviews/sec):674.4 #Correct:309 #Tested:617 Testing Accuracy:50.0%\r",
      "Progress:61.7% Speed(reviews/sec):674.6 #Correct:309 #Tested:618 Testing Accuracy:50.0%\r",
      "Progress:61.8% Speed(reviews/sec):674.8 #Correct:310 #Tested:619 Testing Accuracy:50.0%\r",
      "Progress:61.9% Speed(reviews/sec):674.9 #Correct:310 #Tested:620 Testing Accuracy:50.0%\r",
      "Progress:62.0% Speed(reviews/sec):674.8 #Correct:311 #Tested:621 Testing Accuracy:50.0%\r",
      "Progress:62.1% Speed(reviews/sec):675.0 #Correct:311 #Tested:622 Testing Accuracy:50.0%\r",
      "Progress:62.2% Speed(reviews/sec):675.1 #Correct:312 #Tested:623 Testing Accuracy:50.0%\r",
      "Progress:62.3% Speed(reviews/sec):675.3 #Correct:312 #Tested:624 Testing Accuracy:50.0%\r",
      "Progress:62.4% Speed(reviews/sec):675.4 #Correct:313 #Tested:625 Testing Accuracy:50.0%\r",
      "Progress:62.5% Speed(reviews/sec):674.7 #Correct:313 #Tested:626 Testing Accuracy:50.0%\r",
      "Progress:62.6% Speed(reviews/sec):674.6 #Correct:314 #Tested:627 Testing Accuracy:50.0%\r",
      "Progress:62.7% Speed(reviews/sec):674.8 #Correct:314 #Tested:628 Testing Accuracy:50.0%\r",
      "Progress:62.8% Speed(reviews/sec):675.1 #Correct:315 #Tested:629 Testing Accuracy:50.0%\r",
      "Progress:62.9% Speed(reviews/sec):674.9 #Correct:315 #Tested:630 Testing Accuracy:50.0%\r",
      "Progress:63.0% Speed(reviews/sec):675.0 #Correct:316 #Tested:631 Testing Accuracy:50.0%\r",
      "Progress:63.1% Speed(reviews/sec):674.5 #Correct:316 #Tested:632 Testing Accuracy:50.0%\r",
      "Progress:63.2% Speed(reviews/sec):674.7 #Correct:317 #Tested:633 Testing Accuracy:50.0%\r",
      "Progress:63.3% Speed(reviews/sec):674.7 #Correct:317 #Tested:634 Testing Accuracy:50.0%\r",
      "Progress:63.4% Speed(reviews/sec):674.4 #Correct:318 #Tested:635 Testing Accuracy:50.0%\r",
      "Progress:63.5% Speed(reviews/sec):674.3 #Correct:318 #Tested:636 Testing Accuracy:50.0%\r",
      "Progress:63.6% Speed(reviews/sec):674.4 #Correct:319 #Tested:637 Testing Accuracy:50.0%\r",
      "Progress:63.7% Speed(reviews/sec):674.5 #Correct:319 #Tested:638 Testing Accuracy:50.0%\r",
      "Progress:63.8% Speed(reviews/sec):674.6 #Correct:320 #Tested:639 Testing Accuracy:50.0%\r",
      "Progress:63.9% Speed(reviews/sec):674.8 #Correct:320 #Tested:640 Testing Accuracy:50.0%\r",
      "Progress:64.0% Speed(reviews/sec):675.1 #Correct:321 #Tested:641 Testing Accuracy:50.0%\r",
      "Progress:64.1% Speed(reviews/sec):675.0 #Correct:321 #Tested:642 Testing Accuracy:50.0%\r",
      "Progress:64.2% Speed(reviews/sec):674.9 #Correct:322 #Tested:643 Testing Accuracy:50.0%\r",
      "Progress:64.3% Speed(reviews/sec):675.0 #Correct:322 #Tested:644 Testing Accuracy:50.0%\r",
      "Progress:64.4% Speed(reviews/sec):674.9 #Correct:323 #Tested:645 Testing Accuracy:50.0%\r",
      "Progress:64.5% Speed(reviews/sec):674.8 #Correct:323 #Tested:646 Testing Accuracy:50.0%\r",
      "Progress:64.6% Speed(reviews/sec):674.6 #Correct:324 #Tested:647 Testing Accuracy:50.0%\r",
      "Progress:64.7% Speed(reviews/sec):674.8 #Correct:324 #Tested:648 Testing Accuracy:50.0%\r",
      "Progress:64.8% Speed(reviews/sec):675.0 #Correct:325 #Tested:649 Testing Accuracy:50.0%\r",
      "Progress:64.9% Speed(reviews/sec):675.1 #Correct:325 #Tested:650 Testing Accuracy:50.0%\r",
      "Progress:65.0% Speed(reviews/sec):675.3 #Correct:326 #Tested:651 Testing Accuracy:50.0%\r",
      "Progress:65.1% Speed(reviews/sec):675.5 #Correct:326 #Tested:652 Testing Accuracy:50.0%\r",
      "Progress:65.2% Speed(reviews/sec):675.7 #Correct:327 #Tested:653 Testing Accuracy:50.0%\r",
      "Progress:65.3% Speed(reviews/sec):675.8 #Correct:327 #Tested:654 Testing Accuracy:50.0%\r",
      "Progress:65.4% Speed(reviews/sec):675.9 #Correct:328 #Tested:655 Testing Accuracy:50.0%\r",
      "Progress:65.5% Speed(reviews/sec):676.0 #Correct:328 #Tested:656 Testing Accuracy:50.0%\r",
      "Progress:65.6% Speed(reviews/sec):676.2 #Correct:329 #Tested:657 Testing Accuracy:50.0%\r",
      "Progress:65.7% Speed(reviews/sec):676.2 #Correct:329 #Tested:658 Testing Accuracy:50.0%\r",
      "Progress:65.8% Speed(reviews/sec):676.5 #Correct:330 #Tested:659 Testing Accuracy:50.0%\r",
      "Progress:65.9% Speed(reviews/sec):676.5 #Correct:330 #Tested:660 Testing Accuracy:50.0%\r",
      "Progress:66.0% Speed(reviews/sec):676.8 #Correct:331 #Tested:661 Testing Accuracy:50.0%\r",
      "Progress:66.1% Speed(reviews/sec):676.9 #Correct:331 #Tested:662 Testing Accuracy:50.0%\r",
      "Progress:66.2% Speed(reviews/sec):677.0 #Correct:332 #Tested:663 Testing Accuracy:50.0%\r",
      "Progress:66.3% Speed(reviews/sec):677.1 #Correct:332 #Tested:664 Testing Accuracy:50.0%\r",
      "Progress:66.4% Speed(reviews/sec):677.2 #Correct:333 #Tested:665 Testing Accuracy:50.0%\r",
      "Progress:66.5% Speed(reviews/sec):677.4 #Correct:333 #Tested:666 Testing Accuracy:50.0%\r",
      "Progress:66.6% Speed(reviews/sec):677.6 #Correct:334 #Tested:667 Testing Accuracy:50.0%\r",
      "Progress:66.7% Speed(reviews/sec):677.7 #Correct:334 #Tested:668 Testing Accuracy:50.0%\r",
      "Progress:66.8% Speed(reviews/sec):677.7 #Correct:335 #Tested:669 Testing Accuracy:50.0%\r",
      "Progress:66.9% Speed(reviews/sec):677.8 #Correct:335 #Tested:670 Testing Accuracy:50.0%\r",
      "Progress:67.0% Speed(reviews/sec):678.1 #Correct:336 #Tested:671 Testing Accuracy:50.0%\r",
      "Progress:67.1% Speed(reviews/sec):678.2 #Correct:336 #Tested:672 Testing Accuracy:50.0%\r",
      "Progress:67.2% Speed(reviews/sec):678.4 #Correct:337 #Tested:673 Testing Accuracy:50.0%\r",
      "Progress:67.3% Speed(reviews/sec):678.4 #Correct:337 #Tested:674 Testing Accuracy:50.0%\r",
      "Progress:67.4% Speed(reviews/sec):678.6 #Correct:338 #Tested:675 Testing Accuracy:50.0%\r",
      "Progress:67.5% Speed(reviews/sec):678.7 #Correct:338 #Tested:676 Testing Accuracy:50.0%\r",
      "Progress:67.6% Speed(reviews/sec):678.9 #Correct:339 #Tested:677 Testing Accuracy:50.0%\r",
      "Progress:67.7% Speed(reviews/sec):678.9 #Correct:339 #Tested:678 Testing Accuracy:50.0%\r",
      "Progress:67.8% Speed(reviews/sec):679.1 #Correct:340 #Tested:679 Testing Accuracy:50.0%\r",
      "Progress:67.9% Speed(reviews/sec):679.2 #Correct:340 #Tested:680 Testing Accuracy:50.0%\r",
      "Progress:68.0% Speed(reviews/sec):679.2 #Correct:341 #Tested:681 Testing Accuracy:50.0%\r",
      "Progress:68.1% Speed(reviews/sec):679.3 #Correct:341 #Tested:682 Testing Accuracy:50.0%\r",
      "Progress:68.2% Speed(reviews/sec):679.4 #Correct:342 #Tested:683 Testing Accuracy:50.0%\r",
      "Progress:68.3% Speed(reviews/sec):679.7 #Correct:342 #Tested:684 Testing Accuracy:50.0%\r",
      "Progress:68.4% Speed(reviews/sec):679.8 #Correct:343 #Tested:685 Testing Accuracy:50.0%\r",
      "Progress:68.5% Speed(reviews/sec):679.9 #Correct:343 #Tested:686 Testing Accuracy:50.0%\r",
      "Progress:68.6% Speed(reviews/sec):680.1 #Correct:344 #Tested:687 Testing Accuracy:50.0%\r",
      "Progress:68.7% Speed(reviews/sec):680.2 #Correct:344 #Tested:688 Testing Accuracy:50.0%\r",
      "Progress:68.8% Speed(reviews/sec):679.9 #Correct:345 #Tested:689 Testing Accuracy:50.0%\r",
      "Progress:68.9% Speed(reviews/sec):680.0 #Correct:345 #Tested:690 Testing Accuracy:50.0%\r",
      "Progress:69.0% Speed(reviews/sec):680.0 #Correct:346 #Tested:691 Testing Accuracy:50.0%\r",
      "Progress:69.1% Speed(reviews/sec):680.2 #Correct:346 #Tested:692 Testing Accuracy:50.0%"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\r",
      "Progress:69.2% Speed(reviews/sec):680.0 #Correct:347 #Tested:693 Testing Accuracy:50.0%\r",
      "Progress:69.3% Speed(reviews/sec):680.0 #Correct:347 #Tested:694 Testing Accuracy:50.0%\r",
      "Progress:69.4% Speed(reviews/sec):679.8 #Correct:348 #Tested:695 Testing Accuracy:50.0%\r",
      "Progress:69.5% Speed(reviews/sec):679.3 #Correct:348 #Tested:696 Testing Accuracy:50.0%\r",
      "Progress:69.6% Speed(reviews/sec):679.5 #Correct:349 #Tested:697 Testing Accuracy:50.0%\r",
      "Progress:69.7% Speed(reviews/sec):679.4 #Correct:349 #Tested:698 Testing Accuracy:50.0%\r",
      "Progress:69.8% Speed(reviews/sec):679.5 #Correct:350 #Tested:699 Testing Accuracy:50.0%\r",
      "Progress:69.9% Speed(reviews/sec):679.5 #Correct:350 #Tested:700 Testing Accuracy:50.0%\r",
      "Progress:70.0% Speed(reviews/sec):679.3 #Correct:351 #Tested:701 Testing Accuracy:50.0%\r",
      "Progress:70.1% Speed(reviews/sec):679.1 #Correct:351 #Tested:702 Testing Accuracy:50.0%\r",
      "Progress:70.2% Speed(reviews/sec):678.9 #Correct:352 #Tested:703 Testing Accuracy:50.0%\r",
      "Progress:70.3% Speed(reviews/sec):678.6 #Correct:352 #Tested:704 Testing Accuracy:50.0%\r",
      "Progress:70.4% Speed(reviews/sec):678.4 #Correct:353 #Tested:705 Testing Accuracy:50.0%\r",
      "Progress:70.5% Speed(reviews/sec):678.3 #Correct:353 #Tested:706 Testing Accuracy:50.0%\r",
      "Progress:70.6% Speed(reviews/sec):678.3 #Correct:354 #Tested:707 Testing Accuracy:50.0%\r",
      "Progress:70.7% Speed(reviews/sec):678.3 #Correct:354 #Tested:708 Testing Accuracy:50.0%\r",
      "Progress:70.8% Speed(reviews/sec):678.2 #Correct:355 #Tested:709 Testing Accuracy:50.0%\r",
      "Progress:70.9% Speed(reviews/sec):678.0 #Correct:355 #Tested:710 Testing Accuracy:50.0%\r",
      "Progress:71.0% Speed(reviews/sec):678.0 #Correct:356 #Tested:711 Testing Accuracy:50.0%\r",
      "Progress:71.1% Speed(reviews/sec):678.1 #Correct:356 #Tested:712 Testing Accuracy:50.0%\r",
      "Progress:71.2% Speed(reviews/sec):678.2 #Correct:357 #Tested:713 Testing Accuracy:50.0%\r",
      "Progress:71.3% Speed(reviews/sec):677.9 #Correct:357 #Tested:714 Testing Accuracy:50.0%\r",
      "Progress:71.4% Speed(reviews/sec):677.0 #Correct:358 #Tested:715 Testing Accuracy:50.0%\r",
      "Progress:71.5% Speed(reviews/sec):677.0 #Correct:358 #Tested:716 Testing Accuracy:50.0%\r",
      "Progress:71.6% Speed(reviews/sec):677.2 #Correct:359 #Tested:717 Testing Accuracy:50.0%\r",
      "Progress:71.7% Speed(reviews/sec):677.3 #Correct:359 #Tested:718 Testing Accuracy:50.0%\r",
      "Progress:71.8% Speed(reviews/sec):677.4 #Correct:360 #Tested:719 Testing Accuracy:50.0%\r",
      "Progress:71.9% Speed(reviews/sec):677.5 #Correct:360 #Tested:720 Testing Accuracy:50.0%\r",
      "Progress:72.0% Speed(reviews/sec):677.7 #Correct:361 #Tested:721 Testing Accuracy:50.0%\r",
      "Progress:72.1% Speed(reviews/sec):677.8 #Correct:361 #Tested:722 Testing Accuracy:50.0%\r",
      "Progress:72.2% Speed(reviews/sec):677.8 #Correct:362 #Tested:723 Testing Accuracy:50.0%\r",
      "Progress:72.3% Speed(reviews/sec):677.9 #Correct:362 #Tested:724 Testing Accuracy:50.0%\r",
      "Progress:72.4% Speed(reviews/sec):678.0 #Correct:363 #Tested:725 Testing Accuracy:50.0%\r",
      "Progress:72.5% Speed(reviews/sec):677.7 #Correct:363 #Tested:726 Testing Accuracy:50.0%\r",
      "Progress:72.6% Speed(reviews/sec):677.5 #Correct:364 #Tested:727 Testing Accuracy:50.0%\r",
      "Progress:72.7% Speed(reviews/sec):677.3 #Correct:364 #Tested:728 Testing Accuracy:50.0%\r",
      "Progress:72.8% Speed(reviews/sec):677.5 #Correct:365 #Tested:729 Testing Accuracy:50.0%\r",
      "Progress:72.9% Speed(reviews/sec):677.2 #Correct:365 #Tested:730 Testing Accuracy:50.0%\r",
      "Progress:73.0% Speed(reviews/sec):677.2 #Correct:366 #Tested:731 Testing Accuracy:50.0%\r",
      "Progress:73.1% Speed(reviews/sec):677.4 #Correct:366 #Tested:732 Testing Accuracy:50.0%\r",
      "Progress:73.2% Speed(reviews/sec):677.3 #Correct:367 #Tested:733 Testing Accuracy:50.0%\r",
      "Progress:73.3% Speed(reviews/sec):677.2 #Correct:367 #Tested:734 Testing Accuracy:50.0%\r",
      "Progress:73.4% Speed(reviews/sec):677.1 #Correct:368 #Tested:735 Testing Accuracy:50.0%\r",
      "Progress:73.5% Speed(reviews/sec):677.1 #Correct:368 #Tested:736 Testing Accuracy:50.0%\r",
      "Progress:73.6% Speed(reviews/sec):677.2 #Correct:369 #Tested:737 Testing Accuracy:50.0%\r",
      "Progress:73.7% Speed(reviews/sec):677.3 #Correct:369 #Tested:738 Testing Accuracy:50.0%\r",
      "Progress:73.8% Speed(reviews/sec):677.3 #Correct:370 #Tested:739 Testing Accuracy:50.0%\r",
      "Progress:73.9% Speed(reviews/sec):677.4 #Correct:370 #Tested:740 Testing Accuracy:50.0%\r",
      "Progress:74.0% Speed(reviews/sec):677.6 #Correct:371 #Tested:741 Testing Accuracy:50.0%\r",
      "Progress:74.1% Speed(reviews/sec):677.8 #Correct:371 #Tested:742 Testing Accuracy:50.0%\r",
      "Progress:74.2% Speed(reviews/sec):677.9 #Correct:372 #Tested:743 Testing Accuracy:50.0%\r",
      "Progress:74.3% Speed(reviews/sec):677.9 #Correct:372 #Tested:744 Testing Accuracy:50.0%\r",
      "Progress:74.4% Speed(reviews/sec):678.0 #Correct:373 #Tested:745 Testing Accuracy:50.0%\r",
      "Progress:74.5% Speed(reviews/sec):677.8 #Correct:373 #Tested:746 Testing Accuracy:50.0%\r",
      "Progress:74.6% Speed(reviews/sec):677.8 #Correct:374 #Tested:747 Testing Accuracy:50.0%\r",
      "Progress:74.7% Speed(reviews/sec):677.4 #Correct:374 #Tested:748 Testing Accuracy:50.0%\r",
      "Progress:74.8% Speed(reviews/sec):677.4 #Correct:375 #Tested:749 Testing Accuracy:50.0%\r",
      "Progress:74.9% Speed(reviews/sec):677.3 #Correct:375 #Tested:750 Testing Accuracy:50.0%\r",
      "Progress:75.0% Speed(reviews/sec):677.5 #Correct:376 #Tested:751 Testing Accuracy:50.0%\r",
      "Progress:75.1% Speed(reviews/sec):677.6 #Correct:376 #Tested:752 Testing Accuracy:50.0%\r",
      "Progress:75.2% Speed(reviews/sec):677.7 #Correct:377 #Tested:753 Testing Accuracy:50.0%\r",
      "Progress:75.3% Speed(reviews/sec):677.8 #Correct:377 #Tested:754 Testing Accuracy:50.0%\r",
      "Progress:75.4% Speed(reviews/sec):677.7 #Correct:378 #Tested:755 Testing Accuracy:50.0%\r",
      "Progress:75.5% Speed(reviews/sec):677.9 #Correct:378 #Tested:756 Testing Accuracy:50.0%\r",
      "Progress:75.6% Speed(reviews/sec):678.0 #Correct:379 #Tested:757 Testing Accuracy:50.0%\r",
      "Progress:75.7% Speed(reviews/sec):678.1 #Correct:379 #Tested:758 Testing Accuracy:50.0%\r",
      "Progress:75.8% Speed(reviews/sec):678.1 #Correct:380 #Tested:759 Testing Accuracy:50.0%\r",
      "Progress:75.9% Speed(reviews/sec):678.2 #Correct:380 #Tested:760 Testing Accuracy:50.0%\r",
      "Progress:76.0% Speed(reviews/sec):678.2 #Correct:381 #Tested:761 Testing Accuracy:50.0%\r",
      "Progress:76.1% Speed(reviews/sec):678.2 #Correct:381 #Tested:762 Testing Accuracy:50.0%\r",
      "Progress:76.2% Speed(reviews/sec):678.3 #Correct:382 #Tested:763 Testing Accuracy:50.0%\r",
      "Progress:76.3% Speed(reviews/sec):678.4 #Correct:382 #Tested:764 Testing Accuracy:50.0%\r",
      "Progress:76.4% Speed(reviews/sec):678.4 #Correct:383 #Tested:765 Testing Accuracy:50.0%\r",
      "Progress:76.5% Speed(reviews/sec):678.4 #Correct:383 #Tested:766 Testing Accuracy:50.0%\r",
      "Progress:76.6% Speed(reviews/sec):678.5 #Correct:384 #Tested:767 Testing Accuracy:50.0%\r",
      "Progress:76.7% Speed(reviews/sec):678.6 #Correct:384 #Tested:768 Testing Accuracy:50.0%\r",
      "Progress:76.8% Speed(reviews/sec):678.7 #Correct:385 #Tested:769 Testing Accuracy:50.0%\r",
      "Progress:76.9% Speed(reviews/sec):678.9 #Correct:385 #Tested:770 Testing Accuracy:50.0%\r",
      "Progress:77.0% Speed(reviews/sec):678.6 #Correct:386 #Tested:771 Testing Accuracy:50.0%\r",
      "Progress:77.1% Speed(reviews/sec):678.7 #Correct:386 #Tested:772 Testing Accuracy:50.0%\r",
      "Progress:77.2% Speed(reviews/sec):678.8 #Correct:387 #Tested:773 Testing Accuracy:50.0%\r",
      "Progress:77.3% Speed(reviews/sec):678.8 #Correct:387 #Tested:774 Testing Accuracy:50.0%\r",
      "Progress:77.4% Speed(reviews/sec):678.9 #Correct:388 #Tested:775 Testing Accuracy:50.0%\r",
      "Progress:77.5% Speed(reviews/sec):678.8 #Correct:388 #Tested:776 Testing Accuracy:50.0%\r",
      "Progress:77.6% Speed(reviews/sec):678.8 #Correct:389 #Tested:777 Testing Accuracy:50.0%\r",
      "Progress:77.7% Speed(reviews/sec):678.8 #Correct:389 #Tested:778 Testing Accuracy:50.0%\r",
      "Progress:77.8% Speed(reviews/sec):678.5 #Correct:390 #Tested:779 Testing Accuracy:50.0%\r",
      "Progress:77.9% Speed(reviews/sec):678.6 #Correct:390 #Tested:780 Testing Accuracy:50.0%\r",
      "Progress:78.0% Speed(reviews/sec):678.1 #Correct:391 #Tested:781 Testing Accuracy:50.0%\r",
      "Progress:78.1% Speed(reviews/sec):678.3 #Correct:391 #Tested:782 Testing Accuracy:50.0%\r",
      "Progress:78.2% Speed(reviews/sec):678.2 #Correct:392 #Tested:783 Testing Accuracy:50.0%\r",
      "Progress:78.3% Speed(reviews/sec):678.4 #Correct:392 #Tested:784 Testing Accuracy:50.0%\r",
      "Progress:78.4% Speed(reviews/sec):678.3 #Correct:393 #Tested:785 Testing Accuracy:50.0%\r",
      "Progress:78.5% Speed(reviews/sec):678.4 #Correct:393 #Tested:786 Testing Accuracy:50.0%\r",
      "Progress:78.6% Speed(reviews/sec):678.4 #Correct:394 #Tested:787 Testing Accuracy:50.0%\r",
      "Progress:78.7% Speed(reviews/sec):678.3 #Correct:394 #Tested:788 Testing Accuracy:50.0%\r",
      "Progress:78.8% Speed(reviews/sec):678.5 #Correct:395 #Tested:789 Testing Accuracy:50.0%\r",
      "Progress:78.9% Speed(reviews/sec):678.3 #Correct:395 #Tested:790 Testing Accuracy:50.0%\r",
      "Progress:79.0% Speed(reviews/sec):678.2 #Correct:396 #Tested:791 Testing Accuracy:50.0%\r",
      "Progress:79.1% Speed(reviews/sec):678.1 #Correct:396 #Tested:792 Testing Accuracy:50.0%\r",
      "Progress:79.2% Speed(reviews/sec):677.9 #Correct:397 #Tested:793 Testing Accuracy:50.0%\r",
      "Progress:79.3% Speed(reviews/sec):677.9 #Correct:397 #Tested:794 Testing Accuracy:50.0%\r",
      "Progress:79.4% Speed(reviews/sec):678.0 #Correct:398 #Tested:795 Testing Accuracy:50.0%\r",
      "Progress:79.5% Speed(reviews/sec):678.2 #Correct:398 #Tested:796 Testing Accuracy:50.0%\r",
      "Progress:79.6% Speed(reviews/sec):678.3 #Correct:399 #Tested:797 Testing Accuracy:50.0%\r",
      "Progress:79.7% Speed(reviews/sec):678.4 #Correct:399 #Tested:798 Testing Accuracy:50.0%\r",
      "Progress:79.8% Speed(reviews/sec):678.7 #Correct:400 #Tested:799 Testing Accuracy:50.0%\r",
      "Progress:79.9% Speed(reviews/sec):678.8 #Correct:400 #Tested:800 Testing Accuracy:50.0%\r",
      "Progress:80.0% Speed(reviews/sec):679.0 #Correct:401 #Tested:801 Testing Accuracy:50.0%\r",
      "Progress:80.1% Speed(reviews/sec):679.2 #Correct:401 #Tested:802 Testing Accuracy:50.0%\r",
      "Progress:80.2% Speed(reviews/sec):679.3 #Correct:402 #Tested:803 Testing Accuracy:50.0%\r",
      "Progress:80.3% Speed(reviews/sec):679.4 #Correct:402 #Tested:804 Testing Accuracy:50.0%\r",
      "Progress:80.4% Speed(reviews/sec):679.6 #Correct:403 #Tested:805 Testing Accuracy:50.0%\r",
      "Progress:80.5% Speed(reviews/sec):679.7 #Correct:403 #Tested:806 Testing Accuracy:50.0%\r",
      "Progress:80.6% Speed(reviews/sec):679.7 #Correct:404 #Tested:807 Testing Accuracy:50.0%\r",
      "Progress:80.7% Speed(reviews/sec):679.6 #Correct:404 #Tested:808 Testing Accuracy:50.0%\r",
      "Progress:80.8% Speed(reviews/sec):679.8 #Correct:405 #Tested:809 Testing Accuracy:50.0%\r",
      "Progress:80.9% Speed(reviews/sec):679.7 #Correct:405 #Tested:810 Testing Accuracy:50.0%\r",
      "Progress:81.0% Speed(reviews/sec):679.6 #Correct:406 #Tested:811 Testing Accuracy:50.0%\r",
      "Progress:81.1% Speed(reviews/sec):679.6 #Correct:406 #Tested:812 Testing Accuracy:50.0%\r",
      "Progress:81.2% Speed(reviews/sec):679.7 #Correct:407 #Tested:813 Testing Accuracy:50.0%\r",
      "Progress:81.3% Speed(reviews/sec):679.8 #Correct:407 #Tested:814 Testing Accuracy:50.0%\r",
      "Progress:81.4% Speed(reviews/sec):679.6 #Correct:408 #Tested:815 Testing Accuracy:50.0%\r",
      "Progress:81.5% Speed(reviews/sec):679.8 #Correct:408 #Tested:816 Testing Accuracy:50.0%\r",
      "Progress:81.6% Speed(reviews/sec):679.9 #Correct:409 #Tested:817 Testing Accuracy:50.0%\r",
      "Progress:81.7% Speed(reviews/sec):679.9 #Correct:409 #Tested:818 Testing Accuracy:50.0%\r",
      "Progress:81.8% Speed(reviews/sec):679.8 #Correct:410 #Tested:819 Testing Accuracy:50.0%\r",
      "Progress:81.9% Speed(reviews/sec):680.0 #Correct:410 #Tested:820 Testing Accuracy:50.0%\r",
      "Progress:82.0% Speed(reviews/sec):679.6 #Correct:411 #Tested:821 Testing Accuracy:50.0%\r",
      "Progress:82.1% Speed(reviews/sec):679.6 #Correct:411 #Tested:822 Testing Accuracy:50.0%\r",
      "Progress:82.2% Speed(reviews/sec):679.7 #Correct:412 #Tested:823 Testing Accuracy:50.0%\r",
      "Progress:82.3% Speed(reviews/sec):679.9 #Correct:412 #Tested:824 Testing Accuracy:50.0%\r",
      "Progress:82.4% Speed(reviews/sec):680.0 #Correct:413 #Tested:825 Testing Accuracy:50.0%\r",
      "Progress:82.5% Speed(reviews/sec):680.0 #Correct:413 #Tested:826 Testing Accuracy:50.0%\r",
      "Progress:82.6% Speed(reviews/sec):679.3 #Correct:414 #Tested:827 Testing Accuracy:50.0%\r",
      "Progress:82.7% Speed(reviews/sec):679.4 #Correct:414 #Tested:828 Testing Accuracy:50.0%\r",
      "Progress:82.8% Speed(reviews/sec):679.3 #Correct:415 #Tested:829 Testing Accuracy:50.0%"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\r",
      "Progress:82.9% Speed(reviews/sec):679.2 #Correct:415 #Tested:830 Testing Accuracy:50.0%\r",
      "Progress:83.0% Speed(reviews/sec):679.2 #Correct:416 #Tested:831 Testing Accuracy:50.0%\r",
      "Progress:83.1% Speed(reviews/sec):678.9 #Correct:416 #Tested:832 Testing Accuracy:50.0%\r",
      "Progress:83.2% Speed(reviews/sec):678.7 #Correct:417 #Tested:833 Testing Accuracy:50.0%\r",
      "Progress:83.3% Speed(reviews/sec):678.7 #Correct:417 #Tested:834 Testing Accuracy:50.0%\r",
      "Progress:83.4% Speed(reviews/sec):678.4 #Correct:418 #Tested:835 Testing Accuracy:50.0%\r",
      "Progress:83.5% Speed(reviews/sec):678.5 #Correct:418 #Tested:836 Testing Accuracy:50.0%\r",
      "Progress:83.6% Speed(reviews/sec):677.8 #Correct:419 #Tested:837 Testing Accuracy:50.0%\r",
      "Progress:83.7% Speed(reviews/sec):677.1 #Correct:419 #Tested:838 Testing Accuracy:50.0%\r",
      "Progress:83.8% Speed(reviews/sec):677.1 #Correct:420 #Tested:839 Testing Accuracy:50.0%\r",
      "Progress:83.9% Speed(reviews/sec):677.1 #Correct:420 #Tested:840 Testing Accuracy:50.0%\r",
      "Progress:84.0% Speed(reviews/sec):677.1 #Correct:421 #Tested:841 Testing Accuracy:50.0%\r",
      "Progress:84.1% Speed(reviews/sec):677.0 #Correct:421 #Tested:842 Testing Accuracy:50.0%\r",
      "Progress:84.2% Speed(reviews/sec):676.9 #Correct:422 #Tested:843 Testing Accuracy:50.0%\r",
      "Progress:84.3% Speed(reviews/sec):676.9 #Correct:422 #Tested:844 Testing Accuracy:50.0%\r",
      "Progress:84.4% Speed(reviews/sec):677.0 #Correct:423 #Tested:845 Testing Accuracy:50.0%\r",
      "Progress:84.5% Speed(reviews/sec):677.0 #Correct:423 #Tested:846 Testing Accuracy:50.0%\r",
      "Progress:84.6% Speed(reviews/sec):677.0 #Correct:424 #Tested:847 Testing Accuracy:50.0%\r",
      "Progress:84.7% Speed(reviews/sec):677.1 #Correct:424 #Tested:848 Testing Accuracy:50.0%\r",
      "Progress:84.8% Speed(reviews/sec):676.9 #Correct:425 #Tested:849 Testing Accuracy:50.0%\r",
      "Progress:84.9% Speed(reviews/sec):675.3 #Correct:425 #Tested:850 Testing Accuracy:50.0%\r",
      "Progress:85.0% Speed(reviews/sec):675.0 #Correct:426 #Tested:851 Testing Accuracy:50.0%\r",
      "Progress:85.1% Speed(reviews/sec):675.1 #Correct:426 #Tested:852 Testing Accuracy:50.0%\r",
      "Progress:85.2% Speed(reviews/sec):675.0 #Correct:427 #Tested:853 Testing Accuracy:50.0%\r",
      "Progress:85.3% Speed(reviews/sec):675.0 #Correct:427 #Tested:854 Testing Accuracy:50.0%\r",
      "Progress:85.4% Speed(reviews/sec):675.1 #Correct:428 #Tested:855 Testing Accuracy:50.0%\r",
      "Progress:85.5% Speed(reviews/sec):675.3 #Correct:428 #Tested:856 Testing Accuracy:50.0%\r",
      "Progress:85.6% Speed(reviews/sec):675.3 #Correct:429 #Tested:857 Testing Accuracy:50.0%\r",
      "Progress:85.7% Speed(reviews/sec):674.9 #Correct:429 #Tested:858 Testing Accuracy:50.0%\r",
      "Progress:85.8% Speed(reviews/sec):674.8 #Correct:430 #Tested:859 Testing Accuracy:50.0%\r",
      "Progress:85.9% Speed(reviews/sec):674.9 #Correct:430 #Tested:860 Testing Accuracy:50.0%\r",
      "Progress:86.0% Speed(reviews/sec):675.0 #Correct:431 #Tested:861 Testing Accuracy:50.0%\r",
      "Progress:86.1% Speed(reviews/sec):675.1 #Correct:431 #Tested:862 Testing Accuracy:50.0%\r",
      "Progress:86.2% Speed(reviews/sec):675.2 #Correct:432 #Tested:863 Testing Accuracy:50.0%\r",
      "Progress:86.3% Speed(reviews/sec):674.6 #Correct:432 #Tested:864 Testing Accuracy:50.0%\r",
      "Progress:86.4% Speed(reviews/sec):674.7 #Correct:433 #Tested:865 Testing Accuracy:50.0%\r",
      "Progress:86.5% Speed(reviews/sec):674.8 #Correct:433 #Tested:866 Testing Accuracy:50.0%\r",
      "Progress:86.6% Speed(reviews/sec):674.8 #Correct:434 #Tested:867 Testing Accuracy:50.0%\r",
      "Progress:86.7% Speed(reviews/sec):675.0 #Correct:434 #Tested:868 Testing Accuracy:50.0%\r",
      "Progress:86.8% Speed(reviews/sec):675.1 #Correct:435 #Tested:869 Testing Accuracy:50.0%\r",
      "Progress:86.9% Speed(reviews/sec):675.2 #Correct:435 #Tested:870 Testing Accuracy:50.0%\r",
      "Progress:87.0% Speed(reviews/sec):675.3 #Correct:436 #Tested:871 Testing Accuracy:50.0%\r",
      "Progress:87.1% Speed(reviews/sec):675.3 #Correct:436 #Tested:872 Testing Accuracy:50.0%\r",
      "Progress:87.2% Speed(reviews/sec):675.4 #Correct:437 #Tested:873 Testing Accuracy:50.0%\r",
      "Progress:87.3% Speed(reviews/sec):675.6 #Correct:437 #Tested:874 Testing Accuracy:50.0%\r",
      "Progress:87.4% Speed(reviews/sec):675.7 #Correct:438 #Tested:875 Testing Accuracy:50.0%\r",
      "Progress:87.5% Speed(reviews/sec):675.9 #Correct:438 #Tested:876 Testing Accuracy:50.0%\r",
      "Progress:87.6% Speed(reviews/sec):676.0 #Correct:439 #Tested:877 Testing Accuracy:50.0%\r",
      "Progress:87.7% Speed(reviews/sec):675.5 #Correct:439 #Tested:878 Testing Accuracy:50.0%\r",
      "Progress:87.8% Speed(reviews/sec):675.6 #Correct:440 #Tested:879 Testing Accuracy:50.0%\r",
      "Progress:87.9% Speed(reviews/sec):675.7 #Correct:440 #Tested:880 Testing Accuracy:50.0%\r",
      "Progress:88.0% Speed(reviews/sec):675.8 #Correct:441 #Tested:881 Testing Accuracy:50.0%\r",
      "Progress:88.1% Speed(reviews/sec):675.9 #Correct:441 #Tested:882 Testing Accuracy:50.0%\r",
      "Progress:88.2% Speed(reviews/sec):676.1 #Correct:442 #Tested:883 Testing Accuracy:50.0%\r",
      "Progress:88.3% Speed(reviews/sec):676.2 #Correct:442 #Tested:884 Testing Accuracy:50.0%\r",
      "Progress:88.4% Speed(reviews/sec):676.3 #Correct:443 #Tested:885 Testing Accuracy:50.0%\r",
      "Progress:88.5% Speed(reviews/sec):676.4 #Correct:443 #Tested:886 Testing Accuracy:50.0%\r",
      "Progress:88.6% Speed(reviews/sec):676.6 #Correct:444 #Tested:887 Testing Accuracy:50.0%\r",
      "Progress:88.7% Speed(reviews/sec):676.6 #Correct:444 #Tested:888 Testing Accuracy:50.0%\r",
      "Progress:88.8% Speed(reviews/sec):676.8 #Correct:445 #Tested:889 Testing Accuracy:50.0%\r",
      "Progress:88.9% Speed(reviews/sec):676.7 #Correct:445 #Tested:890 Testing Accuracy:50.0%\r",
      "Progress:89.0% Speed(reviews/sec):676.8 #Correct:446 #Tested:891 Testing Accuracy:50.0%\r",
      "Progress:89.1% Speed(reviews/sec):676.8 #Correct:446 #Tested:892 Testing Accuracy:50.0%\r",
      "Progress:89.2% Speed(reviews/sec):676.9 #Correct:447 #Tested:893 Testing Accuracy:50.0%\r",
      "Progress:89.3% Speed(reviews/sec):677.0 #Correct:447 #Tested:894 Testing Accuracy:50.0%\r",
      "Progress:89.4% Speed(reviews/sec):677.2 #Correct:448 #Tested:895 Testing Accuracy:50.0%\r",
      "Progress:89.5% Speed(reviews/sec):677.3 #Correct:448 #Tested:896 Testing Accuracy:50.0%\r",
      "Progress:89.6% Speed(reviews/sec):677.4 #Correct:449 #Tested:897 Testing Accuracy:50.0%\r",
      "Progress:89.7% Speed(reviews/sec):677.5 #Correct:449 #Tested:898 Testing Accuracy:50.0%\r",
      "Progress:89.8% Speed(reviews/sec):677.2 #Correct:450 #Tested:899 Testing Accuracy:50.0%\r",
      "Progress:89.9% Speed(reviews/sec):677.3 #Correct:450 #Tested:900 Testing Accuracy:50.0%\r",
      "Progress:90.0% Speed(reviews/sec):677.3 #Correct:451 #Tested:901 Testing Accuracy:50.0%\r",
      "Progress:90.1% Speed(reviews/sec):677.0 #Correct:451 #Tested:902 Testing Accuracy:50.0%\r",
      "Progress:90.2% Speed(reviews/sec):677.1 #Correct:452 #Tested:903 Testing Accuracy:50.0%\r",
      "Progress:90.3% Speed(reviews/sec):677.2 #Correct:452 #Tested:904 Testing Accuracy:50.0%\r",
      "Progress:90.4% Speed(reviews/sec):677.3 #Correct:453 #Tested:905 Testing Accuracy:50.0%\r",
      "Progress:90.5% Speed(reviews/sec):677.3 #Correct:453 #Tested:906 Testing Accuracy:50.0%\r",
      "Progress:90.6% Speed(reviews/sec):677.5 #Correct:454 #Tested:907 Testing Accuracy:50.0%\r",
      "Progress:90.7% Speed(reviews/sec):677.5 #Correct:454 #Tested:908 Testing Accuracy:50.0%\r",
      "Progress:90.8% Speed(reviews/sec):677.6 #Correct:455 #Tested:909 Testing Accuracy:50.0%\r",
      "Progress:90.9% Speed(reviews/sec):677.7 #Correct:455 #Tested:910 Testing Accuracy:50.0%\r",
      "Progress:91.0% Speed(reviews/sec):677.8 #Correct:456 #Tested:911 Testing Accuracy:50.0%\r",
      "Progress:91.1% Speed(reviews/sec):677.9 #Correct:456 #Tested:912 Testing Accuracy:50.0%\r",
      "Progress:91.2% Speed(reviews/sec):677.9 #Correct:457 #Tested:913 Testing Accuracy:50.0%\r",
      "Progress:91.3% Speed(reviews/sec):677.8 #Correct:457 #Tested:914 Testing Accuracy:50.0%\r",
      "Progress:91.4% Speed(reviews/sec):677.9 #Correct:458 #Tested:915 Testing Accuracy:50.0%\r",
      "Progress:91.5% Speed(reviews/sec):678.0 #Correct:458 #Tested:916 Testing Accuracy:50.0%\r",
      "Progress:91.6% Speed(reviews/sec):677.9 #Correct:459 #Tested:917 Testing Accuracy:50.0%\r",
      "Progress:91.7% Speed(reviews/sec):677.9 #Correct:459 #Tested:918 Testing Accuracy:50.0%\r",
      "Progress:91.8% Speed(reviews/sec):678.0 #Correct:460 #Tested:919 Testing Accuracy:50.0%\r",
      "Progress:91.9% Speed(reviews/sec):678.0 #Correct:460 #Tested:920 Testing Accuracy:50.0%\r",
      "Progress:92.0% Speed(reviews/sec):678.0 #Correct:461 #Tested:921 Testing Accuracy:50.0%\r",
      "Progress:92.1% Speed(reviews/sec):678.1 #Correct:461 #Tested:922 Testing Accuracy:50.0%\r",
      "Progress:92.2% Speed(reviews/sec):678.2 #Correct:462 #Tested:923 Testing Accuracy:50.0%\r",
      "Progress:92.3% Speed(reviews/sec):678.3 #Correct:462 #Tested:924 Testing Accuracy:50.0%\r",
      "Progress:92.4% Speed(reviews/sec):678.4 #Correct:463 #Tested:925 Testing Accuracy:50.0%\r",
      "Progress:92.5% Speed(reviews/sec):678.6 #Correct:463 #Tested:926 Testing Accuracy:50.0%\r",
      "Progress:92.6% Speed(reviews/sec):678.6 #Correct:464 #Tested:927 Testing Accuracy:50.0%\r",
      "Progress:92.7% Speed(reviews/sec):678.7 #Correct:464 #Tested:928 Testing Accuracy:50.0%\r",
      "Progress:92.8% Speed(reviews/sec):678.9 #Correct:465 #Tested:929 Testing Accuracy:50.0%\r",
      "Progress:92.9% Speed(reviews/sec):678.9 #Correct:465 #Tested:930 Testing Accuracy:50.0%\r",
      "Progress:93.0% Speed(reviews/sec):679.0 #Correct:466 #Tested:931 Testing Accuracy:50.0%\r",
      "Progress:93.1% Speed(reviews/sec):678.8 #Correct:466 #Tested:932 Testing Accuracy:50.0%\r",
      "Progress:93.2% Speed(reviews/sec):679.0 #Correct:467 #Tested:933 Testing Accuracy:50.0%\r",
      "Progress:93.3% Speed(reviews/sec):678.8 #Correct:467 #Tested:934 Testing Accuracy:50.0%\r",
      "Progress:93.4% Speed(reviews/sec):678.9 #Correct:468 #Tested:935 Testing Accuracy:50.0%\r",
      "Progress:93.5% Speed(reviews/sec):679.0 #Correct:468 #Tested:936 Testing Accuracy:50.0%\r",
      "Progress:93.6% Speed(reviews/sec):679.2 #Correct:469 #Tested:937 Testing Accuracy:50.0%\r",
      "Progress:93.7% Speed(reviews/sec):679.3 #Correct:469 #Tested:938 Testing Accuracy:50.0%\r",
      "Progress:93.8% Speed(reviews/sec):679.4 #Correct:470 #Tested:939 Testing Accuracy:50.0%\r",
      "Progress:93.9% Speed(reviews/sec):679.5 #Correct:470 #Tested:940 Testing Accuracy:50.0%\r",
      "Progress:94.0% Speed(reviews/sec):679.7 #Correct:471 #Tested:941 Testing Accuracy:50.0%\r",
      "Progress:94.1% Speed(reviews/sec):679.7 #Correct:471 #Tested:942 Testing Accuracy:50.0%\r",
      "Progress:94.2% Speed(reviews/sec):679.8 #Correct:472 #Tested:943 Testing Accuracy:50.0%\r",
      "Progress:94.3% Speed(reviews/sec):679.9 #Correct:472 #Tested:944 Testing Accuracy:50.0%\r",
      "Progress:94.4% Speed(reviews/sec):679.7 #Correct:473 #Tested:945 Testing Accuracy:50.0%\r",
      "Progress:94.5% Speed(reviews/sec):679.7 #Correct:473 #Tested:946 Testing Accuracy:50.0%\r",
      "Progress:94.6% Speed(reviews/sec):679.7 #Correct:474 #Tested:947 Testing Accuracy:50.0%\r",
      "Progress:94.7% Speed(reviews/sec):679.8 #Correct:474 #Tested:948 Testing Accuracy:50.0%\r",
      "Progress:94.8% Speed(reviews/sec):679.9 #Correct:475 #Tested:949 Testing Accuracy:50.0%\r",
      "Progress:94.9% Speed(reviews/sec):680.1 #Correct:475 #Tested:950 Testing Accuracy:50.0%\r",
      "Progress:95.0% Speed(reviews/sec):680.2 #Correct:476 #Tested:951 Testing Accuracy:50.0%\r",
      "Progress:95.1% Speed(reviews/sec):680.0 #Correct:476 #Tested:952 Testing Accuracy:50.0%\r",
      "Progress:95.2% Speed(reviews/sec):680.0 #Correct:477 #Tested:953 Testing Accuracy:50.0%\r",
      "Progress:95.3% Speed(reviews/sec):680.1 #Correct:477 #Tested:954 Testing Accuracy:50.0%\r",
      "Progress:95.4% Speed(reviews/sec):679.8 #Correct:478 #Tested:955 Testing Accuracy:50.0%\r",
      "Progress:95.5% Speed(reviews/sec):680.0 #Correct:478 #Tested:956 Testing Accuracy:50.0%\r",
      "Progress:95.6% Speed(reviews/sec):680.1 #Correct:479 #Tested:957 Testing Accuracy:50.0%\r",
      "Progress:95.7% Speed(reviews/sec):680.3 #Correct:479 #Tested:958 Testing Accuracy:50.0%\r",
      "Progress:95.8% Speed(reviews/sec):680.4 #Correct:480 #Tested:959 Testing Accuracy:50.0%\r",
      "Progress:95.9% Speed(reviews/sec):680.5 #Correct:480 #Tested:960 Testing Accuracy:50.0%\r",
      "Progress:96.0% Speed(reviews/sec):680.4 #Correct:481 #Tested:961 Testing Accuracy:50.0%\r",
      "Progress:96.1% Speed(reviews/sec):680.6 #Correct:481 #Tested:962 Testing Accuracy:50.0%\r",
      "Progress:96.2% Speed(reviews/sec):680.7 #Correct:482 #Tested:963 Testing Accuracy:50.0%\r",
      "Progress:96.3% Speed(reviews/sec):680.8 #Correct:482 #Tested:964 Testing Accuracy:50.0%\r",
      "Progress:96.4% Speed(reviews/sec):681.0 #Correct:483 #Tested:965 Testing Accuracy:50.0%\r",
      "Progress:96.5% Speed(reviews/sec):681.1 #Correct:483 #Tested:966 Testing Accuracy:50.0%\r",
      "Progress:96.6% Speed(reviews/sec):681.2 #Correct:484 #Tested:967 Testing Accuracy:50.0%\r",
      "Progress:96.7% Speed(reviews/sec):681.3 #Correct:484 #Tested:968 Testing Accuracy:50.0%\r",
      "Progress:96.8% Speed(reviews/sec):681.4 #Correct:485 #Tested:969 Testing Accuracy:50.0%"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\r",
      "Progress:96.9% Speed(reviews/sec):681.5 #Correct:485 #Tested:970 Testing Accuracy:50.0%\r",
      "Progress:97.0% Speed(reviews/sec):681.0 #Correct:486 #Tested:971 Testing Accuracy:50.0%\r",
      "Progress:97.1% Speed(reviews/sec):680.9 #Correct:486 #Tested:972 Testing Accuracy:50.0%\r",
      "Progress:97.2% Speed(reviews/sec):680.9 #Correct:487 #Tested:973 Testing Accuracy:50.0%\r",
      "Progress:97.3% Speed(reviews/sec):680.7 #Correct:487 #Tested:974 Testing Accuracy:50.0%\r",
      "Progress:97.4% Speed(reviews/sec):680.5 #Correct:488 #Tested:975 Testing Accuracy:50.0%\r",
      "Progress:97.5% Speed(reviews/sec):680.5 #Correct:488 #Tested:976 Testing Accuracy:50.0%\r",
      "Progress:97.6% Speed(reviews/sec):680.5 #Correct:489 #Tested:977 Testing Accuracy:50.0%\r",
      "Progress:97.7% Speed(reviews/sec):680.1 #Correct:489 #Tested:978 Testing Accuracy:50.0%\r",
      "Progress:97.8% Speed(reviews/sec):680.2 #Correct:490 #Tested:979 Testing Accuracy:50.0%\r",
      "Progress:97.9% Speed(reviews/sec):680.1 #Correct:490 #Tested:980 Testing Accuracy:50.0%\r",
      "Progress:98.0% Speed(reviews/sec):680.3 #Correct:491 #Tested:981 Testing Accuracy:50.0%\r",
      "Progress:98.1% Speed(reviews/sec):679.8 #Correct:491 #Tested:982 Testing Accuracy:50.0%\r",
      "Progress:98.2% Speed(reviews/sec):679.7 #Correct:492 #Tested:983 Testing Accuracy:50.0%\r",
      "Progress:98.3% Speed(reviews/sec):679.0 #Correct:492 #Tested:984 Testing Accuracy:50.0%\r",
      "Progress:98.4% Speed(reviews/sec):679.1 #Correct:493 #Tested:985 Testing Accuracy:50.0%\r",
      "Progress:98.5% Speed(reviews/sec):679.2 #Correct:493 #Tested:986 Testing Accuracy:50.0%\r",
      "Progress:98.6% Speed(reviews/sec):679.1 #Correct:494 #Tested:987 Testing Accuracy:50.0%\r",
      "Progress:98.7% Speed(reviews/sec):678.9 #Correct:494 #Tested:988 Testing Accuracy:50.0%\r",
      "Progress:98.8% Speed(reviews/sec):678.8 #Correct:495 #Tested:989 Testing Accuracy:50.0%\r",
      "Progress:98.9% Speed(reviews/sec):678.2 #Correct:495 #Tested:990 Testing Accuracy:50.0%\r",
      "Progress:99.0% Speed(reviews/sec):678.2 #Correct:496 #Tested:991 Testing Accuracy:50.0%\r",
      "Progress:99.1% Speed(reviews/sec):678.3 #Correct:496 #Tested:992 Testing Accuracy:50.0%\r",
      "Progress:99.2% Speed(reviews/sec):678.4 #Correct:497 #Tested:993 Testing Accuracy:50.0%\r",
      "Progress:99.3% Speed(reviews/sec):678.5 #Correct:497 #Tested:994 Testing Accuracy:50.0%\r",
      "Progress:99.4% Speed(reviews/sec):678.4 #Correct:498 #Tested:995 Testing Accuracy:50.0%\r",
      "Progress:99.5% Speed(reviews/sec):678.4 #Correct:498 #Tested:996 Testing Accuracy:50.0%\r",
      "Progress:99.6% Speed(reviews/sec):678.4 #Correct:499 #Tested:997 Testing Accuracy:50.0%\r",
      "Progress:99.7% Speed(reviews/sec):678.1 #Correct:499 #Tested:998 Testing Accuracy:50.0%\r",
      "Progress:99.8% Speed(reviews/sec):678.1 #Correct:500 #Tested:999 Testing Accuracy:50.0%\r",
      "Progress:99.9% Speed(reviews/sec):678.2 #Correct:500 #Tested:1000 Testing Accuracy:50.0%"
     ]
    }
   ],
   "source": [
    "mlp.test(reviews[-1000:],labels[-1000:])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "运行下方代码框，开始训练网络。训练时，代码会不断输出模型的准确性，以便你观察模型的训练效果。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Progress:0.0% Speed(reviews/sec):0.0 #Correct:1 #Trained:1 Training Accuracy:100.%\n",
      "layer_0 shape = (1, 72810)\n",
      "layer_1 shape = (1, 10)\n",
      "layer_2 shape = (1, 1)\n",
      "layer_2_delta shape = (1, 1)\n",
      "layer_1_delta shape = (1, 10)                    \n",
      "weights_1_2 shape = (10, 1)\n",
      " weights_0_1 shape = (72810, 10)\n",
      "Progress:10.4% Speed(reviews/sec):185.3 #Correct:1251 #Trained:2501 Training Accuracy:50.0%\n",
      "Progress:20.8% Speed(reviews/sec):184.6 #Correct:2501 #Trained:5001 Training Accuracy:50.0%\n",
      "Progress:31.2% Speed(reviews/sec):184.6 #Correct:3751 #Trained:7501 Training Accuracy:50.0%\n",
      "Progress:41.6% Speed(reviews/sec):184.5 #Correct:5001 #Trained:10001 Training Accuracy:50.0%\n",
      "Progress:52.0% Speed(reviews/sec):184.5 #Correct:6251 #Trained:12501 Training Accuracy:50.0%\n",
      "Progress:62.5% Speed(reviews/sec):184.5 #Correct:7501 #Trained:15001 Training Accuracy:50.0%\n",
      "Progress:72.9% Speed(reviews/sec):182.0 #Correct:8751 #Trained:17501 Training Accuracy:50.0%\n",
      "Progress:83.3% Speed(reviews/sec):181.2 #Correct:10001 #Trained:20001 Training Accuracy:50.0%\n",
      "Progress:93.7% Speed(reviews/sec):179.7 #Correct:11251 #Trained:22501 Training Accuracy:50.0%\n",
      "Progress:99.9% Speed(reviews/sec):179.4 #Correct:12000 #Trained:24000 Training Accuracy:50.0%"
     ]
    }
   ],
   "source": [
    "mlp.train(reviews[:-1000],labels[:-1000])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "很可能训练效果并不好，部分原因可能在于学习率太高了。运行下方代码框，将学习率调小为 `0.01`，重建并重新训练网络。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Progress:0.0% Speed(reviews/sec):0.0 #Correct:1 #Trained:1 Training Accuracy:100.%\n",
      "layer_0 shape = (1, 72810)\n",
      "layer_1 shape = (1, 10)\n",
      "layer_2 shape = (1, 1)\n",
      "layer_2_delta shape = (1, 1)\n",
      "layer_1_delta shape = (1, 10)                    \n",
      "weights_1_2 shape = (10, 1)\n",
      " weights_0_1 shape = (72810, 10)\n",
      "Progress:10.4% Speed(reviews/sec):182.8 #Correct:1248 #Trained:2501 Training Accuracy:49.9%\n",
      "Progress:20.8% Speed(reviews/sec):180.7 #Correct:2498 #Trained:5001 Training Accuracy:49.9%\n",
      "Progress:31.2% Speed(reviews/sec):181.7 #Correct:3748 #Trained:7501 Training Accuracy:49.9%\n",
      "Progress:41.6% Speed(reviews/sec):182.7 #Correct:4998 #Trained:10001 Training Accuracy:49.9%\n",
      "Progress:52.0% Speed(reviews/sec):183.2 #Correct:6248 #Trained:12501 Training Accuracy:49.9%\n",
      "Progress:62.5% Speed(reviews/sec):183.6 #Correct:7487 #Trained:15001 Training Accuracy:49.9%\n",
      "Progress:72.9% Speed(reviews/sec):183.8 #Correct:8737 #Trained:17501 Training Accuracy:49.9%\n",
      "Progress:83.3% Speed(reviews/sec):183.9 #Correct:9987 #Trained:20001 Training Accuracy:49.9%\n",
      "Progress:93.7% Speed(reviews/sec):184.1 #Correct:11237 #Trained:22501 Training Accuracy:49.9%\n",
      "Progress:99.9% Speed(reviews/sec):184.1 #Correct:11986 #Trained:24000 Training Accuracy:49.9%"
     ]
    }
   ],
   "source": [
    "mlp = SentimentNetwork(reviews[:-1000],labels[:-1000], learning_rate=0.01)\n",
    "mlp.train(reviews[:-1000],labels[:-1000])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "可能效果还是不好。运行下方代码框，再次将学习率调小为 `0.001`，再次重建并重新训练网络。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Progress:0.0% Speed(reviews/sec):0.0 #Correct:1 #Trained:1 Training Accuracy:100.%\n",
      "layer_0 shape = (1, 72810)\n",
      "layer_1 shape = (1, 10)\n",
      "layer_2 shape = (1, 1)\n",
      "layer_2_delta shape = (1, 1)\n",
      "layer_1_delta shape = (1, 10)                    \n",
      "weights_1_2 shape = (10, 1)\n",
      " weights_0_1 shape = (72810, 10)\n",
      "Progress:10.4% Speed(reviews/sec):186.6 #Correct:1270 #Trained:2501 Training Accuracy:50.7%\n",
      "Progress:20.8% Speed(reviews/sec):185.7 #Correct:2609 #Trained:5001 Training Accuracy:52.1%\n",
      "Progress:31.2% Speed(reviews/sec):186.1 #Correct:4032 #Trained:7501 Training Accuracy:53.7%\n",
      "Progress:41.6% Speed(reviews/sec):186.2 #Correct:5530 #Trained:10001 Training Accuracy:55.2%\n",
      "Progress:52.0% Speed(reviews/sec):186.3 #Correct:7119 #Trained:12501 Training Accuracy:56.9%\n",
      "Progress:62.5% Speed(reviews/sec):186.3 #Correct:8687 #Trained:15001 Training Accuracy:57.9%\n",
      "Progress:72.9% Speed(reviews/sec):186.4 #Correct:10311 #Trained:17501 Training Accuracy:58.9%\n",
      "Progress:83.3% Speed(reviews/sec):186.3 #Correct:11909 #Trained:20001 Training Accuracy:59.5%\n",
      "Progress:93.7% Speed(reviews/sec):186.3 #Correct:13569 #Trained:22501 Training Accuracy:60.3%\n",
      "Progress:99.9% Speed(reviews/sec):186.3 #Correct:14627 #Trained:24000 Training Accuracy:60.9%"
     ]
    }
   ],
   "source": [
    "mlp = SentimentNetwork(reviews[:-1000],labels[:-1000], learning_rate=0.001)\n",
    "mlp.train(reviews[:-1000],labels[:-1000])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "当学习率为 `0.001` 时，网络应该终于开始随着训练进行改进了，虽然效果还是不尽人意，但起码说明了这个解决方案是可行的。我们会在下一课进一步改进这个网络。\n",
    "\n",
    "# 项目 3 到此结束。\n",
    "\n",
    "# 理解神经噪声<a id='lesson_4'></a>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from IPython.display import Image\n",
    "Image(filename='sentiment_network.png')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [],
   "source": [
    "def update_input_layer(review):\n",
    "    \n",
    "    global layer_0\n",
    "    \n",
    "    # clear out previous state, reset the layer to be all 0s\n",
    "    layer_0 *= 0\n",
    "    for word in review.split(\" \"):\n",
    "        layer_0[0][word2index[word]] += 1\n",
    "\n",
    "update_input_layer(reviews[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[18.,  0.,  0., ...,  0.,  0.,  0.]])"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "layer_0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [],
   "source": [
    "review_counter = Counter()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [],
   "source": [
    "for word in reviews[0].split(\" \"):\n",
    "    review_counter[word] += 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[('.', 27),\n",
       " ('', 18),\n",
       " ('the', 9),\n",
       " ('to', 6),\n",
       " ('high', 5),\n",
       " ('i', 5),\n",
       " ('bromwell', 4),\n",
       " ('is', 4),\n",
       " ('a', 4),\n",
       " ('teachers', 4),\n",
       " ('that', 4),\n",
       " ('of', 4),\n",
       " ('it', 2),\n",
       " ('at', 2),\n",
       " ('as', 2),\n",
       " ('school', 2),\n",
       " ('my', 2),\n",
       " ('in', 2),\n",
       " ('me', 2),\n",
       " ('students', 2),\n",
       " ('their', 2),\n",
       " ('student', 2),\n",
       " ('cartoon', 1),\n",
       " ('comedy', 1),\n",
       " ('ran', 1),\n",
       " ('same', 1),\n",
       " ('time', 1),\n",
       " ('some', 1),\n",
       " ('other', 1),\n",
       " ('programs', 1)]"
      ]
     },
     "execution_count": 88,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "review_counter.most_common(30)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 项目 4：减小输入数据的噪声<a id='project_4'></a>\n",
    "\n",
    "**待办：** 尝试为输入数据减小噪声，具体操作内容如下：\n",
    "* 将你之前创建的类 `SentimentNetwork` 复制到下方的代码框中。\n",
    "* 修改 `update_input_layer`，使其不再统计各单词出现的次数，而只记录单词是否出现过。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [],
   "source": [
    "import time\n",
    "import sys\n",
    "import numpy as np\n",
    "\n",
    "# Encapsulate our neural network in a class\n",
    "class SentimentNetwork:\n",
    "    def __init__(self, reviews,labels,hidden_nodes = 10, learning_rate = 0.1):\n",
    "        \"\"\"Create a SentimenNetwork with the given settings\n",
    "        Args:\n",
    "            reviews(list) - List of reviews used for training\n",
    "            labels(list) - List of POSITIVE/NEGATIVE labels associated with the given reviews\n",
    "            hidden_nodes(int) - Number of nodes to create in the hidden layer\n",
    "            learning_rate(float) - Learning rate to use while training\n",
    "        \n",
    "        \"\"\"\n",
    "        # Assign a seed to our random number generator to ensure we get\n",
    "        # reproducable results during development \n",
    "        np.random.seed(1)\n",
    "\n",
    "        # process the reviews and their associated labels so that everything\n",
    "        # is ready for training\n",
    "        self.pre_process_data(reviews, labels)\n",
    "        \n",
    "        # Build the network to have the number of hidden nodes and the learning rate that\n",
    "        # were passed into this initializer. Make the same number of input nodes as\n",
    "        # there are vocabulary words and create a single output node.\n",
    "        self.init_network(len(self.review_vocab),hidden_nodes, 1, learning_rate)\n",
    "\n",
    "    def pre_process_data(self, reviews, labels):\n",
    "        \n",
    "        # populate review_vocab with all of the words in the given reviews\n",
    "        review_vocab = set()\n",
    "        for review in reviews:\n",
    "            for word in review.split(\" \"):\n",
    "                review_vocab.add(word)\n",
    "\n",
    "        # Convert the vocabulary set to a list so we can access words via indices\n",
    "        self.review_vocab = list(review_vocab)\n",
    "        \n",
    "        # populate label_vocab with all of the words in the given labels.\n",
    "        label_vocab = set()\n",
    "        for label in labels:\n",
    "            label_vocab.add(label)\n",
    "        \n",
    "        # Convert the label vocabulary set to a list so we can access labels via indices\n",
    "        self.label_vocab = list(label_vocab)\n",
    "        \n",
    "        # Store the sizes of the review and label vocabularies.\n",
    "        self.review_vocab_size = len(self.review_vocab)\n",
    "        self.label_vocab_size = len(self.label_vocab)\n",
    "        \n",
    "        # Create a dictionary of words in the vocabulary mapped to index positions\n",
    "        self.word2index = {}\n",
    "        for i, word in enumerate(self.review_vocab):\n",
    "            self.word2index[word] = i\n",
    "        \n",
    "        # Create a dictionary of labels mapped to index positions\n",
    "        self.label2index = {}\n",
    "        for i, label in enumerate(self.label_vocab):\n",
    "            self.label2index[label] = i\n",
    "        \n",
    "    def init_network(self, input_nodes, hidden_nodes, output_nodes, learning_rate):\n",
    "        # Set number of nodes in input, hidden and output layers.\n",
    "        self.input_nodes = input_nodes\n",
    "        self.hidden_nodes = hidden_nodes\n",
    "        self.output_nodes = output_nodes\n",
    "\n",
    "        # Store the learning rate\n",
    "        self.learning_rate = learning_rate\n",
    "\n",
    "        # Initialize weights\n",
    "\n",
    "        # These are the weights between the input layer and the hidden layer.\n",
    "        self.weights_0_1 = np.zeros((self.input_nodes,self.hidden_nodes))\n",
    "    \n",
    "        # These are the weights between the hidden layer and the output layer.\n",
    "        self.weights_1_2 = np.random.normal(0.0, self.output_nodes**-0.5, \n",
    "                                                (self.hidden_nodes, self.output_nodes))\n",
    "        \n",
    "        # The input layer, a two-dimensional matrix with shape 1 x input_nodes\n",
    "        self.layer_0 = np.zeros((1,input_nodes))\n",
    "    \n",
    "        \n",
    "    def update_input_layer(self,review):\n",
    "\n",
    "        # clear out previous state, reset the layer to be all 0s\n",
    "        self.layer_0 *= 0\n",
    "        \n",
    "        for word in review.split(\" \"):\n",
    "            # NOTE: This if-check was not in the version of this method created in Project 2,\n",
    "            #       and it appears in Andrew's Project 3 solution without explanation. \n",
    "            #       It simply ensures the word is actually a key in word2index before\n",
    "            #       accessing it, which is important because accessing an invalid key\n",
    "            #       with raise an exception in Python. This allows us to ignore unknown\n",
    "            #       words encountered in new reviews.\n",
    "            if(word in self.word2index.keys()):\n",
    "                ## New for Project 4: changed to set to 1 instead of add 1\n",
    "                self.layer_0[0][self.word2index[word]] = 1\n",
    "                \n",
    "    def get_target_for_label(self,label):\n",
    "        if(label == 'POSITIVE'):\n",
    "            return 1\n",
    "        else:\n",
    "            return 0\n",
    "        \n",
    "    def sigmoid(self,x):\n",
    "        return 1 / (1 + np.exp(-x))\n",
    "    \n",
    "    def sigmoid_output_2_derivative(self,output):\n",
    "        return output * (1 - output)\n",
    "    \n",
    "    def train(self, training_reviews, training_labels):\n",
    "        \n",
    "        # make sure out we have a matching number of reviews and labels\n",
    "        assert(len(training_reviews) == len(training_labels))\n",
    "        \n",
    "        # Keep track of correct predictions to display accuracy during training \n",
    "        correct_so_far = 0\n",
    "\n",
    "        # Remember when we started for printing time statistics\n",
    "        start = time.time()\n",
    "        \n",
    "        # loop through all the given reviews and run a forward and backward pass,\n",
    "        # updating weights for every item\n",
    "        for i in range(len(training_reviews)):\n",
    "            \n",
    "            # Get the next review and its correct label\n",
    "            review = training_reviews[i]\n",
    "            label = training_labels[i]\n",
    "            \n",
    "            #### Implement the forward pass here ####\n",
    "            ### Forward pass ###\n",
    "\n",
    "            # Input Layer\n",
    "            self.update_input_layer(review)\n",
    "\n",
    "            # Hidden layer\n",
    "            layer_1 = self.layer_0.dot(self.weights_0_1)\n",
    "\n",
    "            # Output layer\n",
    "            layer_2 = self.sigmoid(layer_1.dot(self.weights_1_2))\n",
    "            \n",
    "            #### Implement the backward pass here ####\n",
    "            ### Backward pass ###\n",
    "\n",
    "            # Output error\n",
    "            layer_2_error = layer_2 - self.get_target_for_label(label) # Output layer error is the difference between desired target and actual output.\n",
    "            layer_2_delta = layer_2_error * self.sigmoid_output_2_derivative(layer_2)\n",
    "\n",
    "            # Backpropagated error\n",
    "            layer_1_error = layer_2_delta.dot(self.weights_1_2.T) # errors propagated to the hidden layer\n",
    "            layer_1_delta = layer_1_error # hidden layer gradients - no nonlinearity so it's the same as the error\n",
    "\n",
    "            # Update the weights\n",
    "            self.weights_1_2 -= layer_1.T.dot(layer_2_delta) * self.learning_rate # update hidden-to-output weights with gradient descent step\n",
    "            self.weights_0_1 -= self.layer_0.T.dot(layer_1_delta) * self.learning_rate # update input-to-hidden weights with gradient descent step\n",
    "\n",
    "            # Keep track of correct predictions.\n",
    "            if(layer_2 >= 0.5 and label == 'POSITIVE'):\n",
    "                correct_so_far += 1\n",
    "            elif(layer_2 < 0.5 and label == 'NEGATIVE'):\n",
    "                correct_so_far += 1\n",
    "            \n",
    "            # For debug purposes, print out our prediction accuracy and speed \n",
    "            # throughout the training process. \n",
    "            elapsed_time = float(time.time() - start)\n",
    "            reviews_per_second = i / elapsed_time if elapsed_time > 0 else 0\n",
    "            \n",
    "            sys.stdout.write(\"\\rProgress:\" + str(100 * i/float(len(training_reviews)))[:4] \\\n",
    "                             + \"% Speed(reviews/sec):\" + str(reviews_per_second)[0:5] \\\n",
    "                             + \" #Correct:\" + str(correct_so_far) + \" #Trained:\" + str(i+1) \\\n",
    "                             + \" Training Accuracy:\" + str(correct_so_far * 100 / float(i+1))[:4] + \"%\")\n",
    "            if(i % 2500 == 0):\n",
    "                print(\"\")\n",
    "    \n",
    "    def test(self, testing_reviews, testing_labels):\n",
    "        \"\"\"\n",
    "        Attempts to predict the labels for the given testing_reviews,\n",
    "        and uses the test_labels to calculate the accuracy of those predictions.\n",
    "        \"\"\"\n",
    "        \n",
    "        # keep track of how many correct predictions we make\n",
    "        correct = 0\n",
    "\n",
    "        # we'll time how many predictions per second we make\n",
    "        start = time.time()\n",
    "\n",
    "        # Loop through each of the given reviews and call run to predict\n",
    "        # its label. \n",
    "        for i in range(len(testing_reviews)):\n",
    "            pred = self.run(testing_reviews[i])\n",
    "            if(pred == testing_labels[i]):\n",
    "                correct += 1\n",
    "            \n",
    "            # For debug purposes, print out our prediction accuracy and speed \n",
    "            # throughout the prediction process. \n",
    "\n",
    "            elapsed_time = float(time.time() - start)\n",
    "            reviews_per_second = i / elapsed_time if elapsed_time > 0 else 0\n",
    "            \n",
    "            sys.stdout.write(\"\\rProgress:\" + str(100 * i/float(len(testing_reviews)))[:4] \\\n",
    "                             + \"% Speed(reviews/sec):\" + str(reviews_per_second)[0:5] \\\n",
    "                             + \" #Correct:\" + str(correct) + \" #Tested:\" + str(i+1) \\\n",
    "                             + \" Testing Accuracy:\" + str(correct * 100 / float(i+1))[:4] + \"%\")\n",
    "    \n",
    "    def run(self, review):\n",
    "        \"\"\"\n",
    "        Returns a POSITIVE or NEGATIVE prediction for the given review.\n",
    "        \"\"\"\n",
    "        # Run a forward pass through the network, like in the \"train\" function.\n",
    "        \n",
    "        # Input Layer\n",
    "        self.update_input_layer(review.lower())\n",
    "\n",
    "        # Hidden layer\n",
    "        layer_1 = self.layer_0.dot(self.weights_0_1)\n",
    "\n",
    "        # Output layer\n",
    "        layer_2 = self.sigmoid(layer_1.dot(self.weights_1_2))\n",
    "        \n",
    "        # Return POSITIVE for values above greater-than-or-equal-to 0.5 in the output layer;\n",
    "        # return NEGATIVE for other values\n",
    "        if(layer_2[0] >= 0.5):\n",
    "            return \"POSITIVE\"\n",
    "        else:\n",
    "            return \"NEGATIVE\"\n",
    "        "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "运行下方代码框，重建网络并对其进行训练。注意，此处重新采纳了较高的学习率，即 `0.1` 。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Progress:0.0% Speed(reviews/sec):0.0 #Correct:1 #Trained:1 Training Accuracy:100.%\n",
      "Progress:10.4% Speed(reviews/sec):190.4 #Correct:1819 #Trained:2501 Training Accuracy:72.7%\n",
      "Progress:20.8% Speed(reviews/sec):190.7 #Correct:3815 #Trained:5001 Training Accuracy:76.2%\n",
      "Progress:31.2% Speed(reviews/sec):191.0 #Correct:5907 #Trained:7501 Training Accuracy:78.7%\n",
      "Progress:41.6% Speed(reviews/sec):191.2 #Correct:8058 #Trained:10001 Training Accuracy:80.5%\n",
      "Progress:52.0% Speed(reviews/sec):190.9 #Correct:10175 #Trained:12501 Training Accuracy:81.3%\n",
      "Progress:62.5% Speed(reviews/sec):191.1 #Correct:12296 #Trained:15001 Training Accuracy:81.9%\n",
      "Progress:72.9% Speed(reviews/sec):191.1 #Correct:14406 #Trained:17501 Training Accuracy:82.3%\n",
      "Progress:83.3% Speed(reviews/sec):190.9 #Correct:16572 #Trained:20001 Training Accuracy:82.8%\n",
      "Progress:93.7% Speed(reviews/sec):189.7 #Correct:18754 #Trained:22501 Training Accuracy:83.3%\n",
      "Progress:99.9% Speed(reviews/sec):187.7 #Correct:20082 #Trained:24000 Training Accuracy:83.6%"
     ]
    }
   ],
   "source": [
    "mlp = SentimentNetwork(reviews[:-1000],labels[:-1000], learning_rate=0.1)\n",
    "mlp.train(reviews[:-1000],labels[:-1000])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "这次的训练效果应该比之前的好多了，虽然还不够完美，但进步很大。运行下方代码框，用那一千条预测对模型进行测试。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\r",
      "Progress:0.0% Speed(reviews/sec):0.0 #Correct:1 #Tested:1 Testing Accuracy:100.%\r",
      "Progress:0.1% Speed(reviews/sec):226.7 #Correct:1 #Tested:2 Testing Accuracy:50.0%\r",
      "Progress:0.2% Speed(reviews/sec):342.5 #Correct:2 #Tested:3 Testing Accuracy:66.6%\r",
      "Progress:0.3% Speed(reviews/sec):408.6 #Correct:3 #Tested:4 Testing Accuracy:75.0%\r",
      "Progress:0.4% Speed(reviews/sec):453.6 #Correct:4 #Tested:5 Testing Accuracy:80.0%\r",
      "Progress:0.5% Speed(reviews/sec):494.2 #Correct:5 #Tested:6 Testing Accuracy:83.3%\r",
      "Progress:0.6% Speed(reviews/sec):505.6 #Correct:6 #Tested:7 Testing Accuracy:85.7%\r",
      "Progress:0.7% Speed(reviews/sec):523.4 #Correct:7 #Tested:8 Testing Accuracy:87.5%\r",
      "Progress:0.8% Speed(reviews/sec):542.9 #Correct:8 #Tested:9 Testing Accuracy:88.8%\r",
      "Progress:0.9% Speed(reviews/sec):520.0 #Correct:9 #Tested:10 Testing Accuracy:90.0%\r",
      "Progress:1.0% Speed(reviews/sec):528.3 #Correct:10 #Tested:11 Testing Accuracy:90.9%\r",
      "Progress:1.1% Speed(reviews/sec):545.9 #Correct:11 #Tested:12 Testing Accuracy:91.6%\r",
      "Progress:1.2% Speed(reviews/sec):529.9 #Correct:12 #Tested:13 Testing Accuracy:92.3%\r",
      "Progress:1.3% Speed(reviews/sec):538.0 #Correct:13 #Tested:14 Testing Accuracy:92.8%\r",
      "Progress:1.4% Speed(reviews/sec):545.8 #Correct:14 #Tested:15 Testing Accuracy:93.3%\r",
      "Progress:1.5% Speed(reviews/sec):554.2 #Correct:15 #Tested:16 Testing Accuracy:93.7%\r",
      "Progress:1.6% Speed(reviews/sec):540.6 #Correct:16 #Tested:17 Testing Accuracy:94.1%\r",
      "Progress:1.7% Speed(reviews/sec):542.6 #Correct:17 #Tested:18 Testing Accuracy:94.4%\r",
      "Progress:1.8% Speed(reviews/sec):550.4 #Correct:18 #Tested:19 Testing Accuracy:94.7%\r",
      "Progress:1.9% Speed(reviews/sec):556.8 #Correct:18 #Tested:20 Testing Accuracy:90.0%\r",
      "Progress:2.0% Speed(reviews/sec):556.8 #Correct:19 #Tested:21 Testing Accuracy:90.4%\r",
      "Progress:2.1% Speed(reviews/sec):565.1 #Correct:20 #Tested:22 Testing Accuracy:90.9%\r",
      "Progress:2.2% Speed(reviews/sec):572.0 #Correct:21 #Tested:23 Testing Accuracy:91.3%\r",
      "Progress:2.3% Speed(reviews/sec):577.4 #Correct:22 #Tested:24 Testing Accuracy:91.6%\r",
      "Progress:2.4% Speed(reviews/sec):570.7 #Correct:23 #Tested:25 Testing Accuracy:92.0%\r",
      "Progress:2.5% Speed(reviews/sec):576.7 #Correct:24 #Tested:26 Testing Accuracy:92.3%\r",
      "Progress:2.6% Speed(reviews/sec):582.5 #Correct:25 #Tested:27 Testing Accuracy:92.5%\r",
      "Progress:2.7% Speed(reviews/sec):578.1 #Correct:26 #Tested:28 Testing Accuracy:92.8%\r",
      "Progress:2.8% Speed(reviews/sec):581.6 #Correct:27 #Tested:29 Testing Accuracy:93.1%\r",
      "Progress:2.9% Speed(reviews/sec):583.7 #Correct:28 #Tested:30 Testing Accuracy:93.3%\r",
      "Progress:3.0% Speed(reviews/sec):586.2 #Correct:29 #Tested:31 Testing Accuracy:93.5%\r",
      "Progress:3.1% Speed(reviews/sec):590.4 #Correct:30 #Tested:32 Testing Accuracy:93.7%\r",
      "Progress:3.2% Speed(reviews/sec):592.7 #Correct:31 #Tested:33 Testing Accuracy:93.9%\r",
      "Progress:3.3% Speed(reviews/sec):594.2 #Correct:32 #Tested:34 Testing Accuracy:94.1%\r",
      "Progress:3.4% Speed(reviews/sec):591.4 #Correct:33 #Tested:35 Testing Accuracy:94.2%\r",
      "Progress:3.5% Speed(reviews/sec):594.3 #Correct:33 #Tested:36 Testing Accuracy:91.6%\r",
      "Progress:3.6% Speed(reviews/sec):593.8 #Correct:34 #Tested:37 Testing Accuracy:91.8%\r",
      "Progress:3.7% Speed(reviews/sec):597.6 #Correct:35 #Tested:38 Testing Accuracy:92.1%\r",
      "Progress:3.8% Speed(reviews/sec):593.0 #Correct:36 #Tested:39 Testing Accuracy:92.3%\r",
      "Progress:3.9% Speed(reviews/sec):590.2 #Correct:36 #Tested:40 Testing Accuracy:90.0%\r",
      "Progress:4.0% Speed(reviews/sec):593.6 #Correct:37 #Tested:41 Testing Accuracy:90.2%\r",
      "Progress:4.1% Speed(reviews/sec):597.2 #Correct:38 #Tested:42 Testing Accuracy:90.4%\r",
      "Progress:4.2% Speed(reviews/sec):590.4 #Correct:39 #Tested:43 Testing Accuracy:90.6%\r",
      "Progress:4.3% Speed(reviews/sec):589.3 #Correct:40 #Tested:44 Testing Accuracy:90.9%\r",
      "Progress:4.4% Speed(reviews/sec):579.9 #Correct:41 #Tested:45 Testing Accuracy:91.1%\r",
      "Progress:4.5% Speed(reviews/sec):568.7 #Correct:41 #Tested:46 Testing Accuracy:89.1%\r",
      "Progress:4.6% Speed(reviews/sec):567.9 #Correct:42 #Tested:47 Testing Accuracy:89.3%\r",
      "Progress:4.7% Speed(reviews/sec):573.0 #Correct:43 #Tested:48 Testing Accuracy:89.5%\r",
      "Progress:4.8% Speed(reviews/sec):573.0 #Correct:44 #Tested:49 Testing Accuracy:89.7%\r",
      "Progress:4.9% Speed(reviews/sec):576.5 #Correct:45 #Tested:50 Testing Accuracy:90.0%\r",
      "Progress:5.0% Speed(reviews/sec):578.2 #Correct:46 #Tested:51 Testing Accuracy:90.1%\r",
      "Progress:5.1% Speed(reviews/sec):581.2 #Correct:47 #Tested:52 Testing Accuracy:90.3%\r",
      "Progress:5.2% Speed(reviews/sec):579.2 #Correct:48 #Tested:53 Testing Accuracy:90.5%\r",
      "Progress:5.3% Speed(reviews/sec):581.1 #Correct:49 #Tested:54 Testing Accuracy:90.7%\r",
      "Progress:5.4% Speed(reviews/sec):584.5 #Correct:50 #Tested:55 Testing Accuracy:90.9%\r",
      "Progress:5.5% Speed(reviews/sec):587.6 #Correct:51 #Tested:56 Testing Accuracy:91.0%\r",
      "Progress:5.6% Speed(reviews/sec):580.5 #Correct:52 #Tested:57 Testing Accuracy:91.2%\r",
      "Progress:5.7% Speed(reviews/sec):583.7 #Correct:53 #Tested:58 Testing Accuracy:91.3%\r",
      "Progress:5.8% Speed(reviews/sec):586.7 #Correct:53 #Tested:59 Testing Accuracy:89.8%\r",
      "Progress:5.9% Speed(reviews/sec):589.1 #Correct:54 #Tested:60 Testing Accuracy:90.0%\r",
      "Progress:6.0% Speed(reviews/sec):588.2 #Correct:55 #Tested:61 Testing Accuracy:90.1%\r",
      "Progress:6.1% Speed(reviews/sec):588.1 #Correct:56 #Tested:62 Testing Accuracy:90.3%\r",
      "Progress:6.2% Speed(reviews/sec):591.0 #Correct:57 #Tested:63 Testing Accuracy:90.4%\r",
      "Progress:6.3% Speed(reviews/sec):587.8 #Correct:57 #Tested:64 Testing Accuracy:89.0%\r",
      "Progress:6.4% Speed(reviews/sec):587.7 #Correct:57 #Tested:65 Testing Accuracy:87.6%\r",
      "Progress:6.5% Speed(reviews/sec):590.3 #Correct:57 #Tested:66 Testing Accuracy:86.3%\r",
      "Progress:6.6% Speed(reviews/sec):591.0 #Correct:58 #Tested:67 Testing Accuracy:86.5%\r",
      "Progress:6.7% Speed(reviews/sec):591.4 #Correct:59 #Tested:68 Testing Accuracy:86.7%\r",
      "Progress:6.8% Speed(reviews/sec):593.8 #Correct:60 #Tested:69 Testing Accuracy:86.9%\r",
      "Progress:6.9% Speed(reviews/sec):595.4 #Correct:61 #Tested:70 Testing Accuracy:87.1%\r",
      "Progress:7.0% Speed(reviews/sec):597.0 #Correct:61 #Tested:71 Testing Accuracy:85.9%\r",
      "Progress:7.1% Speed(reviews/sec):600.1 #Correct:62 #Tested:72 Testing Accuracy:86.1%\r",
      "Progress:7.2% Speed(reviews/sec):602.7 #Correct:63 #Tested:73 Testing Accuracy:86.3%\r",
      "Progress:7.3% Speed(reviews/sec):604.9 #Correct:64 #Tested:74 Testing Accuracy:86.4%\r",
      "Progress:7.4% Speed(reviews/sec):607.1 #Correct:64 #Tested:75 Testing Accuracy:85.3%\r",
      "Progress:7.5% Speed(reviews/sec):608.4 #Correct:65 #Tested:76 Testing Accuracy:85.5%\r",
      "Progress:7.6% Speed(reviews/sec):610.7 #Correct:66 #Tested:77 Testing Accuracy:85.7%\r",
      "Progress:7.7% Speed(reviews/sec):612.9 #Correct:67 #Tested:78 Testing Accuracy:85.8%\r",
      "Progress:7.8% Speed(reviews/sec):614.6 #Correct:68 #Tested:79 Testing Accuracy:86.0%\r",
      "Progress:7.9% Speed(reviews/sec):616.1 #Correct:69 #Tested:80 Testing Accuracy:86.2%\r",
      "Progress:8.0% Speed(reviews/sec):618.1 #Correct:70 #Tested:81 Testing Accuracy:86.4%\r",
      "Progress:8.1% Speed(reviews/sec):616.1 #Correct:70 #Tested:82 Testing Accuracy:85.3%\r",
      "Progress:8.2% Speed(reviews/sec):617.6 #Correct:71 #Tested:83 Testing Accuracy:85.5%\r",
      "Progress:8.3% Speed(reviews/sec):619.3 #Correct:72 #Tested:84 Testing Accuracy:85.7%\r",
      "Progress:8.4% Speed(reviews/sec):621.6 #Correct:73 #Tested:85 Testing Accuracy:85.8%\r",
      "Progress:8.5% Speed(reviews/sec):622.5 #Correct:74 #Tested:86 Testing Accuracy:86.0%\r",
      "Progress:8.6% Speed(reviews/sec):623.2 #Correct:75 #Tested:87 Testing Accuracy:86.2%\r",
      "Progress:8.7% Speed(reviews/sec):625.2 #Correct:75 #Tested:88 Testing Accuracy:85.2%\r",
      "Progress:8.8% Speed(reviews/sec):627.2 #Correct:76 #Tested:89 Testing Accuracy:85.3%\r",
      "Progress:8.9% Speed(reviews/sec):626.0 #Correct:77 #Tested:90 Testing Accuracy:85.5%\r",
      "Progress:9.0% Speed(reviews/sec):626.8 #Correct:78 #Tested:91 Testing Accuracy:85.7%\r",
      "Progress:9.1% Speed(reviews/sec):628.2 #Correct:79 #Tested:92 Testing Accuracy:85.8%\r",
      "Progress:9.2% Speed(reviews/sec):630.0 #Correct:80 #Tested:93 Testing Accuracy:86.0%\r",
      "Progress:9.3% Speed(reviews/sec):631.5 #Correct:81 #Tested:94 Testing Accuracy:86.1%\r",
      "Progress:9.4% Speed(reviews/sec):633.3 #Correct:82 #Tested:95 Testing Accuracy:86.3%\r",
      "Progress:9.5% Speed(reviews/sec):634.6 #Correct:83 #Tested:96 Testing Accuracy:86.4%\r",
      "Progress:9.6% Speed(reviews/sec):635.6 #Correct:84 #Tested:97 Testing Accuracy:86.5%\r",
      "Progress:9.7% Speed(reviews/sec):635.6 #Correct:85 #Tested:98 Testing Accuracy:86.7%\r",
      "Progress:9.8% Speed(reviews/sec):637.3 #Correct:86 #Tested:99 Testing Accuracy:86.8%\r",
      "Progress:9.9% Speed(reviews/sec):639.0 #Correct:87 #Tested:100 Testing Accuracy:87.0%\r",
      "Progress:10.0% Speed(reviews/sec):640.5 #Correct:88 #Tested:101 Testing Accuracy:87.1%\r",
      "Progress:10.1% Speed(reviews/sec):641.9 #Correct:89 #Tested:102 Testing Accuracy:87.2%\r",
      "Progress:10.2% Speed(reviews/sec):643.3 #Correct:90 #Tested:103 Testing Accuracy:87.3%"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Progress:99.9% Speed(reviews/sec):627.6 #Correct:849 #Tested:1000 Testing Accuracy:84.9%"
     ]
    }
   ],
   "source": [
    "mlp.test(reviews[-1000:],labels[-1000:])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 项目 4 到此结束。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAl4AAAEoCAYAAACJsv/HAAAABGdBTUEAALGPC/xhBQAAACBjSFJNAAB6JgAAgIQAAPoAAACA6AAAdTAAAOpgAAA6mAAAF3CculE8AAAB1WlUWHRYTUw6Y29tLmFkb2JlLnhtcAAAAAAAPHg6eG1wbWV0YSB4bWxuczp4PSJhZG9iZTpuczptZXRhLyIgeDp4bXB0az0iWE1QIENvcmUgNS40LjAiPgogICA8cmRmOlJERiB4bWxuczpyZGY9Imh0dHA6Ly93d3cudzMub3JnLzE5OTkvMDIvMjItcmRmLXN5bnRheC1ucyMiPgogICAgICA8cmRmOkRlc2NyaXB0aW9uIHJkZjphYm91dD0iIgogICAgICAgICAgICB4bWxuczp0aWZmPSJodHRwOi8vbnMuYWRvYmUuY29tL3RpZmYvMS4wLyI+CiAgICAgICAgIDx0aWZmOkNvbXByZXNzaW9uPjE8L3RpZmY6Q29tcHJlc3Npb24+CiAgICAgICAgIDx0aWZmOk9yaWVudGF0aW9uPjE8L3RpZmY6T3JpZW50YXRpb24+CiAgICAgICAgIDx0aWZmOlBob3RvbWV0cmljSW50ZXJwcmV0YXRpb24+MjwvdGlmZjpQaG90b21ldHJpY0ludGVycHJldGF0aW9uPgogICAgICA8L3JkZjpEZXNjcmlwdGlvbj4KICAgPC9yZGY6UkRGPgo8L3g6eG1wbWV0YT4KAtiABQAAQABJREFUeAHsvQv8HdPV/7+pKnVrUCSoS5A0ES1CCC2JpPh7KkLlaV1yafskoUk02hChjyhyERWXIMlTInFpRCVBCRIJQRJFtQRJCXVLUOLXoC7Vnv+8t67T9Z3Muc85Z+actV6v+c6cmX1Z+7NnZn++a63Ze4NMIM7EEDAEDAFDwBAwBAwBQ6DqCGxY9RqsAkPAEDAEDAFDwBAwBAwBj4ARL7sRDAFDwBAwBAwBQ8AQqBECRrxqBLRVYwgYAoaAIWAIGAKGgBEvuwcMAUPAEDAEDAFDwBCoEQJGvGoEtFVjCBgChoAhYAgYAoaAES+7BwwBQ8AQMAQMAUPAEKgRAka8agS0VWMIGAKGgCFgCBgChoARL7sHDAFDwBAwBAwBQ8AQqBECRrxqBLRVYwgYAoaAIWAIGAKGgBEvuwcMAUPAEDAEDAFDwBCoEQJGvGoEtFVjCBgChoAhYAgYAoaAES+7BwwBQ8AQMAQMAUPAEKgRAka8agS0VWMIGAKGgCFgCBgChoARL7sHDAFDwBAwBAwBQ8AQqBECRrxqBLRVYwg0GgLvvfee22CDDdzWW2+diqahZ+fOnVOhqylpCBgCjYuAEa/G7VtrmSFgCPwbgT59+jiIookhYAgYAvVGwIhXvXvA6jcEGgiB2267zbVt29ZbwrCGDRo0KNs6Of/kk09mz2GFIh3nIEYQJH6LJW38+PHrpZ06daq3so0cOTJ7LdcBaSgLvUwMAUPAEEgCAka8ktALpoMh0AAIQJwgWi+99FK2NZAkyBTSo0cPv1+wYEF2T57999/fbz179mxBkLgGcdLkjYyc41oxMm7cOJfJZNysWbOKSW5pDAFDwBCoOgJGvKoOsVVgCDQHAliVIEQDBw70ZGft2rWuVatWTojWiSee6IGQ32L54jwEjd+QM4gS2xNPPOF23333FmSMAiQNpMrEEDAEDIG0IWDEK209ZvoaAglFQAgXZAmrVNgyBWGCiAnhEgIG8RIrGefE1UggPOchc5KHpp999tkJRcDUMgQMAUOgMAJGvApjZCkMAUOgCAQgScRsCaGCgEG0tECyIFKkgUzhZiSdyJQpU7IWL7F8sSediSFgCBgCjYCAEa9G6EVrgyGQAARwF0KqIFe4ASFU/NYicV7ylSFpESFflCHWL53Pjg0BQ8AQaBQEjHg1Sk9aOwyBOiMg1i2C4XEXQq7knKgG0eKcEDIhXrgpsWphBZOvH7XLUfLb3hAwBAyBtCNgxCvtPWj6GwIJQQDyJBYtyBVuQ7F66ekchGyRVixdNGH+/Pk+MF83hzI5b2IIGAKGQKMgsEEQP5FplMZYOwwBQyD5CDA3F4H3uCMtUD75/WUaGgKGQLwImMUrXjytNEPAEMiBAG5E3IeQLixiWLMqEaxo4o6M2lOPiSFgCBgCSUNgo6QpZPoYAoZA4yOApSsc/1Vqq3FZmsG+VNQsvSFgCNQbAXM11rsHrH5DwBAwBAwBQ8AQaBoEzNXYNF1tDTUEDAFDwBAwBAyBeiNgxKvePWD1GwKGgCFgCBgChkDTIGDEq2m62hpqCBgChoAhYAgYAvVGwIhXvXvA6jcEDAFDwBAwBAyBpkHAiFfTdLU11BAwBAwBQ8AQMATqjYARr3r3gNVvCBgChoAhYAgYAk2DgBGvpulqa6ghYAgYAoaAIWAI1BsBm0C13j1g9RsCTYTAgw8+6F555RW3/Nnn3RtvvOHWrH7DPfjgovUQ+MFJp7gtttjCdezYwX1t553c4Ycf7r7yla+sl85OGAKGgCGQNgRsAtW09ZjpawikDIG5c+e6hx9Z4u6+607Xuk0b981993e77b6b23PPPd1WW27puh7cpUWLnn1uhXv1tdfc6tVr3EsvveSeefpP7q475zrI2JHf6eF69eplJKwFYvbDEDAE0oSAEa809ZbpagikBIH/9//+n5tx401uzuzZXuPeJ3zPHdG9u+vYoX1ZLVi95k0379773WPLlrr/mzrZ/XzE2e4npw92u+66a1nlWSZDwBAwBOqFgBGveiFv9RoCDYrAlVde5a65+mq3X+cD3Kl9+7qjj+wZa0uxiP3619e5yydeagQsVmStMEPAEKgFAka8aoGy1WEINAECxG9dcMEv3RZbbuVOO/302AlXGEIhYPPuvsudM+oc169fv3AS+20IGAKGQOIQMOKVuC4xhQyB9CFw5VWT3DWTJrnThw5zw4acXtMGzLtvvrtk3NggfmzHwNJ2lcV/1RR9q8wQMARKRcCmkygVMUtfFgKdO3d2G2ywgXvyySfLyl+LTFOnTnVbb7211xNdx48fX4tqU10HsVyDBp/uFix4wF1/w/Saky7Aw5V58y23uO2338Ed1OUg98c//jHVmJryhoAh0NgI2HQSjd2/1roiEYAQDho0qEXqkSNH+t9nn312i/P243MEIF19+w1wm2++uZs8+VrXpvUOdYOGuideNsF/Lfn9//6+m3nrTPfNb36zbvpYxYaAIWAI5ELALF65kLHzVUWAaQJ69uyZtS5xzDmkT58+/ry2OElaOcceq5RskKb33ntvvfxY2tgKCdYu5MQTT3SZTMaNGzfO/16wYIHf25+WCAjpatt2D3fLzTfWlXRpzXBzjhg5ykG+zPKlkbFjQ8AQSAoCRryS0hNNpgfkSpMajiFXSI8ePfxeX8ci1apVKzdw4EBvmRJrlE8Y/IE4SX45Bzkr1rUp6SBeiOzlvJRpe+c06cLKlDT50YC+Rr6S1immjyFgCGQRMOKVhcIOaoUAli0Izf777++tS1iYOJbzkCtIlpAeCBjWLNKwh2RxvGrVKp9/7dq1nqyRXpO13Xff3XHtiSeeKNg0sZZRLyJ7zsu1goU0SYIxYz+PfUsi6ZIugHwR6D98+JmeKMp52xsChoAhUG8ELMar3j3QhPVDiCBbECixXImbUeCAWEGi2ISAYYWSY/Zt27aV5Nm9XOcE6YVAZRPYQUUITJ8+3d05d45bGEwdkXTB7bj8mWfc6T8Z6t2hSdfX9DMEDIHmQMAsXs3Rz4lrJXFXEq+FcuJeFEW1qw/yBYGSc6TBKgZ5C2/lBsILQRMCKFYuzss10a1Z93/5y1/c2DFj3cRggtR6BtKXgv/o0ef79SAhjCaGgCFgCCQBASNeSeiFJtPhtttu85YryBZB7JAobakCDrFWQc4gXljAIEDsEcpgi0ukXOpCpGw5H1c9aS5n1Lm/cCf0+X7VJ0aNEyMI4lkjz/GEkdg0E0PAEDAE6o2AEa9690AT1i8WJFyNfJWIy1AsTAKHkCw5L9Yu3JQQNc7L14/yZSNzcEl6KafYPWUiEC7KExdo2BJXbHmNlo5Z6f/wxON+fcS0tY15vo4+5rtOYtPSpr/pawgYAo2FgBGvxurPVLQGMqNdghxr4iONELIFCZPrXJsyZYq3lAmB4xxlzp8/v2y3IJYtytVlYo3TelJPs8rU/7vOB6unxcUY7qcf//hHbsIl4xzuUhNDwBAwBOqJgC0ZVE/0re68COB+JBZMSFXexHaxaghg7Ro8aLBbsXJF1eqoRcHDzxzhvvjFjdwl48fWojqrwxAwBAyBSATM4hUJi52sNwK4DWXiU23tKkcv3I/ijozaSz3llN0MeW75za2u74Afpb6pWL34ItNivVLfldYAQyDVCJjFK9Xd17jKS7wW7sZZs2Y1bkMT3jJICu7X5c8+7zp2aJ9wbQurd2yv3u743r1c//79Cye2FIaAIWAIVAEBs3hVAVQrsnIEmPiUqSKMdFWOZSUlzJ071/3PwMENQbrAoUewOsKSpY9VAonlNQQMAUOgIgSMeFUEn2U2BBobAUjK3p06NUwjj+je3f3f1MkN0x5riCFgCKQPASNe6esz09gQqBkCix9c5Dr/e+60mlVaxYpwl3732OMcHwyYGAKGgCFQDwSMeNUDdavTEEgBAjL1QteDu6RA2+JVbNt2D/f0088Un8FSGgKGgCEQIwJGvGIE04oyBBoJAYjXfp0PaKQm+bbstvtu7rXX32i4dlmDDAFDIB0IGPFKRz+ZloZAzRHAHbf99jvUvN5qV7jnnnu6N94w4lVtnK18Q8AQiEZgo+jTdtYQMASaHYFWW2/jNt5ks2aHwdpvCBgChkCsCJjFK1Y4rTBDoHEQ+Ne//tU4jVEt+cY+ndxvbrlJnbFDQ8AQMARqh4ARr9phbTUZAoZAAhBI63qTCYDOVDAEDIEYEDDiFQOIVoQhYAgYAoaAIWAIGALFIGDEqxiULI0hYAg0DAJMCttur3YN0x5riCFgCKQLASNe6eov09YQqAkCrNG4ukG//PvbunUNOU1GTW4Mq8QQMAQqRsC+aqwYQiug1ggwzcEf//hHvzHXFNsrr7zSQo2tttrKffOb33Rf+cpX/P7www93bCa5EYBsgSsCbh07dmjIdQ3ff//93CDYFUPAEDAEqoyAEa8qA2zFx4MAizXfcMMNfqkXSAEkCmLVv3//LLnSNQkhYw+Z+OlPf+r+9Kc/uV69ernjjjvOb5TT7CI4gYPgKphAxGbPuUN+Nsz+xRdXuS5dDmyY9lhDDAFDIF0IbJAJJF0qm7bNggAD/+WXX+43SAHkCdK06667lgUB5QmBg4xR1ujRo8surywlEpBJky2wzIfnBhts4N5YvcY10peAJ518qvtOzyM8aU9Ad5gKhoAh0GQIWIxXk3V4GpoLQRJChFsRsgRZgHjlIwmF2gZ5w0ImrkrS77bbbv4cdTayQDRpNxuCxZCtEJ4sKP3Io0t8nkb584cnHvdtb5T2WDsMAUMgXQgY8UpXfzW8thADXIjsIVzsIQhxC4QD1+XLL7/sIF38xrrWSAJ2stE+cGTjuFjpcuCBgYv26WKTJz7dvPvmu9Zt2pSEQeIbZQoaAoZAqhCwGK9UdVdjK4tFCzKEtYvjWggkRAieWMPQIa3xXxAtEUhWpXLMMUe74cPPrLSYxOR/5JFHXZsdd0qMPqaIIWAINB8CFuPVfH2euBZjcRKSAAkqxSITZ2PQA/KFWxPyheUt6YLO8iUiugqOcerdrVt3d9pPhrg+3zs+zmLrUlb7du3d5CmTIz/IqItCVqkhYAg0HQIbNl2LrcGJQkBIl7gX60W6AAUrF8QP8sKmCU2SQIMYiguRY9GXfTWk9/HHuwXz51ej6JqWed20GW6v9l/3eHGfaetgTRWxygwBQ6CpETCLV1N3f30br0kXFqYkCfrg7mRwToLlC4LFhkAa2GohkE8I3d/+9je3/NnnXccO7WtRbVXqwHI3dOhQd/zxvbPl07+0z8QQMAQMgVohYMSrVkhbPS0QSDLpEkXrTb4gPeCE1JJsSfshJUy5AenaY489XbfuR7ipU66Vy6naY+26acYNbtGihevpbeRrPUjshCFgCFQRASNeVQTXis6NAAM6pIJBL8kiVi/0rEXAvcYDS1st6ozCH9I5YMCAFpd23mlnN+XX17mjj+zZ4nzSf6xe86Y7+aST1rN2ab3BvZ54a13s2BAwBBobASNejd2/iWwdXy0ysGPRqRexKAUYXFES/1VKvmLTarKVBLcXZPOKK67Iqs/yS8S+4eqcPn2Gu/mWW1I1oeq5vxjtXn5plbvl5huzbYo64H7EspiGezJKfztnCBgC6UDAiFc6+qlhtGRw23fffd1TTz2ViNipYoDFMseADFnEUlepUB44iCSBbKELekG6pk+fLqq5XXbZxZOub3zjG45FLpj1ve0ee7qLLxydTZPkA+btGj5sqLv3vnt9HxbS1chXIYTsuiFgCFSKgBGvShG0/CUhAMlgw+qVJsHiI1NNlGMR0WSL/EkI2Nf4ox/9wnqWIpCtRYsWeQvQv/71L/fPf/7Tvfjii8F6l8e5kaPOcz8a0FeSJnL/7HMr3Am9j3Njxo5tEVBfSFkjX4UQsuuGgCFQCQJGvCpBz/KWhAAWIwgXA1s55KWkyqqQGGLCVixpxDXHhiSRbHnFgj/0B5a8V155RU65fv36uYkTJ3q9IVxs//jHP9ynn37qZs2a5X71q8vc9Bk3uq4Hd8nmSdIBcV2DB5/m2rdv7y4ZP7Zk1eQexdJpYggYAoZAnAgY8YoTTSsrLwIMYpAWLEdpFAZjiBdkKhdxJA3WI4T2siVZiC+TLxdFzzPOOMOTLn4L6YJwffTRR+7jjz92K1eudA8uetDNvHWmu/GmWxJHvoR07RgsDXTttVdLs0reC2lOeh+W3DDLYAgYAnVFwIhXXeFvnsrF2iWDWVpbDmlkINZWL0220vRlHH0S/nJx2rRp3tpFPBfuRbFyQbr+/ve/uw8//NDde++97pBDDnG/u+tud+usZJEvIV3cXzOmT8tJkIu9/+R+NfJVLGKWzhAwBAohsGGhBGm9PnXqVLfBBhv4rWfPnqlrhtafdmiRdrFfsGCBvlTwuG3btllcxo8fXzB9XAmEeMVVXr3KgXixmDaWItkYlLGEseWyhNVL31z1EkSvSRdfLhLPhYsR0oWlCyvXJ5984gnX+++/7+fzItaNjyPWrVvnDu92mPveCSe6Q7oe5Jgnq96yZOljWffinXfMiaUv6FsEcm1iCBgChkAcCGwURyFWhiGQDwGsBg899JD/Oi5furRcY0JR3IlxfOFY6zajd74vFyWI/rPPPsuSLqxcEK9nnnnGbbPNNt7qBTnbeOON3dH/31Fu+x22c2MuusAtD66PGPGzukw1AfGbMG6MO33IEDds6JBYYYV8gRvkK2kfRcTaUCvMEDAEaoJAw1q8aoKeVVIUAlhJevXqFYsFoqgKq5gIqxZB57QpbQJ5QH89XQRfLjK1B3shXbgXcS1+8MEHnnBBNP2SQcuXu+22285bwkiLxXWjjTZyPXr0cNdff737y19edid9/weOKRxqJXy5OHDQaZ50sfh13KRL2oElEwJmli9BxPaGgCFQLgJGvMpFrsr5Bg4c6F0+WBbY0iyQlLitQ08++aTDVdqnTx+HK1m7X+UYtyrXRo4c6W677Tb33nvvxQIj5CVtxEusNXq6CNyKMl0ErkWxcgnpwp0opOuuu+5yXbp08WkAEWvXJpts4rdNN93U7bXXXu6GadcF83yd5OfNYr6vahIwYrnGjJvgp4uAFC17bJknlbF0cI5CjHzlAMZOGwKGQEkINBXxYqCWQZn9oEGD3EsvvZQTMOKnSKPzbL311n7AzzWI67TkZ+vcubMvQ2KqikmTL8YrrDCkQuqgbI45V45EtRnygj7lCm5GyEocInjSRiFUnIsS+pZrQtAgYuTJ1XdRZUSdE3dTWqwfxKKBv54ugi8XCaSHTIS/XMStqEnX8sDS1bp1a5/uC1/4QpZsbb755o4N4vWlL33JffGLXwzixvq7JUuXBCTtwCwBm/Xb2VEwlnWOOK7hZ45w3YP2LH/maf9lJdNF0I5aiJAvMDUxBAwBQ6AsBAJrSkPKlClTMBP5LXCFZNjkt963atUqs2rVqvUwOPvssyPTS17yPfHEE+vlk+vsw2WMGzfOpy8mjdaf9Fp0/sAyllNPqU/n3X333bPpw9f5rcsOH1NXqRK4sTLB7OelZotMX0i/sL65foNBVJ9HVprjZOA6zQTEJcfV5JxGxzAOnAtchZmAcGUCt2Im+FoxE7ghM2vWrPG4BIQy8/DDD2fmzZuXmT17dmbw4MGZ3/zmN5mAzGcCy1dm4cKFmccffzzz/PPPZ1577bXMu+++mwniwDJBIH4msJr5skEgILiZ4KOKzOGHd8u026td5qfDf5659bbbM8uffb4kgO659/7MqPPOz5Yz4qyRmZdffrmkMqqROLAWVqNYK9MQMAQaHIGmCK7PZREJBiRv/cCqNX/+f+JSsJCIdYo0UYLVBEtQMIC7gIRFJSlYBpkK1RNZsDqZzxKFdWf//ff3MTgqS+QhFjLS5xPqCkiLCwhlvmQtrmEVIjamUilGv2LrwBKGizIgzsVmWS8dVi8+Gkiy5Fpz8bDDDst+uainiyCmS+K6CKjH5Xj//fd7axnWLCxbm222mdtiiy38xrG2dmENE2suuGAdwp3Jxn2w+OFH3Nw5c9x/n3hCUGY317rNjm777XdwXw3ixsKCNQtd7rpzrvvusce5Aw88wJ1xxrDYXdbhekv5jRVRrIml5LO0hoAh0NwIbNgszYeAMNAGRNqtXbvWExJpuyZmECpNhiAakDLysRF7JRJOK+f1Xsdq5SIsxaTRZYaPA0tQVr/AUtbicj5iphNq0qWxglgSPC0CNrS7WIGcMEBVKrpPpCz0pO1s9FF4Y4Z1roEv/aiFGLFy3bGUA/FKqruJIHqmvdALXbPmIvpCugiMJ54L0sV0EfLVorgX2XPuz3/+s9ttt918PNeXv/xlT7aYdoINFyPniPMi3itMujTWgheB7yxUzXM0ceJlbuD//MhtteVmbvMvb+I23WRjv20WHH/68YeuT0DOzhx+hk/L1BDnnTsqUaRL2ifkC8xNDAFDwBAoCoHgJdiQEnbVhV1LwSDdwgUTkDGPQzgf6cKi3Za4HLUEoGfLJV2UFJMmrIcuR+cPSIW+5I8DspHVgbTSNi7iZpP8pENwmco59mGsyE87JQ26FSvnn39+hq0SoX6pm30uN2+hOsK44AouV3AzBSSm3OxVyxeQ4kzwhWILvPgNhrgXcQXiEgysSZl33nkn8+qrr3qX4e9///vMAw88kLnzzjszAWH1rkVcjLgagwlTM4888kgmCMz398abb76ZCYLuM4FFzLsqcVlSdjMLLnWwNzEEDAFDoBACTWHxwtoRtniEf4sVB0uICC5Ebe2R81hQRMin88h59lF59fVi04Tz6N8nnnii/umPw/Xm+4CADFp/rEhhbMBB11OoPK0QVhYJRtfnSznW+pEPKxZ6lipYHHXbwuWWWl7S0uPOA+tyv1wkqJ4lgQi2X7x4sTvqqKO8ZWvLLbf0bkMsXbgZsXQRTM9UEoUsXUnDqFr6iOvZLF/VQtjKNQQaB4GmiPHSg22hrtOkItfgDhHRIqRNn+M4nC58vdg0UfnkXFTbwvXm0k/K0NchI8Tp5BOdPl86uRb3F2dRbZa6Cu3Jq/u4UPq0XIfglrLmIq5Eiediz3JAzFSPGzIImPeLS8tXi5Atjonpkq8XIV0bbvj5/22F7pe0YFipnpAviWmM+56vVDfLbwgYAslBoCksXsmB2zSJA4FyLVUQxnLzxqF3tcpgOaZu3br5ObekjuDLRT/Ra2Dy9hYs4rmwZgnhCsdzEesF6YJQvfXWW65Tp04+lgsrFxYviJfEcwnp0oH0Um+z78XylfQPL5q9n6z9hkA9ETDiFUJfW1NyDdJhi0/YwhQqsqo/o3QM66fbFKWM1h83JYN1vi2I8YoqJvIcXzRiBahEwq5TAu2jgu3z1YGVi69XNTbhcvPlT+q1ctdcZGJUXItYuiBd9DdfLgaxXu673/1uC9KFpSuKdCUVk3rrBflCjHzVuyesfkMgmQg0hauxFOi1e5FBmi8ewwO0/lIQ0qLzlFJXHGnRT8dfUab+ShP9ChEvrT9EjnZrMlaJnhCvOOJeaKN8hYh+fIXJRt/k6gPS0R5IV5R7MdyvlbSz1nnBtNw1FyFcuBexgPF1I5YrSJe2dEksl54uAtdipVYuyAhu0b+te9899tjvPWzowrQRSDDfl9uv8wH+uH27doG1bXPHl4NCZvyFFPzhvqetbBybGAKGgCEgCBjxEiT+vWeAZ0Bn0EawkmDhkUGa35rYhEnPv4up2S48txa/9dQQxegH8ZLYJ9rN/GS0WQiZlCmYcE1/YFCosXEQLwLjwV10kDqlL4SUyflCe/SX9hVKG3VdYnmirlX7HHhCRnQQPWstBl9a+iB4XIYEyIt7EauWTBkB6eJYgughUkwHQcB88OWjO+SQQ7LxXJAurlXqWgSr3919j3sg6L81q1d7YrV3p33cET16ujZtWnu4mDICYe3FV4MYM+Spp/7onnt+pbvjjjt9vm8Hc391PbiLnyrDJ0j4HyFftD9txDHh0Jp6hkCqETDiFeo+rCcM8kJesJRARKKEtHxhV29BV9E3rAttKUZIB6lEsBKxJE+UQNBKIV0QhNGjR0cVVdI5SBKEL+wuLKmQfyeGjFbab/WyZDCIE0Svl/9hglIW7iagG8LFRqC8zNGFRYlNSBfXSIMFi2B5SBdz3FFu1PxcfLmIQNJKkdmz57irrrrKk6YT+nzfnTXyHHf0kdHPkpTbsUN7x4botBCyBxYudLPn3OHGjR3nTh8yxB373f9ykJskC/pBlI18JbmXTDdDoLYIWIxXBN6QkELkAtLFhJ3s6yn5iBXkopCbUXSnvYXICGXR5lKEgYe1GuMQCBMTutLmcnDHasmkqmzl5NdtYCCFVNZScNFRpyZdpay5CPmCjEG6IFPEbUG0sHRhMZOvF7WlqxzSBeHq1q27J12n9O3vVqxc4S6+cHQLIlUqbpCxYUNOd1jGJl55lXv55Vf85K5XXjUpFld2qfqUkh5CzHPAPWNiCBgChoBZvHLcA+JexJUVjukSYlbp4J2j6pJOQ5ggRASbSxwT1iF0LMbNqCsjD+QEt50OXqd86uF6qcKAw6zpcf3HD+YQRDYsc+JqlL3WD71lo11x9RcWDMhkLd1HfLk4YMAA3Ty/yDXWLgLjcS/q5X9wL2Lhkngulv/B0kVaXIeQLln+54UXXvAuRpmfi3gvCJfEdLWoNM8P+viSCb8KLFxvOAjXjwb0zZO6/EtYwth+/OMfuYsvvthdM2mSuyrYevbsUX6hVc6pyVct75sqN8uKNwQMgTIQ2CB4ETPLtYkhUDUE+gfL10DA4nA5Vk3JEgqeO3eub0utLBhxrLkI6ULCay5CXo855pi8ay4WA8306dPd2DFjXd8BP3L9+53q2rTeoZhssaSZ9dvZ7n+DJYW6dT/CjR17sXe5xlJwFQoRt2OtraVVaIoVaQgYAmUiYK7GMoGzbMUjQOwQZKURREgXZLLawiBNPZWuuQjp0kH0uBSZn4sPFfbee++S1lyMavNZZ5/jSRcuwFEjR9SUdKFPn+8d7xb6LyXXub79BiTapYflC9KF29jEEDAEmhMBs3g1Z7/XvNUMOAw2aXezQIZwWTJBKVY8kbgtGNRDmXF/uUhMl8RyPf744+7oo48u+8tFdIToIJMnX1tzwiXY6/3wM0e4eXff5WbeOjPx9xrPQ9z3jcbCjg0BQyCZCBjxSma/NJxWuMsYqG8IYpXSLJdffrm33oUtFuHfEEzIZjmCC7MaXy5CumQWeiZKZS1GposgnqvUIPokki7B+rppM9yEcWOMfAkgtjcEDIFEIWDEK1Hd0bjKMP3CbrvtFnyN9nILS1HaWoyVC/IFMconkCfIiQj52AoJBI6ymVlehC8XmS4C4YtEJj3FfaiXAJIger3mImSK6SIIoteWrr/+9a/+/J577pmdo4uyS5ku4qSTT/VTVCTF0oX+WtJGvioh6rrddmwIGALJR8CIV/L7qGE0JF4JSavVC8LFBoksVcij82ENC7tdwaVaXy5CvNj4cnHp0qV+brpyvlyk3RddPCZYWujxxLgXc/XFub8Y7Z55+k9uxvRpZVsfc5Ud93mIOsS8XCtp3PpYeYaAIVA9BIx4VQ9bKzmEAMQDq9dTTz21HukIJU3cT6xXDIwE18cRl0N5+qtIXLE6novgd+o67LDD/BQQWLr0dBHhSVFlugiAC3+5SEwXVi/m59KkCwtXKVYuyp4/f4EbGkxeevucudmJTjmfVMEyt1WwyPe1116dVBWzehn5ykJhB4ZAQyNgxKuhuzd5jWNKCQiFJh3J03J9jcS1iO5xCgQsvOYi5eNaxMUoy//gXsS1qJf/EfIVXnMRgiWuRUgXVi7OrVmzxpMy5jYrh3Sh60FdDnK/DCxefEmYBlm95k3XPfhIIenzfAmWPBdYvSD5JoaAIdCYCBjxasx+TXSrcLFBZNIyrxdkCzepWCTiAhcig/VMW7r0mousvSgxXVi72rZt6yc1lYlRIWFRay4K6WIvli6C6B955BHXvXv3skgXbR40+HRvfZs65dq4IKhJOTLP17LHlqXClScuaSNfNbk9rBJDoOYIGPGqOeRWIQQGwkFMk1iSkopKtXSlXNqul//JteaiWLpYT/Gtt97yVi+W/sEyss0227RYcxGyJV8uYulihnpI18MPP+yOOOIID3Op7kUyEfQ/eNBgP19WLSdHjeu+wOXYsUMHd+6558RVZFXLMfJVVXitcEOgrggY8aor/M1bOaQLFxsDejjIPCmoiEUKkkhQfVxCmyFdpXy5KF8t4l788MMP/VeNb7/9tnv33Xc9sYJgffWrX3UsFyWWLr5oJN7r9ddf9+QMC0o5pIt2Q1z27rSPnyA1LhxqWQ6LbO/d8eup+qrWyFct7xCryxCoHQJGvGqHtdUUQgAyg7sxieQL0sVEqVihIIlxCWWV8uUiJEtci5AuHUTPV4nEbsmai6z+xWANCYNwsSZjt27d3OLFi/2+3DbQP2m2dkm7mVx12222To3VC73pT+7FpP5zItja3hAwBIpHwIhX8VhZyiogQOwUMVRJIl9i6cJChFUOi1ccQll6+Z9ivlwU0gUBg3QR64VArCBYEs+lv1wUSxfE7IILLmhBuhjAS52yYMRZI12rrbdJrbVL+m7J0sfcD/v3cytWrpBTqdhzP0LAjHylortMSUOgIAJGvApCZAmqjQBWIEgJ+3rHfEnslcSg0XYhhaUSFsGNgZP2sZC0yC677OIJJ8H0xXy5COki0B4hZkt/uSiuRc4J6dpwww19/BiuRQikCO1DHxGu6etyXvakxfK3/NnnUzF9hOida39sr97u+N69HIQ/TWLkK029ZboaAvkR+ELg6hmdP4ldNQSqiwD/ye+www5u8ODB7s033/RL2VS3xujScX0yII8cOdKNGzcumwhismLFCv8FYankiwETEnffffdly4NsMZ8W5UK6mCoCS5YE0eNSXLdund841l8uykz0WLgIomcP8SKQXpMuCBdfS4atJOBMvbKhH2QMiwobv0kjMnPmTPevzAbujGFD5FSq9399d6178sk/uO9+979S1Q6sm2zLli3zfZcq5U1ZQ8AQaIGAWbxawGE/6okABEAsEZAgCEstBMJBveyxuuWqV4hJmMzk0lGsZ6V8uQjREvci00Xw9SLkDAsWxAqCpd2L+stFXItsyEMPPZSzHbn05bwQMUlz2cQrXI+ePd2wIafLqVTvCbI/ofdxqXM3atCxwOa6R3U6OzYEDIFkIrBhMtUyrZoRAQiNkBVcjkKGqoUFJAMXILPpS935BjSxEjHwFRIZHDXpYkLUadM+X75G5ueCWOFGhHDxlSMb1i6xdEG6IFMSz4WVi9gwmTIC9yKuRwLphXRRJ7qWI1j0wEC299f9zXUOvpRsFOnYob1r3aaNK6YPk9pm+ibN+icVV9PLEKgVAka8aoW01VM0Ani/sS4hkCJIWJwzxotljdglyBcLd2NhK8aNKMSEgY+8UYLVjK8J9XQREC5mo+fLQ1n+B9KFVQsLl5Au9pAurpEWMgW5wqUI4dKkCzImpAuLGO5FNrArl3jp9lDOgw8ucl0P7qJPp/74m/vu32L+tDQ2yMhXGnvNdDYEPkfAiJfdCYlEAIIDgXnvvfe8NQrLFGQCKxgkLBfpydUYiJKUwaBF+XPmzPGEqxySQhkQEzYt1KGni4AoMQM901II6fr00089sQqTLggY57iO8OUirsQw6WL6iCjSRR7aiW5xCG37wUmnxFFUosr46nbb+Y8FEqVUGcrQz/R3qc9CGVVZFkPAEIgRAYvxihFMK6q6CGCpgoyxJ4aJLwMhTbgJo6xVMigRZE5AOwMVm/5yslKiAjlh4EMPSFetv1wUKxfICwlElzgEK+Arr77hJl42IY7iElPGvPvmuxtnzHC33HxjYnSqRBGeB/o86hmopFzLawgYAtVBYKPqFGulGgLxIwDBggyIMOBAeiBPUQIRYjCCbOUSrlVCvhjwIDy4LbXoNRf1l4viXtRB9MzRxXlckBAp3Ic6iF5PFxHlWpR60SNfWyVdsfsNNvyCwzpkkmwEJD7RyFey+8m0MwQEAbN4CRK2b1oEKrEUQf6woOkg+kJrLmrSVcmXi5A0kUrIo5QR3k+8/Ar30cefpn7i1HC7Vq950+3YprV3/Yavpfk39yL/aEDATAwBQyC5CGyYXNVMM0OgNggwUGE5KzVWRsiOJl2HHXaYD6JnAKzml4uadEEcbbAt/l5J4yLfxbQOyxci/0gUk8fSGAKGQO0RMOJVe8ytxgQiIO6aYlUj1izqy0UC6flK8qWXXvKTooprMe4vF7WeRrw0Gs19LATcyFdz3wfW+mQjYMQr2f1j2tUQgWLJV6EvFzt16uS/THziiSfWmy4iji8XNSRiddPn7Dg/Akyi2m6vdvkTpfiqka8Ud56p3hQIGPFqim62RhaDAO5BtlzWAlyRTGehF7rmy0rIDy5GHUS//fbbu2222cYtWLAgOykqpItAelnoWoLow5Oihmej118u6nagpwyy+nxcxzIha1zlJaWcV197ze3X+YCkqFMVPeS+IO7LxBAwBJKFgH3VmKz+MG2KQADCAdmRPVkgRUwbgcg0ExxjxWIQ4ms/iYHhfC4hLWWz10L5lCF1cK3Ql4sQpnbt2rnFixe71q1b+9nlK/1yUetE+9GpWrLlFpu75cufrVbxdSsXAtwMwj3MfQv5KubeL4QJ9xtlyUbZ+rljzjqpR5479tW8RwvpbNcNgSQiYF81JrFXTKf1EOBlT1yVTJ4qL3T2WKkQecGTVgYFGSTkHF8gyrZeJeoE5EuXV+mXiytXrvQz0GMJK2XNRR1Er9Tz5FD00+fjPAaDSy651N1zz+/iLLbuZY0ZN8Ft9uVNgoW/h9Zdl1oowLMAaeJZKVX0c8dHJFh2ue8gdWyI3IfyjHGOe4c62VM/aeS5k+eVdCaGQDMiYMSrGXs9JW3mhQ3RYgkhhBc3rr5yBhDyMzAwEDAXGGUTqyVzfXFdiwxW7KlXL//Dmoss/4PIl4vMNv/xxx97VyIWFaaMYMO1yDVmrV+7dq13M+69997Zha5lji7IGDPVs+Yiy/8guUgXAxoiA5//EdMfwQic7rjjDl8qujeSDBx0mnv7rTVlr1qQRiy4j+lbCFAxwj85PCfca0KY2JcjlMFzTJkc8wzz3FXj/i1HP8tjCNQaASNetUbc6isKAV7SvJwhWbyo2eIUiAWEjsEoFwHjvI7non7WXJTlf4jpIl4LYsVC15AsSJcQL87J8j+y5iIkZvXq1d5yAOkinktIF2lkzcV8bUX3YgfQfOVwjYGQ8tgYHDXB5Pp+++7nLho7zh19ZE9+NoS0b9fezbx15nrxfHFhmmSQ6Od87eQe4L5HeD7ifu543iB0rPDAc8SxWcCSfMeYbtVAwIhXNVC1MstGgBczL3v+Q4d85Rskyq5EZWQgYoCBgDAIyH/1YdJF/AqDEq4WWXNRky49KSoEDNIlQfRYslhbEaLFuotsTz/9tNt///3ddsHM8FyHdOUKolfqeoJUCSbgSptpC3s9B5muR44hXkcd81138YWj5VSq90uWPubOHXVOsH7mwvXaAR4iWGMa1SJDO8P3EPc/z50QI46rKdTHM4YuPH9C9qpZp5VtCCQFASNeSekJ08MTn+HDh7vzzz/fv4xrCQkkj5c/xAtyIm420eGpp57ywfRYucS9iGuRmefF0iXuRUgXaRC+XNx0002zpEtci5wj7mvbbbd1u+22W1Gki8EKKZUQCMlikNMfB/jCIv4ce+yxfmBmcGY+skmTro4kKhFZE3/q3F+Mdv/49BN3yfixeXUFa8GbhGGikjdzCi5yL0ib5N6HbEGCammBQg/qxbKNHrWsOwXdZCo2KAJGvBq0Y9PULIiO/PcLSSg3hqvSNvPf/r777tuiGL5cxL3IgPC1r33NffbZZ96SJROjhi1dpa65+Oqrr3r3XrjeFkr8+4ceLKOuyznSycZi4oXk29/+th+EGYhlWgysekIyv/GNb7orA/LVCO7Gbt26B8T+f7OkoxA2ch08RSC+pZJfyZukPW3Cysue547+r4fw/EO+eP7q+fzXo+1WZ3MisFFzNttanRQEeOnKC58Xbz3/46VuXIoS56Sni1i4cKFr06aNj9kSS5cmXeWuuYi1i/oY/ASHqL7Jdx3cuC6b6B9VDudol3ydxp42i/tUiCMWO7HsHXPMMe7+++5PPfG6btoMD0k+nHNhpvNgCQNrhHumXv8oeAUq+IOFCcsu1tx6tgEMIVxY28AZbOupTwWQWlZDoCgEzOJVFEyWqBoICOniJcsgkATRVi8ICUsAMRM9li7IV+fOnddzLeovF4nVIlh+s802W8+9KEH0ub5clAEnTD7F5SVWFhn4Sc+AVYhoMa+ZEK1evXpliRYWLbFqaaJFW/X2wgsvOMjX8mefdx07tE9CN5Wlw0knn+q+d8Lx7vjje5eVPyoT9zD3jAj3crj/5FqS9mJh4h6iDXJv1VtH3gNi/TbyVe/esPqrhYARr2oha+XmRSCJpEsUhthAVhicsAgQi0Vw/FtvveX+/Oc/u5133tmtW7fOTxcR9eUipIsAeuK5Sv1yUax+eiCEXCHsGSgLBcRDGCFaxKthQcBFKq7DMNnSBItjPgjQe7k+PpjPa/fd27qJl00QmFK1n3fffDc8mLdr2WPLqkqM6D/ubSSp1jBNupJIEo18perRMmXLQMCIVxmgWZbKEUj6y19cbwMGDPAWDZb+wbXI+oss8YPgXsz35SIEjCB6sXQV++UixI/BhwE8PJ1FLuR1QPw+++zjiZaQLbFmyV7IVJhghUkXv8Xd+OKLL7pLJ1zqZs66zXU9uEsuNRJ7ntiuoUOHxmrtKtRY+i9p1jDcedxbQvALtaFe14k9Y0u6nvXCx+pNNwJGvNLdf6nUnhcqAwAEI4n/cQOquOCYh6tLly5+Y+JULF2PPvqo23PPPbNzdOX7clFIl8zPlWtSVCxZssUREI/+ECtNtoRIRREsIWOSRvIKDmAybdp0P8/Y7353Jz9TI8xUv2zpEnfnHXPqqnO9rWHcX1hB2afBjSdfGKOviSHQSAgY8Wqk3kxBWyBbvPRxm+EGS6JgKWKDhBBsvmLFCtejR49g+ZxLPOEipusPf/iD69ixo59pnklQZY4uPV0EhEziucJzdDEIM6DIVihOK19AvJAjTbI4FoKlLVv6nD4vedlTnmAgRBFrHeSRJYT+69hebtTIEUnsuvV0Yt6uQ7oeVPcA8rBitbaGUR/ua/7hScucWejMuwJ906JzuJ/ttyEQhYARryhU7FzVEIBs8TLF6pVUERcdJIUYLojWpEmT/GzbV199tSdTb775pp/0lPgpIV3EdUHCiAeDdEFW2BDisoRksS8Up0Wevn37enIqMVvsIUVRREssVrIXghXey3X2QraEaFGnCCSLTdogk7w+++yz7rLLJrpLLv2V6/O94yV5Iver17zpTj7pJNe/fz8/S3oilfy3UtoaBkFii1MgLkL24yy32mXxrGD5Qve4Mam27la+IZALASNeuZCx87EjIC/RJLsYaTTESyxGQrxYBuiUU07xXziefPLJHpvnnnvOde3aNRtET0yXuBaJB1u8eLGjzQToFyJaQq6EmELoCPAXosXAA7HbcccdW3xxCIHKRa7kvCZbUh57LUK02GOli9pkLclLL73Uu1tvvOmWxMZ7QboGDz7NtW/fvuBkqRqHJBzzfLCJcE9UIpTFtCUvv/xyKslL/+AjF4TYNBNDoBEQMOLVCL2YkjbwHyuuDnmRJlVtbfGSObuwehF7xcz6t99+u5+SAZL1zDPPuG7dunlL19KlS93dd9/tHn74Ybd8+fKCzWPiUvnyUAfEM23Ft771raxFChIIeWLgfO+997y7U8iUkCu9l/Sk4VjIllYIF6K2aIWJlpAsvcf6RXwbU2rcc888N3nyZDd9xo2JJF/DzxzhVq160c2Y/vnkt7rtaTuGvIvwDLGVIjxv5OHZS6PERRz5QKZnz54egnHjxrmzzz47VXC0bdvWrySB0qXoP3XqVDdo0KBsW3m/1VJ4Z/HOYBWME0880c2aNauW1SeyLiNede6WfA9Tvmt1Vrvk6onpwt3BSzTpwouJDTJDcD3kSzasXQcccIAbNmyYw+LFS+TWW2/16Qu1C3IlRIvpHiBEQvKEIDE4COmCOKED14RYrV271tdLWZp8CdmSctgjlC/xZZpoQaIgW+JC1ARLSJg+h8UPMnnEEUdk3Y+jRp3r5t1zj7v+humJIV9YukaPvsDhCm4E0hW+p3h+9DNUyBpGWqxdDH5J/ZAl3Mao3/LPWiVWL3mf7r777gEpXxVVTaLPif4oGSZeEovJtSlTpriBAwdy6KXexAsltA68MyFgzSwbNXPjre3FI5DvwS6mFF6YaQmQpa0QFiEqxGtxDvchZOSaa67xW6F25wuIj5ohnsFg66239hOiCunSe44hVAwcuDGxYhBPJmQLnYVooRvkStogREvIVnjPdSFaco1zbK+99pqDeDGJKuWBBfvLLvtVMAv+Pu6HQQzVLy8eU/eYL3Ev0vZGJF20iz5nEylkDcPK1a9fv1STLtpKOyCQxIaWQyDHjx+ftRalzdIlfZ3mPURQ+mDkyJFGvNLcmaZ7OhDgv27inCr5b7XWLYVcsEFCEIgGE6guWbIkpyrEZR0exOOwYdEiRgsiJK4+rGaQJDaxVuk9S7cceuih3joRJlySTvLz3y+uR+LKvvrVr2Z11EQLIqUJVZhYaYIl14RsyZ5FtSGDHTp08BhI48EG6R+4sbbccis36pxz3Isvrqrb144yQeqxx/VOXUyXYFrOnntNhOdMiBjkRL4elnOSLo17yCbPFJZz7rlSBGsfgz7SqlWrFtagUsqpd9pyrXSQHm0Bq1c70AHShcuR/mhmArxhvTrB6m0eBHhZslRNOf+p1hMlIR9YvNgOPvhgt/3222dVgvRgBfrVr37lXRcspn3dddc53JGs64hVi+B8JlqVdR2ZB4ypI/hUno1BgW3OnDl+oORYrpGODWsTMWYySz7kC0KH5Qsd33jjDR9jxteVTO5KoD5Ys2eg4Vjv5VjSkI7AfdrDV5ky6Sskk/nKIHnUI2RUSJcAwRI8M2+d6efKOrZXb8cUDrUSrFzn/mK0n5V+zNixTUW6whhDTiBibBxjYeb+gYA1gkC4yvnnDTcXzxUSJiAQALmviYMiHeRAzrEnjeSPwpHwAPLqPPyzUigfehFzpvPxm/NRwnMoaSkbkfw6vegi5bCXfOxFwu188skn5VJ2r9MQp6Ulqt359NfuRdFNl9dMx6kkXtx0+ibkZuIcTFqL3IBc50EIX9dlcBwWHjbK1Tdtrrp03mL103nKOS71xtftBQvy8zBJ++RloXUp5sHW6aOO+Y+b2KY0CZggYkHCIsTGi/vUU0/11yBRd955p4/3YhkhvjhkSSH9JWSYaAnZ0nssXZAgzjFQkgeiBmEjxgxrF1YzdIIAQQIhRxAl+nSPPfbw9zZlCMmCXNGf8luuQbIgZ0K0KEeIFuXSRogeHwjw0QDlCBa+0Tn+MLgzQWmPHkd41yPB7c8+tyJH6spPQ7iYGLV7QDLe+evb7t777q3prPSVt6C6JdDfCJP+NorwDuEDF56TUkQP8szHl0943/H+1gL5kOBwfZ5jrkWRDSFwPJ9RhIaxiY13sBZ5p1NmtUUTIeoK68I5jZ1OD0ZR7Rb9aVtY+Edx//3396cZf2677bZwkqb5nSriRWfxAHCzh0mUPBz6JseUycCBCImSnuWG0mUQkKiFcnhoKDcsnONa+EYtVb9wuaX8LufG1+Vz0/PgaLzkZRH3Q4+bkf/C0yZCSNlDwNh++ctfuhkzZnhrElNEiBuR4HesYbgjV69e7ckTJEoTLPBlk3NcZ+PLyG233dZbtbCSUVYuogVhgjyxCZmC9HXv3t2TPoiZkC1JI0QLi5i2aPFVJmSLPGy0jzaxHV5mfw0bOsSToI02+oLbu+PXHQQsTgsYZE4I1/JnnnaTp0x2UyZf43YNLDwmLRFI4z88LVvQ8hf3NXGTtKtY4f2m3/P5iBdjgn4f6jooI0wmeAeHSZrOwzHPO+9T9iLUowmNnNd7xpZCZev05RxDgiBDIuG281vrLdgxdkSNi1IOe9oXpb+UQRojXqCQAunTp0/OBwP1wzc5N5X2I3MzyEOobwqYvL4hwuXkgib8IJaqX65yC52v5MaXsvM9ODz0cT0UzD9FrFM9B0ZefEKipP2V7rHwMADg8sMiJV8/EgD8+OOPe+IlBEvvtUUL9+G9997r5wLDfYiIRYugeT0jvpAocRNGWbOOPPJIT+SoD5LFJu5DyhOiRWyXEC2NC32FVOqaoq8nXDLOx6Btu83W3gLWrVt37xIkFqtUgbhdOekaN3DQaZ7MvfKXlz3huuXmG8smiKXqkMb0xOeVS6DjaG81njvaI/dpMTrqf47F2pIvH2mIpeK5Zq/HBcqS8hgj9BjCWDN//nyfj7zapRlOq9+t1MeXylKf1lGny6Wz1KmvY0QI66Cv62NtxZK2yXX9G71ENz12cI71a0V/jRf40HYtmujp8nWaZjjeMC2NhDRpRs7ntHQ2m7ZW0dH6vwmIl+5sbgZNwBjIKEsL1/UNo+vSRA4SJw9Hufrpeos9ruTG13XodoXnVhGsK32wCfitJ+nS7eVY92v4Wim/GQBoG5YpSJMQL8jUTjvt5O9VsWixj4rTIjieexP3HqQoF9ESsqX3EDH5zTFWLYjWQQcd5F2HuDzzES0IlxYGM/qJLS6hrHPPPcetWLkicHkNc5tusrG7ZNxYT4KJBYNIYb2K2ojbOunkU137du09cVseWAVZnJv+w8JVT0IRFz7VLId/CrAOJUXieu74p6AU4iXvMXDQ40AuXHgPSjr24feikAXe+7pNjEGadIR/6zFJ/vlHB4gPzzFCfXp80br7BFX4o4kX7dF16mNJxzmtP/gIIRO8pD2UJ+OjqC7Y8pvruixJ0wz71BAvueHpFP6b0DcovzV50jc56TUx45r+TyVMzEivb5ZwXdSjbx65OSvRjzqLlUpvfKkn3C4eLHm4SKNfKpKnnD0vySQNktJf5bRF58GKx+AG6ZIvEPlqkfguXHbs//rXv64XEM81LE64+N5++2239957xxoQT7nEfHGPEqclFq0w0dJtoR2QJIkL0tfiOiY+57xzR7lFixb6e+vM4We4Dl9v5zbfLIgxCwhZePviF4Ln/H9+5N2WELepU651/YPg6mrqGFdbk1AOVs8kYRXXc8d9StuKFV2vfm9H5YdAhNNAIvR7UYiCLpc0mnRJ2bxjRTSpEaLCNf6JZhPrEHWJQYF9tSXcZj2O6WNpn253OC+65sJL2hHGV5cnaZph//m38iloqe4guQm02tywYgni4eBGF+ZNeh4CIWTy8HATaAIn5em69EMi16M+69V5StVPyi1mr+vJd+OH2xouO6pdnNOkM5ynEX6DX1T/lNo2BgARsXoJCWPPdYLmmYYBkTgqSBfbI4884r/0xNrFb70nrfyW9JIf4sbG7zCp0uSKe//wwCoHqcJKEDUIM4DVgxijC7qxmVQHgXr0a76WxPncEWBfrOh/IGU8yJU36p1IWk0WpDwZQ7ieK1+4PsnLmKPfs2IIkPFLxitN+KinWkI9ooOML+xFX9onbZRz6EIa/c6J0k+nL+d6VJ5GOJcai5e+0Yml0oMOxwS7awl3ODd7+EHQljDJq+vhnH7oJE3UXucrR7+oMqPO6XbJjR/GQkgX+XV6XV6x7dJ5yjkWa0o5eauRhxeMvFziLF8TInEdMsP9iy++6L8gxB0ocVoQnn333dffj5AQ7kvZS5pSAuKl/6PaA7nBJcqmRc4Z+dGo2HG1EIjrudP/8BSja4eXj40AADlaSURBVK73XzF5q5UGEkNclLaI6bqwNDGGCBHT16pxrP8RFSuXJoa1IoDVaFtSy0wN8aoUQB7A8ENYjQG4Uj0bMX+pL8soDHhxC8EodS8vE8rlHiDol5daHP2PLlifsExJnBYB7fL14V577eWXG4JYSUA8MV9ihYJ06RitUgPio7AKn5NgeYmNkb2cD6e334aAIJDU5070K7Qv5R/M8PggZet/qqU82ZMm13skXJ7+xx/yxT/+4lYkhIVNp4mLrEo7cu0hXlIvOtMe/c7UxEzSURbnRf9c+yjjRi49mul8aoiXvtEl4DtXZ3Nep6dDo/574MbWDxXp9I3F7/B1zkWJrq8c/aLKjDqn9bMbPwqhwud4udD3UfdE4dwtU+iYLebDIkAea5VYrr7+9a/7DJAzzvGlGfFOYbKlp3kgTgsiRx7K10SzZe3F/4L8so0Oll6R4+JzW0pDoHIE4nzuitVGvy/DRChcBlae8Pue39r6I+95cb1RBuVqoiLlas8DepCH8vTzLKQNjwwbYSxaZ7kuZVZrr61v6C31orNuqz4mTSFMC+kreBZK12jXU0O8wh1eSkdwI8mDIQ8A+eVFoMviur7xww8iabFcyMPDAI5Uop8voMg/4XoqvfGLrLbsZFh6xMJSdiEJzaitXbgXmbIBaxdWK4iVkC9mvGduLPqqU6dO2SkewhOXQrTY5N5iH5dIPBfEi/4oJUA5Lh2sHEOg1gjogT3qXR7WBxefpGPPby1i/cH9pseJ8GSo4d/irkMfnY9//qQ+6mGc0u90nVbrke9Y58+XTl+TdnFOE0bRW9Iy/gim1IP3QEia5NXjoy6L67qt/NbjGb+bRVJDvPSNwc0qhIeOkgdEBiwd78XNoS0b/FcR/gJSSJl0ur7ZqEf/x0NZ+sYWvWRPGaXoJ3UWu6/0xi+2nnzpdPvzpeMa7qxGHOSFFLHHOoWVSlyNEC823I0yQzzxXvfff79r166dTwtRqybR0v0SjufKFfel81TrmHuBuL8rr7wqmIz2Ij9lBNNGhLcRZ410V141ya/NF45Pq5ZujVZuIz53/NPAPzTFih7Yw4N+VBmQCMYPnmv2mlQwLkh5ECLGEhHK1vOWacJBWj3maOsSY4/UR52a6JFPjytSV6E94w9laR0K5aGeKJIXVb9uN/jo1U8gnDI+QNB0W9FB4wmWUXUW0rURrqfmq0Y6EBIkDw83F1uU6BuDNHIj0MmUIze0EC5uFv2lIvn1TasfBl2ffhDL1U+XV+wx+qEzIjd+VN6oGz8qXannBHv89+EHK6qsOAYAjXVUHeWcq+ShZwDYNXDd4QrEtc2LDiIVFs6zPffcc8GSNse71157ze0a5KuVoCdWx3A8F7+FkFVbH+p58MGH3Ow5c91dd8513z32uGCw2cN9dbvt3Kl9+0ZC8cILLwTLJn3oZt12u+vdu7c7/PBu7ohgcDj0kK7B8eGReezkfxAAI6yblUrSnjveJeF7OV8b0V/GCd6VjAW5nntIBtc1OZCyIQnheCXew4xHeqyQ9LKnLkJPdJ3kY+yJqkfysWeOLJ1PXwsfo1+h8sJ5wr9lDJPzlMkWFtKBk+Aavs5vxh7aHRYZizkfRerC6Rv2dzBopEYCcpQJbgQmN8m5Bf9ZZNsTfDnSIl2x1yggeMha5A3XiR7BjMPZujgoVT/yBDdoth6tX6FrpA3rpH9TLvpo0XUFD4W+5I91mcHD1eJ6FO5gVEgWLVqUOeywwwolS931YA28zPnnn+/1DqaTyOTaSBBMlOo3jsGjVkJdwYsub3XoFkx7kTdNuReDhb8zPzjpFH+f/nT4zzO33nZ75o3Va8oq7p5778+MOu/8TEDA/HbDDTcUbFtZFTVIJvo0mGuuQVrzn2ZMnDgx069fv/+cKOJIv7sCMtMih37nBUTAv9N59+l3afi93KKA4AdlhvMEhClDvvAYofNyXesmdXKesSss+v0d1on0Aclsobe8n8NjWbhc+U07RAf24TokneypMyCRLfKgY758ur1RY5CU3eh7/ltPndCxugO5Sbjxwx2p03BDhEU/LDwoYaLCjaXTUE+hG4s6itWPtPkepnzXyFvqja/LC2NFeegtDx7t1pLvwdbpwscM7IFrIHw69b8hkxCLYiRMtsK/iymjlDSQrVLqKDV9IV2oWwjSFVddXTbZylUPBA5C126vdpkrrrgyV7KmP8+zXIh4pw0kSBfkqxTRxCP8XtPvPIiXSfUQYAyR8YWxqJkllcSrmTssjW3nP+9qWVXqhUexg1oUAdIWsLj1r8SCha6VDNTUHSwD9DkhCghXtQUrGAQMkheFc7XrT3r5pfxzkPS2iH68S0rta6xO/GPNM8teW6GMeAmy1d9r65hY46pfazJrSE1wffDQmKQUAeJNCKhuBCFuRmKiiJ3KJ8Q2SVqdjnPEqsQR+6bLJZ4LKSUGRuennyTuS58v5nj+/AXuqCOPCqbT2MwtDPp62JDTi8lWUZqjj+zpWCi79wnfc4MHDXYXXTymovIaLTP9OXfu3IZpFvcmzwztKkUCspUNhA/+scgbk1VKuZa2eAQ07oG1q6jY4OJLT1/KDdOnsmmcNgR4UQYxOWlTO1Lfyy+/3E8NwUWC5mkbZExIj86Ui3iRBnIUlUfnL+WYsiB0bJWIkLZSdLv44rFu6JAh7pcB8Zl42QTXpvUOlahQcl5I3u1B4P7vf/+4Y/HtuAltyQolIAMY8I/B9OnTGwYPSCRz4JUjBLQHoSc+a75g+HLKtjyFEQBzyBcSWBkLZ2j0FMk0xJlWjYQA7ivivHBFpVlwlwbvg5xbMHFq5thjj81cdtllmeuvvz4bcJ+rzeAShwsW1wtlxSmUV4xLJ5j2IRN8pZh5dMmyOKsvuyyC+HE9xoFr2UrUKSNtps/YpP245oqNRayT2kVXW2lbdIwRLkbEXI1Fw192Qu3qxd1okslsAAiNTi6tffVHoH///l6JNFu+sCLwHzeL9AaDQNbylQvdHXfc0VvEmEaiW7du2YWqsZSJYBVDyrFUoQ+WKaxu1RJcxFjBotyqZ519jluxYoWbPPnamlu58rV3+Jkj3Ly773Izb51Ztts1X/lJuRZ2C0f1ExZaLEVpd/WjP8+eWTOTcveZHpUgYMSrEvQsb9EIMEgwMLCPGsSLLqjOCSFIuBYhkoEFz82ePdstXLjQbx9//HFe7Tp06OAJmBAxEkPCGFRKJU/gyCAkrsG8FVd4EXJHn2lymFTSJU0V8rXssWWpvt+kPbLXBIr+0H0iafSee4Q04orW19J0zPPBxrNnYgikHQEjXmnvwRTpD1lhEEjryxNrHbpDejAUs/3zn//02z/+8Q/3+OOPu5/85Cf+NxOAFpJDDjnETw6KNYyFsxlYtDUsV/4oIpQrbVznNdFjRvk5AeG8+ZZbEmXpCrcV8rVq1YtuxvRpqSVf9LW28nCPlCrcsxA2TdpKLaOe6dEbaxf3YJr/aasnhlZ3shAw4pWs/mhobXhx7rbbbt5SBAFLk4h1CdcNgwCkK5g01X322WcO0vXJJ5+4lStXOqxeuBi5FsTWuMWLF7uHH37Y/f3vfy/Y3J122skTu+7du3uCSoYwEWMQinIpFSw8hgRgQPtvvvkWN33Gja7rwV1iKLW6RRBsf+CBB7jzzh1V3YpiKh2MIVsicfQ1ZfK84XIsh7iJLvXaozMbBNLEEGgEBIx4NUIvpqgNP/3pT/3Akrb/vsN6i7VLSNdHH33k5s2b51iTERIGIYN8QZxYSujDDz90v/vd79yjjz7qHnvssYI91qZNm/XckgzIWMfqJQzgB3U5yI0YOcr9aED0Uj/10i1Xvc8+t8Kd0Ps4N278OE+Yc6Wr53n9LGDRqYb7GMLMJtbSera3lLpFb/5pMzEEGgUBI16N0pMpaQeDNwMLRIYtDSKuDgYtLAeIEK9PP/3UQbruuecev1js+++/739DvnBDIhAvFtJmYWzZL1++3JOwRx55xAeo+4QF/gwJpmxg3UIhX2FrWIHsFV8mrov+mzrl2orLqmUB102b4W6acUNggZydCFcVJEITiVpZoaiHZw8ykwbheUPntFrq0oCx6VgfBIx41Qf3pq6VF+q+++7rgk/eq/LffZzgipuGwYoYNRFNvJ5//nlv0dpmm23cunXrvFsRMoY1TKxeLKYN6YoiYZzHEgYJg8BhHSskxxxzjCdgkDCwRKpJxOiz7//39/18WR07tC+kXuKun3Tyqe6gg7q4YUOH1EU3bdWCvAuBr6UykD0hXvperqUOxdbFcwfpwq0/2lyMxcJm6VKCgBGvlHRUo6lJoDoWLwakarhW4sBLXv7oh75aNPG67777gjiiAx3Wrg8++MBvEC+sYZAv0rJBjNggYRAwTcI4FosYgfmQsGnTpvkydL1Rx5tuuqmTLyXzxYdF5S32HMRl7077uFEjRxSbJVHp5t033w0fNtTV6itHiCr3jwgkIgmC9QjSlXQrkkwdoQlrEvAzHQyBOBAw4hUHilZGWQgwADBA8XJN2tdKQrrQK+rlD5HCmrVgwQLXtWtX716EbEG82LPhbhTyRVrZNFiQsDARo4ybb77Z/fznP/dk7PXXX89axIqND8MlCQnDIibYlmsRE2sXSwHVelZ6jVWlx9W0enG/gJMIZF1wl3NJ2WO9HT58eGItzkl+LySlD02PdCNgxCvd/Zd67eUli0UpKZYvIV2Am4sUQryYx4s4Lr5ihGBBtPiqkY1jTbwItpdNpqCAiFGOlldffdVbzlje5LnnnvNuRIkLk32p8WEdO3bMxoaVEx9GbNcXN/6Su/jC0VrV1B2L1WvFyhWx6K4JOSQrKfdvvsbhbhSSmESLs7wPcj13+dpm1wyBtCBgxCstPdXAevKyxfXBy7begxe6sL5dr169vHsxn9UiWJrFffvb3/bkS6aVEAuX3ss12UO8cEHyW0gY+yeffNJbSZgVH+sUU1C88847bo899ljPNSkkTMeH3X333UVNWyHxYVjEBO9c1jAGab5kZC3ENMZ2hR+bbt26uzPOGFbWF46QFjaRpLgPRZ9Ce9Fd4sv4ZwfyFY5fLFRONa4X889ONeq1Mg2BeiBgxKseqFud6yHAIDBgwAA3ceLEun3tSFzJHXfc4XUjvgoSlksgiYcddpi/LJYrIVEQKr1BsjTZEtKl98E6co55vDbffHOfVtySDJbbbrutPx8VHwbx0iSMwHziwwjWv//++3Opnz1fKD4MQnz9tOnuzjvmZPOk+eDKSde4VwJMf3XpJQWbIZYhSQhhEdIi59KyD5Mu0VtivrjX6/W1I88S9UNk0SHfPzuit+0NgTQjYMQrzb3XYLoTIwP5YXCDiNVqkGOA5cUvpEtg7devn9eD39olKIMYk8Hq8xwLCWMvREz2YTLGb8gXcWK4AyFdQsZ02qefftqx3BBlaomKD9MkjGD9SuPDxowZ51pts21qg+o1XhzLvF653I3cg9wPSFrch17ZPH/kfs31PHGd5w6B+NTKkgfOfLHIs84+LdPL5IHaLhkCRSFgxKsomCxRrRDgZczL/4ILLnAQH17IuQaMSnWSumSw4cUPAXvllVeyRe+zzz4OlyKDsJAs/kMnVkq75+RY0lCAJmEcazIGscKNiKWLpYOEhEXtOYcbEnImJE7Kpj6pmy8jJVAfAqa/lJQvJkuJD9tkk01clwMPcmPGjUvFLPXZTitwgLtx4sTLvJuVeyAtQfEFmhV5uRDp0pl41ngWIGHVfO6oE7LF84arm+NqPeO6fXZsCCQFASNeSekJ06MFAgwYvPyJt4KAyUu6RaIyf1A2L3sGGV781CP/5TMQc/ynP/0pWzqzyLMYNiRMky5Ijrj/2AsBymYMDoSIsZcN8vTSSy+5tWvXuk6dOmXdkpzXFi85Zv/aa6/5dLgd+U1aCBl7IXS6XtEL8iWbkC9tFZP5w6Liw2gv1jZpgy4/zccDB53m/rzyed/vjWLViuqPUkiX5Of+51mT545/RHge4hD5R4dnDxGS53/YH0OgiRAw4tVEnZ3GpgoBIxaF/4r5b5yBoNTBAKsGpImXPqQKMpdvUOH6jBkzspBBZIYOHer69u3r15sUMiOWJX7nIl/ZQoIDSAy6bLXVVu5rX/ua/y3ESfZCqMLWL6xV2223nSMuS5MyIWGSj3LYtAhJ1HpDxPgthCwcH/aNb3zD7blnO3fbbbfqolJ/PGbcBPfZp5+4//3f81LfllwNKId06bLknxMhSTx38uzpdIWOKYfnjucXVz4frUDsSn1+C9Vj1w2BNCFgxCtNvdXkuvLyZuNFjjuQ4HbIGBuC9YJNBh1xI4kriZe9DCCkyycQl+uuu84NHDiwRTLyX3HFFVnCsvHGGzs2IWBCcFpkUj/QHSsb9WtLEsfUKXvIlN6EhGGhYqoJ+R2155xsQsKiiJi4JSFf6K8tYZCxSZMmua232c5NvGyCakH6D5lW4saAVN9y843pb0xEC+T+l+ciIklJp+SZ497lnxYIOWXLF7EUxrE8Z7meO56/uHQqqQGW2BBIGAJGvBLWIaZOcQjolzvHCAMOx3pAkJd9KS98yA8bVqXf//73fsoIrVXbtm3djTfe6K1PX/rSl7wFir1YjiA0iHY9ir7ok0uoU0STMCFPELF3333Xzx/Wvn17T8y05SsXCRPXpBA5KZv6RMcwCYOMzZhxk+uwd6eGCawXbBuZePEMIKXc7z5DkX/kPoZk5XvueAbRQT+LRVZhyQyBhkfAiFfDd7E1sFQEICSQE5kUlfUTTz755PWK+fWvf+0OPvhgt9lmm7kvf/nLjmB0rF/iziMDxIbBMEwI1yss4oQQMfayQZ4Y9HbeeWf/FaRYtjgvJCyKgMk1IWGk0USM6qlD3KUQsZkzZ7lv7rd/wxGv1WvedDu2ae3bGwF7ak9Vm3SlFhhT3BBIGAKf/2ueMKVMHUOg3ggI6YGAQVBw8bVr166FWj/+8Y8dMTB/+9vf/GzzTHjKrPVCbiiDhcCRcv7z1yRIyBzE7oADDnAszE2sF6SPaSi22GILt+WWW/rYMdyYrVq18u7MqD3xZWzkIS+kUSx2EC70ps20vRElzcse5eoPcfNVy9KVq147bwgYAqUjsFHpWSyHIdD4CAjpWbx4sZ86ggWwIVkXXnihwwImMmHCBLdq1Sp33nnneaKi3XgSjwX5iUPQCWHPrPPE3OC6lDplL5Ys2Yu1qxhLmKQlr9QXh+5WRvUQgHRBuArFLVZPAyvZEDAESkHAiFcpaFnapkEA0kEAP/FcEnjO/mc/+5lr3bq1D7wXMJhqgnm2cD3iAiQOa+XKle6oo47ycV8Qoqi4L8lfzh79mMAVHXcNBl2sVFjFECFg7NmEgLEX16TeC9kK7zf64hfLUS3xeZhEtd1eLa2XiVc6h4JGunIAY6cNgQQjYMQrwZ1jqtUHAbH0sGA1k5t+9NFH3hUn0zj06dPHEyzm/xKBAPXs2dONHz/eL/1z6KGH+nwQHx33BUGS8iVvuXsIFwMv8WPa2gEBEyLGXjaIVxQR43yYdPF7y8AV2YjyajAn2n6dD0h904x0pb4LrQFNioARrybt+DQ3W76swtUmx1HtgZiwEV8lX1lFpYs6R9m487AMQZyEtEBcIDIE1eN67B9MMKnl7LPPdmPGjPETo0oe0lMGErfli3ahKy5HLULu2FM/IoRM2hAmYeirSdjGG2/k/vLyS7rYhjjGbZx2MdKV9h40/ZsZASNezdz7KWo7X2wxnxBkR+YSEjKlLU+6SQxO5GOG7Iceesjtsssufh4vyBJ5cwl5IGyQJMiKkCZIDJuc5xqTQp5zzjnuueeeyxY3atQoH/c1ZMiQrJsPkkM5MmkpiYUcZTOWeUBbaGuuNul6pA1SlSZhEDQhX+yZJ21asEB2o8mLL65y7UMfSqSpjUa60tRbpqshsD4CNp3E+pjYmQQhANFiY7CRyU+x7mjXWrHqykSQlEd+CBtlhssSC5JYioSM4H775JNPHF8vMsv7Bx984L9mZLmdZcuW+S8ftS6Qt9/+9rf+60G+PsRVKV8PQtritH5BFhHqLFWkneTTRIwljYhn09dLLTuJ6VkyqOvBXVz/kLUyibqGdTLSFUbEfhsC6UPAiFf6+qwpNIYksbQIkosgVQKEJnRYxGQQFtIlZQvpELccrkfIF3Ffb7zxhnvkkUf8TPIQsaVLl/qvHiWv7O+77z4fEybzfUG+sH5BvtgQbZWSfKXuw7qXml/SS5vZH3FED3fWyHPc0Uf2lMup37dv197NvHVmTgthUhtopCupPWN6GQKlIWDEqzS8LHWVEcByAwlikNGEqFrVQlaoD6uXrCEXZTWChLBh/YJ88dXim2++GaxluKe3fGH9YluzZo0bMGDAeupec8017lvf+lZ23iyZbJUvJbF8hV2A6xVQ5Im4yJdUN+KskW7jL23iLr5wtJxK9X7J0sfcD/v3cytWrkhVO4x0paq7TFlDIC8CNoFqXnjsYi0RwApFnBKbELBq14/bkrpwOUKY0CFKhBhhoXr22Wf9LPVdu3b1bkSZkJQJTHfccUcfi8akpFpOP/10v8ZjvslWxdKk85V6DGmkPXEI5fzj04/dQ4seiKO4RJRx9z3z3LHH9U6ELsUqAZmmX8Mu8WLzWzpDwBBIFgJm8UpWfzStNlidcC+yQYbqIVgVxPqFHlED3aJFizwxZNZ3rF+yrJDEffHFHJYvfl977bUtJlulTfvuu6+f74v8Ou4Ly5fEfVXqdizXOkI+vhIVYbBnwzV3/Q3TfVyUXEvrvlu37u6MM4Z5op2GNsRtwUxDm01HQ6DRETDi1eg9nPD2MdBjbWIP2WGgr6egB+QLaw+DnpAvzkNMIIVimZK4Lwm6J+6LWC8hXxL3ddFFF63XpPnz5/v5vnTcl3zxGEfcVzEDdphoYWmU9mqFL754rHvn3bVu4mUT9OnUHc/67Wx37dWT3KJFC1OhezF9mIqGmJKGgCHQAgEjXi3gsB+1RAAyI9YtTXJqqUOuuiBfEBP0Qk+28HQNOu4L8oX1S8iXfPEI+SLu64c//OF6VU2ePNkx0aqslyhxXxCvSskX+kIetc60RUsuoqXTcEw5zJK//NnnXccO7cOXU/P7pJNPdQd1OdANGzY08TrTV/JsJF5ZU9AQMARKQsCIV0lwWeI4EcDSxaDOIBNlaYmzrnLKgnxNnz7dL3StCYwuS8gX1i+C7iFfLJQN4RLyJUH3p512midmOv+IESPcKaec4skX1i8hX1i/Kg26l3g1sSJWMpATZP/ZZ/9MrdVr3n3z3fCAcC17bFki7zV9Txjp0mjYsSHQeAgY8Wq8Pk1Fi/iCkAGGLYmkCxBxffJlJYKeuURcjzLfl5CvqLivcePGuSVLlrQoihnyL730Uk++sH4x35dMtgr5EgLWIlPoBxYuLHRaIFroXQnhkvLSbvU6tldv1+OI7om3dsXVX9JvtjcEDIHkIWDEK3l90vAaQWjElSfWmKQ2GkIDccE6x3xiuUTIl477wvIlrkcd93Xvvfe6SZMmtSiKWfVZZHunnXbyBAzyhfVLFuiGfCESeB8mWpDXXFa5uAbzK6+aFEwU+5i75eYbW+ie9B9XTrrGzbn9t4mP7Yqrn5LeH6afIdDsCBjxavY7oMbthzBAtnCDQWbSIBJUD2GEhOUTCBjki03HfeFu1Nvzzz/vfvazn61X1MyZM/06jxJ0L65HiNszzzzj00O+8hGtcKFgjsUqFzELp8/1m3J69z7e9T7he27YkNNzJUvUeZm3a/KUyQX7rp6KG+mqJ/pWtyFQWwSMeNUW76avTcgWJCZNgsuRDQJTSHTcl5AvHfclBIygeyx/Ybn44os9+XrnnXcc84Fh/dpuu+1c586ds25HsXyF8+b6DXmE8Fbq1qUcpsR4dMmyxE8vsXrNm27w4NNcjx5HuGFDh+SCpu7njXTVvQtMAUOgpggY8aop3M1dGQOMBNRXSgDqgSQWI4iSLGWUTwdxPUbFfQnxYk8QfniRbcrdb7/93HXXXZcNutdxX3zxCPEqlXzFNcDPnj3HjQoWBk/63F6syQjGSXaNxtUn+e5Fu2YIGALJQsCIV7L6o6G1wU3Hli9WKskAlEocw+RLz/fFGo+vv/561v0Ytcg2cV+33357lnxh/apkke24XI700Vlnn+NWrFjhJk++1rVpvUPium34mSPcqlUvuhnTp1Vs5atG4+gLcWFXo3wr0xAwBJKLgBGv5PZNQ2lWKmlJauMhjljtirF6SRsgYH/4wx/cu+++66ecYJHtdu3aOaaMwCJD/JZMtnrhhRdKtuw+zkW2xVWK27FSEfI1duzYRM3vlQbSFUfMXaX9Z/kNAUOgPghsWJ9qrdZiEWjbtm3WrTR+/PgW2cTdxH7BggUtruX7MXXq1GyZpbqr8pWb7xrxUZCVNLoYdbtog0wxoc+HjyGasj300ENu9913D2KNeriePXu6o446yrVp0ya7ziOYsM7jIYcc4qZNmxYuyh155JG+rHXr1nmixpeSkDfmDcOVKTFl62WMOAHhEvIVcbmkU5eMH+vat2/vTuh9nCOIvd5CTBeTpCbd0mWkq953itVvCNQXgaYnXpAZITCQHJP4EcCtcscdd/j4qPhLr22J8nEApEqLkCzZi1tV9q1atfL3GfFZWLr4WpEvF5m3C9IlC20znQQfHkDMtLDINoRPFtnGQkbAPnOGlUq+0Cmsv66rlGPI15jA4nVI14Mc0zbUSyB+J590ktsqwDLJ7kUjXfW6Q6xeQyA5CGyUHFVMk0ZFACLRq1cvF4d7KwkYQVzCli/OFRKxLurlgDjHb70xZ9eUKVOC+KnJ7u67784WS7A9Lkvm+5IpKySOjD1lhOf7ymYOHfChADFGlU4xQbHHH987mCNrkbvggl+6ZUuXunPPPbdmrkesXDdMv9Gde85ZfoqSfv36hVqajJ9xxtclo0WmhSFgCJSLgBGvcpGrUb5Vq1bVqKbqVVPM/FfVqz3ekqUtWIyKIVvh2iFaQpLE0ip7SJOQJ/ZYufi6Ucd9PfXUU27//fd3999/v9t5552zBEwH3ZOXOig3l4jLF0Igx7nSFnMeLCBxV199rdu749fdxWMvcf37nVrVwPvrps1wN824wbVus2PeZZ2K0b+aaYx0VRNdK9sQSB8CG6ZP5Xg0JiaKgWnkyJHZAl966SV/LsrliEtSx1uRl3PkCYt2XxLTg1CPDLCSXn6zRx825mqSskmn66TcfHLbbbdl81MGZXGuHCmlvYXKh6SIi65Q2qRfpx39gyklZDAtR1/pd4gWM9OzPJC4HrfYYgtPhHA94oLs2rWru/7669er5jvf+Y4DV4n7YnkicTviekTEGrZe5n+fEKtXruulnofAnXvuOZ4ELX/madc9IGPn/mK0e/a5FaUWlTM9Fi4IV7du3T3pGjp0qJ8uIg7LXc5KK7gg90lS9augaZbVEDAEykSgaYlXsXhBrCAwEKcwyeIc15588sm8xZGmEGmCdEHSCpWVqyIIVp8+fVrkpyzOFapblxlHe3V5uLOQXWP4ik6Xi558JDBo0CCP29Zbb50ltkJsOAempCFtuP90eaUex0FaRE8sVGHyJTFf7CXui+kktLDoNot4E/clc4IR98W0FcXGfWGpgsDFKWDD3Fkzb53pPv3kY28BY61EYsDKCcIXssXXipC5BxbMd/369fVLAOHmTKoY6Upqz5hehkB9ETBXYwH8w2QmnPy9997zgzsuQQKowwKhKkZKIUdR5UEscgkEEfcUX9UVkkrbGy4/7mBiyBPtKcaSR9+E8T/xxBMdC1XzlWElAmGBVFZqyYN8IZAvRMiYuB05zzEbywmFF9meMGGCJ9sssg3Z0rFfBPFLXqnHVxL6Aymmn+ImxxAwtnNHjfQfDEC6rrnqSl/7fp0PcHt32scf77FHW/+Fp6j1wgsvBETyQ/eXl19yL/x5ZUAMF7kfnHSKO/I7PdwZw34Su55Sb5x7I11xomllGQKNhcCGjdWc4lsDCcEVw0AmwmDMOYmrgsxoCxRpuc5G8LMIA3w+4kO58+fPz+aVfOH9wIEDs2nOPvvs8OWCv3PpR8Z8+knBcbVXymMf5ySR4kothnRpHfRxHGVQHiRFrHm6/HKOhRRBsnA9Eq/FTPV89Yi7ERceli/ckKNGjXJDhrRc/mbhwoWB662be+WVV7zrkS8emXIC1yNTTkDG5L6N0o+2QLyqJejfP3DPTp1yrVuxcoW3hPU58QS3+Wabus8+/cTNnTPH3ThjRnZ77dVX3WZf3sQvSXT++f/rdceCRuA8uiZdwJIN0mliCBgChsB6CAQv5KaWgKxkAlD8FhCkFlgE1pHstYAUtbjGj1x59XnKDkjXenk5IfWyD4hgZBp0knSUq0XOsy+kH2nWrl3rswekMVsm50XKba/kj9qff/75GbZKJSDDmcCi2EJv3f5SjymLMsuV4Cu+zGGHHVZu9pz5ApKUCSxXmYA0ZQIClQlIfWb16tWZwAqUCb5ozCxevDgzb968zGWXXRaJRfAlZGb58uWZYODPvP3225kgBiwTuB8zgfsxQ9lsuYQ2mVSGwMsvv5xhMzEEDAFDIBcCTWvxCgbqgqKtXVFuOtxWIrjAsHxFSVTecLpi0oTz6N9aFzkfLrNQjFNc7ZX62WMVisNKgTUujC+WRCyDASH1FkWsiuGNa6TB1aqlkJVSp63lsbgaJe4L6xexXVi7JO4LK1jHjh399AnhuK/Bgwf7OdNkvi+C7on7KmayVSw0cVnxaolZUurCyoXEcb/7guyPIWAINCQCFuOVp1s1USH2qZAwmIfjvCAHxUg4XzF5dJqoesJlhomLzs9xHO0Nl0msSxwDUThWC1cvrtlCosmnfMAgecJlyvl677XrEV0kTos9hExvxH0R8/bcc89l1WYeLUj0L37xixYxXwTwE/dFfkTqkYwyrQR9Jsdyzfb5ETDSlR8fu2oIGAL/QcAsXv/Bwo4SjIC2xkG4iiFd4eZAwnQ+XWY4bb1/CymCJMmUE8R9MdO9WL4k7uuSSy5xp5xySguVZ8+e7QP/Je6Lrx5lpvt8cV9m9WoBY1E/jHQVBZMlMgQMgX8jYMQrz62grUg6OD7w22aDlfWxTp+n2KpciiIRYQtXIf309bjai+VEBqa4Gq71LLXMSvKWWlel6cXtiKVLky8JuhcChuvxpGC5HCxcWiBdkE1NvsLrPJKee1jL4YfHP8WELr+RjuXejsOq20i4WFsMAUMgNwLmasyNjY8LEvcbxEa7rfJkq8sl3GbhOC/tSsPtWIh0EAcVd3uxoMjgFBcwtKucrz6pX2MSlz7VLgcCBvkK77XrkeNDDz3UL7I9YMCAFiqxyPY111zjvvWtb2WnnIBs4XpEyIuIlY1jiAT9xj5uIY6Mbd37H7jXXnvdvfHGGy2qIJ6tfft2Lpjj338ZCBFMosh9XQ2Mkthe08kQMATiQcAsXgrHsIVIEy3iaPRcWxAU4r7EKhE1270quuqHBJ9r/fiNziJhUibn9b5a7SVmqFLRukGeaFu4v/LVQVrw0cRLl5kvb9Q1iEMt46DkPsP1SJwWQfeyyDaWL3TB8iWTreZaZJuZ7t9//31XaJFtyAT9FkffUcYNN9zgBg46zbVv194NH36mu3/+A+6DDz9yrbbexp3at2+L7cAuB7mPPv7UvfLqG+6yiVf4Z8xPwHrVJE8Go/qj1ueMdNUacavPEGgcBMzipfqSwZkBDssQc3kRD8RgLVYgBntNZlTWsi0wuoxKjyvVrxrtxeJ1+eWXV9o0b23UpIl+YcNKhzUvF4kiD/0a5YrNlacYZSETtK2Wwr2JpUoHx3OO39r6xe9ci2w/8MAD7vbbb89avpjjC5FytfWL9j0YzGpfrsWJvHffc6+7dMJ4PwHqQQcf7M4444ySF9Bm5vpHHl3ili5Z6o468ii3V/uvu+N793L9g7nB6iFGuuqButVpCDQQAsELt6ll1qxZ682HFBCvLCbM9RQM7uulCW6B7Lnw/Fr8luu6rGyh/z6QNOyZWytKyC/pwvXIefaB6y2bTp/nOJwv1zxe1F9Oe6P0lnPMaRRYZORn2XvmICvUD+F25/tNWTKvWTlKMYfXnDlzyslacR6ZjysIks988sknfr6vd999N/P6669nVq5cmQlIZiYgPZm77747E8R9Rd4XwSLbmeeffz7z6quvZt55551MYAWLnO8rIK2ZYGHuknRmPrBgpvlMu73aZYLFsjPLn32+pPz5Er+xek3m19dPzxx+eDe/3X777HzJY79m83TFDqkVaAg0HQJN72rEBRcQk/WmgQgGbS9Yv5544gmfBuuKFixEBKGXG2+ky6r0GF2wcqCvCPoGxLIk/eJuLy4rpNL5obBq0b5wH/jCS/xDGZQVnm6jlGIeeuihmlu8RD9xO2KdIuge16NeZFu7HotZZBvXoyyyHZ7vCxdmsR9IYAW86OIxbvCgwX45oIWBxWvUyBElW7iknVH7Nq13cD8a8Pk6jaf07e+uuuoqhxuy0vsrqq7wOalD7unwdfttCBgChkAxCGwA1SwmoaUxBMpFgPUMcVf99Kc/LbeIFvlwMRLDJi7gFhfz/IBUQlArJcpz5871bRGXU54qq36Jx5cNlyGkiWWCmDaCGC42SBVTSUCs+PKRvZYf//jHbujQoX6aCiZjhcARPwahw2UpJK+Qy5HrF1zwS9e6zY6OecQ6dmivq6nq8ZhxE9y555zlrrjiSjds2NCq1AXpgnDVMq6vKg2xQg0BQ6DuCBjxqnsXNL4CBFYT5yUWg7haDPHSMVzEcmnBooV1S2LAtDVQpyv1WAhkHLFrpdYdlV7+d2KRbDbIV+CC9CQL0sUmVq1gSSF3zz33tCime/fujkW2mSOMgH3mC9PkC8saBCwX+Zo+fbobO2asO33oMDdsyOktyq7VDxbgxnLdunVrN37cmFgJkpGuWvWi1WMINAcCRryao5/r2kpcUJCfID7GWw3qqkwMlWP1gITUOrg+n+pCvrB8Qb6CtRmz5AvLl5AvjpctW+YuuuiiFsXhnvztb3/rv4qEgAn5wo2J9QvyhYUPArarmmLirLPPcXfOneOuv2G6X9S6RaE1/kEQ/ujRF7g333zTzZg+LRbyZaSrxp1o1RkCTYBA08d4NUEf172JEJV+/frF8nVjvRuD9Y72JIl0gYm4BCXuizm6cBtCophmQsd9HXLIIS5YZLsFlKzt2LNnT0fsGscQNSZbxXomcV8QLqyKEGkE0rVixQpHLFfXg7u0KK8eP4j/mjrlWte27R6ub78BWT3L1cVIV7nIWT5DwBDIh4BZvPKhY9diQwALEbFeWE3SHCcDwTn//PMDy8ro2LCJu6Bw3BduR4n7Etcj+zVr1rjTTjvNEyytw4gRI/wSROJ6hMDJOo8QO8jZvHvv96Rr8uRrHYQnaTL8zBHBlDAvlm35MtKVtB41fQyBxkHAiFfj9GXiW0KAPVuSSUs+ELF2oTskEgKphXYlScT1qOO+IF8E12vyxW9io5YsWdJC/eOPP96dd9553mImrkchXzfddFMQRzXe3T5nbk2D6FsoWMQPJmylrbfcfGMRqf+TxEjXf7CwI0PAEIgfASNe8WNqJeZAQKxeMrDlSJbY07jaIF5RE3fSNi1JcEcK+ZIvHnEZQr5wIWryRdzXzJkzHYRKyy677OJ+/etfu5133jkbdI9rkaWJHl2yLBHuRa1v+JiYr8GDT3NdDjww+NLynPDlyN9yb6bZKhvZMDtpCBgCiUHAiFdiuqI5FIG0ECPElAxpEggXOjMwFyO0MZyWuLB6DOgQMMgXmwTdQ74k6F5I2NKlS92FF164XvMgZZ06dfKxXqef/hPX5/s/qNvXi+spV+AEXzv+sH8/N3nKZG9tzZfcSFc+dOyaIWAIxIWAEa+4kLRyikIAQoLliKkYoixHRRVS40QMyPvuu68LZnCvKKieciQwXZpQKxeljvuCfBE0D/kKux5Xr17twotsoyuLbC9b9nv3j8BqVqrrTtpar/2Vk65xc27/rVu0aGFOFbBY1osY51TKLhgChkBDImDEqyG7NdmNEpejDHZJ1haixIDM3F0yf1ec+oZdlJBStmqIuB513BfkS1yPerJVgu4hYWEJlv9JdFxXWF/5zez2PY7oHjnBKn1QKwIs+tjeEDAEmhcBI17N2/d1bTmuO4LVsQLVw/1WbOMhXWzoWgshaD8cuB+nJSZMvsT1iOUL16OQL46ZbDVY79E3+8ADurgewQLVF184uhYwxF7HvPvmu+HBrPbLHlvW4n4z0hU71FagIWAIFEDAiFcBgOxy9RDA1QjxYvBLIvlKin5hFyVYQcbKFSFfMtkqQfcy0z0ETJMvHfcVLFCdyKkjisXhpJNPdQd1OTBr9TLSVSxyls4QMATiRMCIV5xoWlklI5AUcqMVx72IWzGppBD90E1LOS5KifuSme7DcV8QMCxf//uL813XQ7/lJl42QVeZumOsXpeMG+tjvYx0pa77TGFDoGEQMOLVMF2Z3oZAvhgI+WqwEktOHAhAaiTeB52SaImLameUi1LaEZVezgn5kiknIF96vi/I13/3+W83c9ZtiZ8+QtqUb9+tW3f3jW/s0xCrKORrp10zBAyB5CKwUXJVM82aBQHip4j5gijU82tHiBaz67OhR1pIF/dJlMWL9miJclEyEz/yhS98we/10kPMUn/XXXe5vTvt0xCkiwZ2PfTb7h+ffuLban8MAUPAEKgHAmbxqgfqVmckAkJ8hIBBJmohWLkk2J99Nb5erEU7CtUR5aKUwP1w3JcE3Q8/8+dup52/ltqg+jAmMq/XipUrwpfstyFgCBgCNUHALF41gdkqKQYBCBcuM4gPhIA9WzUtT2Jtg+QRN1UrslcMHnGnAUcw1hIO3IeAHXbYYdlFt1e9+IL7/g9+oLOk+lgW86bd9XZrpxpIU94QMATKRsAsXmVDZxmriQDWL6xPDJCQL+LA4iJFWH4gXLgTEfa4F00+R2DRokUOArZ27Vp34okn+uNGwoY1HDt8vZ2/rxqpXdYWQ8AQSAcCG6ZDTdOy2RDAMgP5IuAeK9huu+2Wjb3id6kiZAuC1apVK18uhIuyjHS1RLNbt26ObZtttgmsXSe3vNgAv3bdbXe3bt0HDdASa4IhYAikEQFzNaax15pIZwgYGyQJEsaGJQy3GRYwriGy9z+CP+JCY8/2yiuveBeaBM7HZT2T+hptT5A9uG2xxRaN1jS3xx5t3dw5cxquXdYgQ8AQSAcCRrzS0U9NryVEC3cjGyKECouVBMf7C//+A7Fig2jhqgwTM53WjqMR+MJGX3RYhxpNGpFMNlofWXsMgUZGwIhXI/duA7eNwGgLjq5uBzOxaiPK13be2f3hiccbsWnWJkPAEEgBAhumQEdT0RAwBAyB2BDo2KG9W/nnlbGVZwUZAoaAIVAKAka8SkHL0hoChoAhYAgYAoaAIVABAka8KgDPshoChkD6EHj2OZs8NX29ZhobAo2DgBGvxulLa4khECsCsoxQrIUmoLBXX3vN/eCkUxKgialgCBgCzYiAEa9m7HVrsyFQBAL/+udn7q9vv11ESktiCBgChoAhUCwCRryKRcrSGQJNhgBfjb711psN1+qnnvqja9+uXcO1yxpkCBgC6UDAiFc6+sm0NARqjgDE6ze33FTzeqtd4V9efsltueXm1a7GyjcEDAFDIBIBI16RsNhJQ8AQ+HxR7W5uydLHGgqMF4KpJGxC3YbqUmuMIZAqBIx4paq7TFlDoLYIHHBgF/fgQ4trW+n/397d9ER1hmEcv+YDsNRUUDMJLEhdYayg3dD6/sZAurOFIU2qSNGkTaoZSXShKGqiCRo7VYkVW0QDTtqmMbY2gcS2GKDMoo3ESGUjfgLWyjNqAqiYyDkz5z7nPyvmDPOc+/nds7hyXp7j495ciHwyOcniuz4aMzQCCMwvQPCa34dPEYi0wNo1lRr8+6/QGAyPjGhHojY082EiCCBgT4DgZa9nVIxA3gTcsy4fjN1XWNa+yvT16sO1VXnzY0cIIIDAXAGC11wR3iOAwCyBmto6XbrUOWubxTe3bv+eO83owiQvBBBAoFACBK9CybNfBIwINO/ZrVu//qLJJ7aXlrja1aXmlhYj6pSJAAJhFSB4hbWzzAsBjwTi8bhWrvpA31+56tGI+R/GHe36Z3hIDfWsWJ9/ffaIAAIzBWJPp18zN/A3AgggMFcgm82qoqJC//53XyveL5/7ceDf7/y0XlVVldq3lyNegW8WBSIQcgGOeIW8wUwPAS8E3GKqR462qa2tzYvh8jpGx7nz09d2PeZoV17V2RkCCLxJgOD1Jhm2I4DALIGWL5tzp+tckLHycndjnj/bocOHD8ktCMsLAQQQKLQAwavQHWD/CBgRcMEl/V06F2SsrGafSqX0WUMDK9Ub+Y1RJgJREOAaryh0mTki4KFAR8dZZTIZ/djdreIl73k4srdDffX1Nxoff6iff8p4OzCjIYAAAgsQIHgtAI+vIhBVgf0HUhobG1M6/W0gw5cLXdnRkemAeJNTjFH9kTJvBAIqwKnGgDaGshAIssDJE8dVXl6upqY9gVvfy4Uut+7YmTOnCV1B/hFRGwIRFSB4RbTxTBuBhQrMDF9BuObLLfD68vRiz/UeHoS90AbzfQQQ8EWA4OULK4MiEA0BF74qV6/W541J3ei9WbBJu7sX3dE3d01X15XLhK6CdYIdI4DA2wQIXm8T4nMEEJhXoLU1pfYT7TrUelC7duf/1KNb3uKTutpcAHQX0rNsxLzt4kMEECiwAMGrwA1g9wiEQcA9eHrw3qBisZg+rq7WsfZTvk/LPQaoJlGnTF9vbpkLFwB5IYAAAkEX4K7GoHeI+hAwJtDf368LFztzq8Vv2LRFjcl6T+987LzcpT/uPH/2YupgSslk0pgQ5SKAQJQFCF5R7j5zR8BHARfAuq9d18ULaX2xq0mVVWu0ZfPGdwph7ujW3bt/qu9Gj5YUF6tx+pqyRCLBaUUf+8fQCCDgjwDByx9XRkUAgRcCExMTGhgY0O3f7uha9w/aUVOr0tIyLVq8WGVlpSoqKnrFanQ0q6mpKT36fzz3nerqj7Ru/Xpt37aVC+df0WIDAghYEiB4WeoWtSIQAgF3JCybzeqpYhoaGn7tjEpKSrRsaYmWL1+WC1rxePy1/8dGBBBAwJoAwctax6gXAQQQQAABBMwKcFej2dZROAIIIIAAAghYEyB4WesY9SKAAAIIIICAWQGCl9nWUTgCCCCAAAIIWBMgeFnrGPUigAACCCCAgFkBgpfZ1lE4AggggAACCFgTIHhZ6xj1IoAAAggggIBZAYKX2dZROAIIIIAAAghYEyB4WesY9SKAAAIIIICAWQGCl9nWUTgCCCCAAAIIWBMgeFnrGPUigAACCCCAgFkBgpfZ1lE4AggggAACCFgTIHhZ6xj1IoAAAggggIBZAYKX2dZROAIIIIAAAghYEyB4WesY9SKAAAIIIICAWQGCl9nWUTgCCCCAAAIIWBMgeFnrGPUigAACCCCAgFkBgpfZ1lE4AggggAACCFgTIHhZ6xj1IoAAAggggIBZAYKX2dZROAIIIIAAAghYEyB4WesY9SKAAAIIIICAWYFnu/zIiInRwogAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 48,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Image(filename='sentiment_network_sparse.png')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [],
   "source": [
    "layer_0 = np.zeros(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0., 0., 0., 0., 0., 0., 0., 0., 0., 0.])"
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "layer_0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [],
   "source": [
    "layer_0[4] = 1\n",
    "layer_0[9] = 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0., 0., 0., 0., 1., 0., 0., 0., 0., 1.])"
      ]
     },
     "execution_count": 52,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "layer_0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [],
   "source": [
    "weights_0_1 = np.random.randn(10,5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([-0.10503756,  0.44222989,  0.24392938, -0.55961832,  0.21389503])"
      ]
     },
     "execution_count": 54,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "layer_0.dot(weights_0_1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [],
   "source": [
    "indices = [4,9]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [],
   "source": [
    "layer_1 = np.zeros(5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [],
   "source": [
    "for index in indices:\n",
    "    layer_1 += (1 * weights_0_1[index])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([-0.10503756,  0.44222989,  0.24392938, -0.55961832,  0.21389503])"
      ]
     },
     "execution_count": 58,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "layer_1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 59,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Image(filename='sentiment_network_sparse_2.png')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [],
   "source": [
    "layer_1 = np.zeros(5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [],
   "source": [
    "for index in indices:\n",
    "    layer_1 += (weights_0_1[index])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([-0.10503756,  0.44222989,  0.24392938, -0.55961832,  0.21389503])"
      ]
     },
     "execution_count": 62,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "layer_1"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 项目 5：提高网络效率<a id='project_5'></a>\n",
    "**待办：** 删除前向传播和反向传播造成的冗余相乘和相加，从而提高类 `SentimentNetwork` 的效率，为此，可进行如下操作：\n",
    "* 将上一个项目的类 `SentimentNetwork` 复制到下方代码框中。\n",
    "* 删除函数 `update_input_layer`——本版本不需要该函数。\n",
    "* 修改 `init_network`；\n",
    ">* 因为你已经不需要单独的输入层了，所以删除任何与 `self.layer_0` 有关的内容\n",
    ">* 因为你要更直接地处理之前的隐藏层，所以要创建一个二维矩阵 `self.layer_1`，使其形状为 1 x hidden_nodes，并将其所有值初始化为 0。\n",
    "* 修改 `train`：\n",
    ">* 将输入参数名从 `training_reviews` 改为 `training_reviews_raw`，为下一步做好准备。\n",
    ">* 你需要在函数开头部分对评论进行预处理，将其转换成一张索引列表，其中的索引（源自 `word2index` ）是在评论中真正出现过的，这与 Andrew 在视频中将具体索引设为 1 的操作有异曲同工之妙。你应该用代码创建一个局部的`列表`变量 `training_reviews`，对于 `training_reviews_raw` 中的 每一条评论，该变量都会建立一个与之相关的列表，而这些列表应包含在评论里出现过的单词的索引。\n",
    ">* 取消对函数 `update_input_layer` 的调用。\n",
    ">* 在前向传播部分，将更新 `layer_1` 的代码替换为新的逻辑代码，只为在评论里出现过的索引添加权重。\n",
    ">* 更新 `weights_0_1` 时，只更新用于前向传播的各权重。\n",
    "* 修改 `run`；\n",
    ">* 取消对函数 `update_input_layer` 的调用。\n",
    ">* 使用 `self` 的 `layer_1` 而非局部的 `layer_1` 对象。\n",
    ">* 和修改 `train` 一样，你需要对 `评论` 进行预处理，以便利用单词索引，然后给在评论中出现过的索引增加权重，从而对 `layer_1` 进行更新。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [],
   "source": [
    "import time\n",
    "import sys\n",
    "import numpy as np\n",
    "\n",
    "# Encapsulate our neural network in a class\n",
    "class SentimentNetwork:\n",
    "    def __init__(self, reviews,labels,hidden_nodes = 10, learning_rate = 0.1):\n",
    "        \"\"\"Create a SentimenNetwork with the given settings\n",
    "        Args:\n",
    "            reviews(list) - List of reviews used for training\n",
    "            labels(list) - List of POSITIVE/NEGATIVE labels associated with the given reviews\n",
    "            hidden_nodes(int) - Number of nodes to create in the hidden layer\n",
    "            learning_rate(float) - Learning rate to use while training\n",
    "        \n",
    "        \"\"\"\n",
    "        # Assign a seed to our random number generator to ensure we get\n",
    "        # reproducable results during development \n",
    "        np.random.seed(1)\n",
    "\n",
    "        # process the reviews and their associated labels so that everything\n",
    "        # is ready for training\n",
    "        self.pre_process_data(reviews, labels)\n",
    "        \n",
    "        # Build the network to have the number of hidden nodes and the learning rate that\n",
    "        # were passed into this initializer. Make the same number of input nodes as\n",
    "        # there are vocabulary words and create a single output node.\n",
    "        self.init_network(len(self.review_vocab),hidden_nodes, 1, learning_rate)\n",
    "\n",
    "    def pre_process_data(self, reviews, labels):\n",
    "        \n",
    "        # populate review_vocab with all of the words in the given reviews\n",
    "        review_vocab = set()\n",
    "        for review in reviews:\n",
    "            for word in review.split(\" \"):\n",
    "                review_vocab.add(word)\n",
    "\n",
    "        # Convert the vocabulary set to a list so we can access words via indices\n",
    "        self.review_vocab = list(review_vocab)\n",
    "        \n",
    "        # populate label_vocab with all of the words in the given labels.\n",
    "        label_vocab = set()\n",
    "        for label in labels:\n",
    "            label_vocab.add(label)\n",
    "        \n",
    "        # Convert the label vocabulary set to a list so we can access labels via indices\n",
    "        self.label_vocab = list(label_vocab)\n",
    "        \n",
    "        # Store the sizes of the review and label vocabularies.\n",
    "        self.review_vocab_size = len(self.review_vocab)\n",
    "        self.label_vocab_size = len(self.label_vocab)\n",
    "        \n",
    "        # Create a dictionary of words in the vocabulary mapped to index positions\n",
    "        self.word2index = {}\n",
    "        for i, word in enumerate(self.review_vocab):\n",
    "            self.word2index[word] = i\n",
    "        \n",
    "        # Create a dictionary of labels mapped to index positions\n",
    "        self.label2index = {}\n",
    "        for i, label in enumerate(self.label_vocab):\n",
    "            self.label2index[label] = i\n",
    "\n",
    "    def init_network(self, input_nodes, hidden_nodes, output_nodes, learning_rate):\n",
    "        # Set number of nodes in input, hidden and output layers.\n",
    "        self.input_nodes = input_nodes\n",
    "        self.hidden_nodes = hidden_nodes\n",
    "        self.output_nodes = output_nodes\n",
    "\n",
    "        # Store the learning rate\n",
    "        self.learning_rate = learning_rate\n",
    "\n",
    "        # Initialize weights\n",
    "\n",
    "        # These are the weights between the input layer and the hidden layer.\n",
    "        self.weights_0_1 = np.zeros((self.input_nodes,self.hidden_nodes))\n",
    "\n",
    "        # These are the weights between the hidden layer and the output layer.\n",
    "        self.weights_1_2 = np.random.normal(0.0, self.output_nodes**-0.5, \n",
    "                                                (self.hidden_nodes, self.output_nodes))\n",
    "        \n",
    "        ## New for Project 5: Removed self.layer_0; added self.layer_1\n",
    "        # The input layer, a two-dimensional matrix with shape 1 x hidden_nodes\n",
    "        self.layer_1 = np.zeros((1,hidden_nodes))\n",
    "    \n",
    "    ## New for Project 5: Removed update_input_layer function\n",
    "    \n",
    "    def get_target_for_label(self,label):\n",
    "        if(label == 'POSITIVE'):\n",
    "            return 1\n",
    "        else:\n",
    "            return 0\n",
    "        \n",
    "    def sigmoid(self,x):\n",
    "        return 1 / (1 + np.exp(-x))\n",
    "    \n",
    "    def sigmoid_output_2_derivative(self,output):\n",
    "        return output * (1 - output)\n",
    "    \n",
    "    ## New for Project 5: changed name of first parameter form 'training_reviews' \n",
    "    #                     to 'training_reviews_raw'\n",
    "    def train(self, training_reviews_raw, training_labels):\n",
    "\n",
    "        ## New for Project 5: pre-process training reviews so we can deal \n",
    "        #                     directly with the indices of non-zero inputs\n",
    "        training_reviews = list()\n",
    "        for review in training_reviews_raw:\n",
    "            indices = set()\n",
    "            for word in review.split(\" \"):\n",
    "                if(word in self.word2index.keys()):\n",
    "                    indices.add(self.word2index[word])\n",
    "            training_reviews.append(list(indices))\n",
    "\n",
    "        # make sure out we have a matching number of reviews and labels\n",
    "        assert(len(training_reviews) == len(training_labels))\n",
    "        \n",
    "        # Keep track of correct predictions to display accuracy during training \n",
    "        correct_so_far = 0\n",
    "\n",
    "        # Remember when we started for printing time statistics\n",
    "        start = time.time()\n",
    "        \n",
    "        # loop through all the given reviews and run a forward and backward pass,\n",
    "        # updating weights for every item\n",
    "        for i in range(len(training_reviews)):\n",
    "            \n",
    "            # Get the next review and its correct label\n",
    "            review = training_reviews[i]\n",
    "            label = training_labels[i]\n",
    "            \n",
    "            #### Implement the forward pass here ####\n",
    "            ### Forward pass ###\n",
    "\n",
    "            ## New for Project 5: Removed call to 'update_input_layer' function\n",
    "            #                     because 'layer_0' is no longer used\n",
    "\n",
    "            # Hidden layer\n",
    "            ## New for Project 5: Add in only the weights for non-zero items\n",
    "            self.layer_1 *= 0\n",
    "            for index in review:\n",
    "                self.layer_1 += self.weights_0_1[index]\n",
    "\n",
    "            # Output layer\n",
    "            ## New for Project 5: changed to use 'self.layer_1' instead of 'local layer_1'\n",
    "            layer_2 = self.sigmoid(self.layer_1.dot(self.weights_1_2))            \n",
    "            \n",
    "            #### Implement the backward pass here ####\n",
    "            ### Backward pass ###\n",
    "\n",
    "            # Output error\n",
    "            layer_2_error = layer_2 - self.get_target_for_label(label) # Output layer error is the difference between desired target and actual output.\n",
    "            layer_2_delta = layer_2_error * self.sigmoid_output_2_derivative(layer_2)\n",
    "\n",
    "            # Backpropagated error\n",
    "            layer_1_error = layer_2_delta.dot(self.weights_1_2.T) # errors propagated to the hidden layer\n",
    "            layer_1_delta = layer_1_error # hidden layer gradients - no nonlinearity so it's the same as the error\n",
    "\n",
    "            # Update the weights\n",
    "            ## New for Project 5: changed to use 'self.layer_1' instead of local 'layer_1'\n",
    "            self.weights_1_2 -= self.layer_1.T.dot(layer_2_delta) * self.learning_rate # update hidden-to-output weights with gradient descent step\n",
    "            \n",
    "            ## New for Project 5: Only update the weights that were used in the forward pass\n",
    "            for index in review:\n",
    "                self.weights_0_1[index] -= layer_1_delta[0] * self.learning_rate # update input-to-hidden weights with gradient descent step\n",
    "\n",
    "            # Keep track of correct predictions.\n",
    "            if(layer_2 >= 0.5 and label == 'POSITIVE'):\n",
    "                correct_so_far += 1\n",
    "            elif(layer_2 < 0.5 and label == 'NEGATIVE'):\n",
    "                correct_so_far += 1\n",
    "            \n",
    "            # For debug purposes, print out our prediction accuracy and speed \n",
    "            # throughout the training process. \n",
    "            elapsed_time = float(time.time() - start)\n",
    "            reviews_per_second = i / elapsed_time if elapsed_time > 0 else 0\n",
    "            \n",
    "            if (i % 500 == 0):\n",
    "                sys.stdout.write(\"\\rProgress:\" + str(100 * i/float(len(training_reviews)))[:4] \\\n",
    "                                 + \"% Speed(reviews/sec):\" + str(reviews_per_second)[0:5] \\\n",
    "                                 + \" #Correct:\" + str(correct_so_far) + \" #Trained:\" + str(i+1) \\\n",
    "                                 + \" Training Accuracy:\" + str(correct_so_far * 100 / float(i+1))[:4] + \"%\")\n",
    "                if(i % 2500 == 0):\n",
    "                    print(\"\")\n",
    "    \n",
    "    def test(self, testing_reviews, testing_labels):\n",
    "        \"\"\"\n",
    "        Attempts to predict the labels for the given testing_reviews,\n",
    "        and uses the test_labels to calculate the accuracy of those predictions.\n",
    "        \"\"\"\n",
    "        \n",
    "        # keep track of how many correct predictions we make\n",
    "        correct = 0\n",
    "\n",
    "        # we'll time how many predictions per second we make\n",
    "        start = time.time()\n",
    "\n",
    "        # Loop through each of the given reviews and call run to predict\n",
    "        # its label. \n",
    "        for i in range(len(testing_reviews)):\n",
    "            pred = self.run(testing_reviews[i])\n",
    "            if(pred == testing_labels[i]):\n",
    "                correct += 1\n",
    "            \n",
    "            # For debug purposes, print out our prediction accuracy and speed \n",
    "            # throughout the prediction process. \n",
    "\n",
    "            elapsed_time = float(time.time() - start)\n",
    "            reviews_per_second = i / elapsed_time if elapsed_time > 0 else 0\n",
    "            \n",
    "            sys.stdout.write(\"\\rProgress:\" + str(100 * i/float(len(testing_reviews)))[:4] \\\n",
    "                             + \"% Speed(reviews/sec):\" + str(reviews_per_second)[0:5] \\\n",
    "                             + \" #Correct:\" + str(correct) + \" #Tested:\" + str(i+1) \\\n",
    "                             + \" Testing Accuracy:\" + str(correct * 100 / float(i+1))[:4] + \"%\")\n",
    "    \n",
    "    def run(self, review):\n",
    "        \"\"\"\n",
    "        Returns a POSITIVE or NEGATIVE prediction for the given review.\n",
    "        \"\"\"\n",
    "        # Run a forward pass through the network, like in the \"train\" function.\n",
    "        \n",
    "        ## New for Project 5: Removed call to update_input_layer function\n",
    "        #                     because layer_0 is no longer used\n",
    "\n",
    "        # Hidden layer\n",
    "        ## New for Project 5: Identify the indices used in the review and then add\n",
    "        #                     just those weights to layer_1 \n",
    "        self.layer_1 *= 0\n",
    "        unique_indices = set()\n",
    "        for word in review.lower().split(\" \"):\n",
    "            if word in self.word2index.keys():\n",
    "                unique_indices.add(self.word2index[word])\n",
    "        for index in unique_indices:\n",
    "            self.layer_1 += self.weights_0_1[index]\n",
    "        \n",
    "        # Output layer\n",
    "        ## New for Project 5: changed to use self.layer_1 instead of local layer_1\n",
    "        layer_2 = self.sigmoid(self.layer_1.dot(self.weights_1_2))\n",
    "        \n",
    "        # Return POSITIVE for values above greater-than-or-equal-to 0.5 in the output layer;\n",
    "        # return NEGATIVE for other values\n",
    "        if(layer_2[0] >= 0.5):\n",
    "            return \"POSITIVE\"\n",
    "        else:\n",
    "            return \"NEGATIVE\""
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "运行下方代码框，再次重建网络并对网络进行训练。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Progress:0.0% Speed(reviews/sec):0.0 #Correct:1 #Trained:1 Training Accuracy:100.%\n",
      "Progress:10.4% Speed(reviews/sec):1098. #Correct:1778 #Trained:2501 Training Accuracy:71.0%\n",
      "Progress:20.8% Speed(reviews/sec):1168. #Correct:3761 #Trained:5001 Training Accuracy:75.2%\n",
      "Progress:31.2% Speed(reviews/sec):1194. #Correct:5859 #Trained:7501 Training Accuracy:78.1%\n",
      "Progress:41.6% Speed(reviews/sec):1203. #Correct:8011 #Trained:10001 Training Accuracy:80.1%\n",
      "Progress:52.0% Speed(reviews/sec):1200. #Correct:10151 #Trained:12501 Training Accuracy:81.2%\n",
      "Progress:62.5% Speed(reviews/sec):1220. #Correct:12281 #Trained:15001 Training Accuracy:81.8%\n",
      "Progress:72.9% Speed(reviews/sec):1206. #Correct:14399 #Trained:17501 Training Accuracy:82.2%\n",
      "Progress:83.3% Speed(reviews/sec):1194. #Correct:16572 #Trained:20001 Training Accuracy:82.8%\n",
      "Progress:93.7% Speed(reviews/sec):1182. #Correct:18748 #Trained:22501 Training Accuracy:83.3%\n",
      "Progress:97.9% Speed(reviews/sec):1182. #Correct:19608 #Trained:23501 Training Accuracy:83.4%"
     ]
    }
   ],
   "source": [
    "mlp = SentimentNetwork(reviews[:-1000],labels[:-1000], learning_rate=0.1)\n",
    "mlp.train(reviews[:-1000],labels[:-1000])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "训练效果应该比之前的要好得多了。运行下方代码框，用那一千条预测来测试你的模型。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\r",
      "Progress:0.0% Speed(reviews/sec):0.0 #Correct:1 #Tested:1 Testing Accuracy:100.%\r",
      "Progress:0.1% Speed(reviews/sec):1173. #Correct:1 #Tested:2 Testing Accuracy:50.0%\r",
      "Progress:0.2% Speed(reviews/sec):1627. #Correct:2 #Tested:3 Testing Accuracy:66.6%\r",
      "Progress:0.3% Speed(reviews/sec):1687. #Correct:3 #Tested:4 Testing Accuracy:75.0%\r",
      "Progress:0.4% Speed(reviews/sec):1978. #Correct:4 #Tested:5 Testing Accuracy:80.0%\r",
      "Progress:0.5% Speed(reviews/sec):1623. #Correct:5 #Tested:6 Testing Accuracy:83.3%\r",
      "Progress:0.6% Speed(reviews/sec):1364. #Correct:6 #Tested:7 Testing Accuracy:85.7%\r",
      "Progress:0.7% Speed(reviews/sec):1326. #Correct:7 #Tested:8 Testing Accuracy:87.5%\r",
      "Progress:0.8% Speed(reviews/sec):1301. #Correct:8 #Tested:9 Testing Accuracy:88.8%\r",
      "Progress:0.9% Speed(reviews/sec):1095. #Correct:9 #Tested:10 Testing Accuracy:90.0%\r",
      "Progress:1.0% Speed(reviews/sec):1129. #Correct:10 #Tested:11 Testing Accuracy:90.9%\r",
      "Progress:1.1% Speed(reviews/sec):1168. #Correct:11 #Tested:12 Testing Accuracy:91.6%\r",
      "Progress:1.2% Speed(reviews/sec):1174. #Correct:12 #Tested:13 Testing Accuracy:92.3%\r",
      "Progress:1.3% Speed(reviews/sec):1202. #Correct:13 #Tested:14 Testing Accuracy:92.8%\r",
      "Progress:1.4% Speed(reviews/sec):1198. #Correct:14 #Tested:15 Testing Accuracy:93.3%\r",
      "Progress:1.5% Speed(reviews/sec):1208. #Correct:15 #Tested:16 Testing Accuracy:93.7%\r",
      "Progress:1.6% Speed(reviews/sec):1169. #Correct:16 #Tested:17 Testing Accuracy:94.1%\r",
      "Progress:1.7% Speed(reviews/sec):1189. #Correct:17 #Tested:18 Testing Accuracy:94.4%\r",
      "Progress:1.8% Speed(reviews/sec):1224. #Correct:18 #Tested:19 Testing Accuracy:94.7%\r",
      "Progress:1.9% Speed(reviews/sec):1240. #Correct:18 #Tested:20 Testing Accuracy:90.0%\r",
      "Progress:2.0% Speed(reviews/sec):1241. #Correct:19 #Tested:21 Testing Accuracy:90.4%\r",
      "Progress:2.1% Speed(reviews/sec):1267. #Correct:20 #Tested:22 Testing Accuracy:90.9%\r",
      "Progress:2.2% Speed(reviews/sec):1276. #Correct:21 #Tested:23 Testing Accuracy:91.3%\r",
      "Progress:2.3% Speed(reviews/sec):1259. #Correct:22 #Tested:24 Testing Accuracy:91.6%\r",
      "Progress:2.4% Speed(reviews/sec):1180. #Correct:23 #Tested:25 Testing Accuracy:92.0%\r",
      "Progress:2.5% Speed(reviews/sec):1098. #Correct:24 #Tested:26 Testing Accuracy:92.3%\r",
      "Progress:2.6% Speed(reviews/sec):1123. #Correct:25 #Tested:27 Testing Accuracy:92.5%\r",
      "Progress:2.7% Speed(reviews/sec):1093. #Correct:26 #Tested:28 Testing Accuracy:92.8%\r",
      "Progress:2.8% Speed(reviews/sec):1109. #Correct:27 #Tested:29 Testing Accuracy:93.1%\r",
      "Progress:2.9% Speed(reviews/sec):1103. #Correct:28 #Tested:30 Testing Accuracy:93.3%\r",
      "Progress:3.0% Speed(reviews/sec):1067. #Correct:29 #Tested:31 Testing Accuracy:93.5%\r",
      "Progress:3.1% Speed(reviews/sec):1087. #Correct:30 #Tested:32 Testing Accuracy:93.7%\r",
      "Progress:3.2% Speed(reviews/sec):1102. #Correct:31 #Tested:33 Testing Accuracy:93.9%\r",
      "Progress:3.3% Speed(reviews/sec):1121. #Correct:32 #Tested:34 Testing Accuracy:94.1%\r",
      "Progress:3.4% Speed(reviews/sec):1130. #Correct:33 #Tested:35 Testing Accuracy:94.2%\r",
      "Progress:3.5% Speed(reviews/sec):1145. #Correct:34 #Tested:36 Testing Accuracy:94.4%\r",
      "Progress:3.6% Speed(reviews/sec):1150. #Correct:35 #Tested:37 Testing Accuracy:94.5%\r",
      "Progress:3.7% Speed(reviews/sec):1174. #Correct:36 #Tested:38 Testing Accuracy:94.7%\r",
      "Progress:3.8% Speed(reviews/sec):1180. #Correct:37 #Tested:39 Testing Accuracy:94.8%\r",
      "Progress:3.9% Speed(reviews/sec):1180. #Correct:37 #Tested:40 Testing Accuracy:92.5%\r",
      "Progress:4.0% Speed(reviews/sec):1180. #Correct:38 #Tested:41 Testing Accuracy:92.6%\r",
      "Progress:4.1% Speed(reviews/sec):1187. #Correct:39 #Tested:42 Testing Accuracy:92.8%\r",
      "Progress:4.2% Speed(reviews/sec):1136. #Correct:40 #Tested:43 Testing Accuracy:93.0%\r",
      "Progress:4.3% Speed(reviews/sec):1139. #Correct:41 #Tested:44 Testing Accuracy:93.1%\r",
      "Progress:4.4% Speed(reviews/sec):1134. #Correct:42 #Tested:45 Testing Accuracy:93.3%\r",
      "Progress:4.5% Speed(reviews/sec):1116. #Correct:42 #Tested:46 Testing Accuracy:91.3%\r",
      "Progress:4.6% Speed(reviews/sec):1116. #Correct:43 #Tested:47 Testing Accuracy:91.4%\r",
      "Progress:4.7% Speed(reviews/sec):1132. #Correct:44 #Tested:48 Testing Accuracy:91.6%\r",
      "Progress:4.8% Speed(reviews/sec):1096. #Correct:45 #Tested:49 Testing Accuracy:91.8%\r",
      "Progress:4.9% Speed(reviews/sec):1108. #Correct:46 #Tested:50 Testing Accuracy:92.0%\r",
      "Progress:5.0% Speed(reviews/sec):1121. #Correct:47 #Tested:51 Testing Accuracy:92.1%\r",
      "Progress:5.1% Speed(reviews/sec):1137. #Correct:48 #Tested:52 Testing Accuracy:92.3%\r",
      "Progress:5.2% Speed(reviews/sec):1133. #Correct:49 #Tested:53 Testing Accuracy:92.4%\r",
      "Progress:5.3% Speed(reviews/sec):1144. #Correct:50 #Tested:54 Testing Accuracy:92.5%\r",
      "Progress:5.4% Speed(reviews/sec):1158. #Correct:51 #Tested:55 Testing Accuracy:92.7%\r",
      "Progress:5.5% Speed(reviews/sec):1174. #Correct:52 #Tested:56 Testing Accuracy:92.8%\r",
      "Progress:5.6% Speed(reviews/sec):1156. #Correct:53 #Tested:57 Testing Accuracy:92.9%\r",
      "Progress:5.7% Speed(reviews/sec):1172. #Correct:54 #Tested:58 Testing Accuracy:93.1%\r",
      "Progress:5.8% Speed(reviews/sec):1187. #Correct:55 #Tested:59 Testing Accuracy:93.2%\r",
      "Progress:5.9% Speed(reviews/sec):1201. #Correct:56 #Tested:60 Testing Accuracy:93.3%\r",
      "Progress:6.0% Speed(reviews/sec):1192. #Correct:57 #Tested:61 Testing Accuracy:93.4%\r",
      "Progress:6.1% Speed(reviews/sec):1178. #Correct:58 #Tested:62 Testing Accuracy:93.5%\r",
      "Progress:6.2% Speed(reviews/sec):1191. #Correct:59 #Tested:63 Testing Accuracy:93.6%\r",
      "Progress:6.3% Speed(reviews/sec):1202. #Correct:60 #Tested:64 Testing Accuracy:93.7%\r",
      "Progress:6.4% Speed(reviews/sec):1201. #Correct:60 #Tested:65 Testing Accuracy:92.3%\r",
      "Progress:6.5% Speed(reviews/sec):1212. #Correct:60 #Tested:66 Testing Accuracy:90.9%\r",
      "Progress:6.6% Speed(reviews/sec):1223. #Correct:61 #Tested:67 Testing Accuracy:91.0%\r",
      "Progress:6.7% Speed(reviews/sec):1231. #Correct:62 #Tested:68 Testing Accuracy:91.1%\r",
      "Progress:6.8% Speed(reviews/sec):1240. #Correct:63 #Tested:69 Testing Accuracy:91.3%\r",
      "Progress:6.9% Speed(reviews/sec):1244. #Correct:64 #Tested:70 Testing Accuracy:91.4%\r",
      "Progress:7.0% Speed(reviews/sec):1248. #Correct:64 #Tested:71 Testing Accuracy:90.1%\r",
      "Progress:7.1% Speed(reviews/sec):1261. #Correct:65 #Tested:72 Testing Accuracy:90.2%\r",
      "Progress:7.2% Speed(reviews/sec):1271. #Correct:66 #Tested:73 Testing Accuracy:90.4%\r",
      "Progress:7.3% Speed(reviews/sec):1284. #Correct:67 #Tested:74 Testing Accuracy:90.5%\r",
      "Progress:7.4% Speed(reviews/sec):1292. #Correct:67 #Tested:75 Testing Accuracy:89.3%\r",
      "Progress:7.5% Speed(reviews/sec):1296. #Correct:68 #Tested:76 Testing Accuracy:89.4%\r",
      "Progress:7.6% Speed(reviews/sec):1296. #Correct:69 #Tested:77 Testing Accuracy:89.6%\r",
      "Progress:7.7% Speed(reviews/sec):1289. #Correct:70 #Tested:78 Testing Accuracy:89.7%\r",
      "Progress:7.8% Speed(reviews/sec):1295. #Correct:71 #Tested:79 Testing Accuracy:89.8%\r",
      "Progress:7.9% Speed(reviews/sec):1302. #Correct:72 #Tested:80 Testing Accuracy:90.0%\r",
      "Progress:8.0% Speed(reviews/sec):1310. #Correct:73 #Tested:81 Testing Accuracy:90.1%\r",
      "Progress:8.1% Speed(reviews/sec):1294. #Correct:73 #Tested:82 Testing Accuracy:89.0%\r",
      "Progress:8.2% Speed(reviews/sec):1299. #Correct:74 #Tested:83 Testing Accuracy:89.1%\r",
      "Progress:8.3% Speed(reviews/sec):1304. #Correct:75 #Tested:84 Testing Accuracy:89.2%\r",
      "Progress:8.4% Speed(reviews/sec):1315. #Correct:76 #Tested:85 Testing Accuracy:89.4%\r",
      "Progress:8.5% Speed(reviews/sec):1322. #Correct:77 #Tested:86 Testing Accuracy:89.5%\r",
      "Progress:8.6% Speed(reviews/sec):1322. #Correct:78 #Tested:87 Testing Accuracy:89.6%\r",
      "Progress:8.7% Speed(reviews/sec):1331. #Correct:78 #Tested:88 Testing Accuracy:88.6%\r",
      "Progress:8.8% Speed(reviews/sec):1337. #Correct:79 #Tested:89 Testing Accuracy:88.7%\r",
      "Progress:8.9% Speed(reviews/sec):1310. #Correct:80 #Tested:90 Testing Accuracy:88.8%\r",
      "Progress:9.0% Speed(reviews/sec):1312. #Correct:81 #Tested:91 Testing Accuracy:89.0%\r",
      "Progress:9.1% Speed(reviews/sec):1319. #Correct:82 #Tested:92 Testing Accuracy:89.1%\r",
      "Progress:9.2% Speed(reviews/sec):1327. #Correct:83 #Tested:93 Testing Accuracy:89.2%\r",
      "Progress:9.3% Speed(reviews/sec):1334. #Correct:84 #Tested:94 Testing Accuracy:89.3%\r",
      "Progress:9.4% Speed(reviews/sec):1341. #Correct:85 #Tested:95 Testing Accuracy:89.4%\r",
      "Progress:9.5% Speed(reviews/sec):1345. #Correct:86 #Tested:96 Testing Accuracy:89.5%\r",
      "Progress:9.6% Speed(reviews/sec):1348. #Correct:87 #Tested:97 Testing Accuracy:89.6%\r",
      "Progress:9.7% Speed(reviews/sec):1349. #Correct:88 #Tested:98 Testing Accuracy:89.7%\r",
      "Progress:9.8% Speed(reviews/sec):1356. #Correct:89 #Tested:99 Testing Accuracy:89.8%\r",
      "Progress:9.9% Speed(reviews/sec):1364. #Correct:90 #Tested:100 Testing Accuracy:90.0%\r",
      "Progress:10.0% Speed(reviews/sec):1370. #Correct:91 #Tested:101 Testing Accuracy:90.0%\r",
      "Progress:10.1% Speed(reviews/sec):1375. #Correct:92 #Tested:102 Testing Accuracy:90.1%\r",
      "Progress:10.2% Speed(reviews/sec):1374. #Correct:93 #Tested:103 Testing Accuracy:90.2%\r",
      "Progress:10.3% Speed(reviews/sec):1353. #Correct:93 #Tested:104 Testing Accuracy:89.4%\r",
      "Progress:10.4% Speed(reviews/sec):1308. #Correct:94 #Tested:105 Testing Accuracy:89.5%\r",
      "Progress:10.5% Speed(reviews/sec):1317. #Correct:95 #Tested:106 Testing Accuracy:89.6%\r",
      "Progress:10.6% Speed(reviews/sec):1319. #Correct:96 #Tested:107 Testing Accuracy:89.7%\r",
      "Progress:10.7% Speed(reviews/sec):1324. #Correct:97 #Tested:108 Testing Accuracy:89.8%\r",
      "Progress:10.8% Speed(reviews/sec):1331. #Correct:97 #Tested:109 Testing Accuracy:88.9%\r",
      "Progress:10.9% Speed(reviews/sec):1333. #Correct:97 #Tested:110 Testing Accuracy:88.1%\r",
      "Progress:11.0% Speed(reviews/sec):1319. #Correct:98 #Tested:111 Testing Accuracy:88.2%\r",
      "Progress:11.1% Speed(reviews/sec):1320. #Correct:99 #Tested:112 Testing Accuracy:88.3%\r",
      "Progress:11.2% Speed(reviews/sec):1323. #Correct:100 #Tested:113 Testing Accuracy:88.4%\r",
      "Progress:11.3% Speed(reviews/sec):1323. #Correct:101 #Tested:114 Testing Accuracy:88.5%\r",
      "Progress:11.4% Speed(reviews/sec):1327. #Correct:102 #Tested:115 Testing Accuracy:88.6%\r",
      "Progress:11.5% Speed(reviews/sec):1333. #Correct:102 #Tested:116 Testing Accuracy:87.9%\r",
      "Progress:11.6% Speed(reviews/sec):1339. #Correct:103 #Tested:117 Testing Accuracy:88.0%\r",
      "Progress:11.7% Speed(reviews/sec):1342. #Correct:104 #Tested:118 Testing Accuracy:88.1%\r",
      "Progress:11.8% Speed(reviews/sec):1348. #Correct:105 #Tested:119 Testing Accuracy:88.2%\r",
      "Progress:11.9% Speed(reviews/sec):1351. #Correct:106 #Tested:120 Testing Accuracy:88.3%\r",
      "Progress:12.0% Speed(reviews/sec):1354. #Correct:107 #Tested:121 Testing Accuracy:88.4%\r",
      "Progress:12.1% Speed(reviews/sec):1359. #Correct:108 #Tested:122 Testing Accuracy:88.5%\r",
      "Progress:12.2% Speed(reviews/sec):1365. #Correct:109 #Tested:123 Testing Accuracy:88.6%\r",
      "Progress:12.3% Speed(reviews/sec):1335. #Correct:110 #Tested:124 Testing Accuracy:88.7%\r",
      "Progress:12.4% Speed(reviews/sec):1336. #Correct:111 #Tested:125 Testing Accuracy:88.8%\r",
      "Progress:12.5% Speed(reviews/sec):1338. #Correct:112 #Tested:126 Testing Accuracy:88.8%\r",
      "Progress:12.6% Speed(reviews/sec):1346. #Correct:113 #Tested:127 Testing Accuracy:88.9%\r",
      "Progress:12.7% Speed(reviews/sec):1347. #Correct:114 #Tested:128 Testing Accuracy:89.0%\r",
      "Progress:12.8% Speed(reviews/sec):1353. #Correct:115 #Tested:129 Testing Accuracy:89.1%\r",
      "Progress:12.9% Speed(reviews/sec):1359. #Correct:116 #Tested:130 Testing Accuracy:89.2%\r",
      "Progress:13.0% Speed(reviews/sec):1364. #Correct:116 #Tested:131 Testing Accuracy:88.5%\r",
      "Progress:13.1% Speed(reviews/sec):1370. #Correct:117 #Tested:132 Testing Accuracy:88.6%\r",
      "Progress:13.2% Speed(reviews/sec):1375. #Correct:118 #Tested:133 Testing Accuracy:88.7%\r",
      "Progress:13.3% Speed(reviews/sec):1381. #Correct:119 #Tested:134 Testing Accuracy:88.8%\r",
      "Progress:13.4% Speed(reviews/sec):1382. #Correct:120 #Tested:135 Testing Accuracy:88.8%\r",
      "Progress:13.5% Speed(reviews/sec):1381. #Correct:120 #Tested:136 Testing Accuracy:88.2%\r",
      "Progress:13.6% Speed(reviews/sec):1385. #Correct:121 #Tested:137 Testing Accuracy:88.3%\r",
      "Progress:13.7% Speed(reviews/sec):1379. #Correct:122 #Tested:138 Testing Accuracy:88.4%\r",
      "Progress:13.8% Speed(reviews/sec):1380. #Correct:123 #Tested:139 Testing Accuracy:88.4%\r",
      "Progress:13.9% Speed(reviews/sec):1379. #Correct:124 #Tested:140 Testing Accuracy:88.5%\r",
      "Progress:14.0% Speed(reviews/sec):1370. #Correct:125 #Tested:141 Testing Accuracy:88.6%\r",
      "Progress:14.1% Speed(reviews/sec):1372. #Correct:126 #Tested:142 Testing Accuracy:88.7%\r",
      "Progress:14.2% Speed(reviews/sec):1378. #Correct:127 #Tested:143 Testing Accuracy:88.8%\r",
      "Progress:14.3% Speed(reviews/sec):1383. #Correct:128 #Tested:144 Testing Accuracy:88.8%\r",
      "Progress:14.4% Speed(reviews/sec):1383. #Correct:129 #Tested:145 Testing Accuracy:88.9%\r",
      "Progress:14.5% Speed(reviews/sec):1389. #Correct:130 #Tested:146 Testing Accuracy:89.0%\r",
      "Progress:14.6% Speed(reviews/sec):1388. #Correct:131 #Tested:147 Testing Accuracy:89.1%\r",
      "Progress:14.7% Speed(reviews/sec):1386. #Correct:132 #Tested:148 Testing Accuracy:89.1%\r",
      "Progress:14.8% Speed(reviews/sec):1387. #Correct:133 #Tested:149 Testing Accuracy:89.2%\r",
      "Progress:14.9% Speed(reviews/sec):1387. #Correct:134 #Tested:150 Testing Accuracy:89.3%\r",
      "Progress:15.0% Speed(reviews/sec):1380. #Correct:135 #Tested:151 Testing Accuracy:89.4%\r",
      "Progress:15.1% Speed(reviews/sec):1384. #Correct:136 #Tested:152 Testing Accuracy:89.4%\r",
      "Progress:15.2% Speed(reviews/sec):1383. #Correct:137 #Tested:153 Testing Accuracy:89.5%\r",
      "Progress:15.3% Speed(reviews/sec):1382. #Correct:138 #Tested:154 Testing Accuracy:89.6%\r",
      "Progress:15.4% Speed(reviews/sec):1387. #Correct:139 #Tested:155 Testing Accuracy:89.6%\r",
      "Progress:15.5% Speed(reviews/sec):1392. #Correct:140 #Tested:156 Testing Accuracy:89.7%\r",
      "Progress:15.6% Speed(reviews/sec):1391. #Correct:141 #Tested:157 Testing Accuracy:89.8%\r",
      "Progress:15.7% Speed(reviews/sec):1393. #Correct:141 #Tested:158 Testing Accuracy:89.2%\r",
      "Progress:15.8% Speed(reviews/sec):1398. #Correct:142 #Tested:159 Testing Accuracy:89.3%\r",
      "Progress:15.9% Speed(reviews/sec):1402. #Correct:143 #Tested:160 Testing Accuracy:89.3%\r",
      "Progress:16.0% Speed(reviews/sec):1402. #Correct:143 #Tested:161 Testing Accuracy:88.8%\r",
      "Progress:16.1% Speed(reviews/sec):1403. #Correct:143 #Tested:162 Testing Accuracy:88.2%\r",
      "Progress:16.2% Speed(reviews/sec):1407. #Correct:144 #Tested:163 Testing Accuracy:88.3%\r",
      "Progress:16.3% Speed(reviews/sec):1411. #Correct:145 #Tested:164 Testing Accuracy:88.4%\r",
      "Progress:16.4% Speed(reviews/sec):1416. #Correct:146 #Tested:165 Testing Accuracy:88.4%\r",
      "Progress:16.5% Speed(reviews/sec):1416. #Correct:147 #Tested:166 Testing Accuracy:88.5%\r",
      "Progress:16.6% Speed(reviews/sec):1419. #Correct:148 #Tested:167 Testing Accuracy:88.6%\r",
      "Progress:16.7% Speed(reviews/sec):1420. #Correct:149 #Tested:168 Testing Accuracy:88.6%\r",
      "Progress:16.8% Speed(reviews/sec):1426. #Correct:150 #Tested:169 Testing Accuracy:88.7%\r",
      "Progress:16.9% Speed(reviews/sec):1430. #Correct:151 #Tested:170 Testing Accuracy:88.8%\r",
      "Progress:17.0% Speed(reviews/sec):1432. #Correct:151 #Tested:171 Testing Accuracy:88.3%\r",
      "Progress:17.1% Speed(reviews/sec):1437. #Correct:152 #Tested:172 Testing Accuracy:88.3%\r",
      "Progress:17.2% Speed(reviews/sec):1436. #Correct:153 #Tested:173 Testing Accuracy:88.4%\r",
      "Progress:17.3% Speed(reviews/sec):1438. #Correct:153 #Tested:174 Testing Accuracy:87.9%\r",
      "Progress:17.4% Speed(reviews/sec):1440. #Correct:154 #Tested:175 Testing Accuracy:88.0%\r",
      "Progress:17.5% Speed(reviews/sec):1443. #Correct:155 #Tested:176 Testing Accuracy:88.0%\r",
      "Progress:17.6% Speed(reviews/sec):1446. #Correct:156 #Tested:177 Testing Accuracy:88.1%\r",
      "Progress:17.7% Speed(reviews/sec):1447. #Correct:157 #Tested:178 Testing Accuracy:88.2%\r",
      "Progress:17.8% Speed(reviews/sec):1451. #Correct:158 #Tested:179 Testing Accuracy:88.2%\r",
      "Progress:17.9% Speed(reviews/sec):1455. #Correct:159 #Tested:180 Testing Accuracy:88.3%\r",
      "Progress:18.0% Speed(reviews/sec):1459. #Correct:160 #Tested:181 Testing Accuracy:88.3%\r",
      "Progress:18.1% Speed(reviews/sec):1463. #Correct:161 #Tested:182 Testing Accuracy:88.4%\r",
      "Progress:18.2% Speed(reviews/sec):1466. #Correct:162 #Tested:183 Testing Accuracy:88.5%\r",
      "Progress:18.3% Speed(reviews/sec):1468. #Correct:163 #Tested:184 Testing Accuracy:88.5%\r",
      "Progress:18.4% Speed(reviews/sec):1465. #Correct:164 #Tested:185 Testing Accuracy:88.6%\r",
      "Progress:18.5% Speed(reviews/sec):1467. #Correct:165 #Tested:186 Testing Accuracy:88.7%\r",
      "Progress:18.6% Speed(reviews/sec):1473. #Correct:165 #Tested:187 Testing Accuracy:88.2%\r",
      "Progress:18.7% Speed(reviews/sec):1474. #Correct:166 #Tested:188 Testing Accuracy:88.2%\r",
      "Progress:18.8% Speed(reviews/sec):1475. #Correct:166 #Tested:189 Testing Accuracy:87.8%\r",
      "Progress:18.9% Speed(reviews/sec):1478. #Correct:167 #Tested:190 Testing Accuracy:87.8%\r",
      "Progress:19.0% Speed(reviews/sec):1482. #Correct:168 #Tested:191 Testing Accuracy:87.9%\r",
      "Progress:19.1% Speed(reviews/sec):1486. #Correct:169 #Tested:192 Testing Accuracy:88.0%\r",
      "Progress:19.2% Speed(reviews/sec):1488. #Correct:169 #Tested:193 Testing Accuracy:87.5%\r",
      "Progress:19.3% Speed(reviews/sec):1484. #Correct:169 #Tested:194 Testing Accuracy:87.1%\r",
      "Progress:19.4% Speed(reviews/sec):1459. #Correct:170 #Tested:195 Testing Accuracy:87.1%\r",
      "Progress:19.5% Speed(reviews/sec):1460. #Correct:170 #Tested:196 Testing Accuracy:86.7%\r",
      "Progress:19.6% Speed(reviews/sec):1461. #Correct:171 #Tested:197 Testing Accuracy:86.8%\r",
      "Progress:19.7% Speed(reviews/sec):1464. #Correct:172 #Tested:198 Testing Accuracy:86.8%\r",
      "Progress:19.8% Speed(reviews/sec):1469. #Correct:173 #Tested:199 Testing Accuracy:86.9%\r",
      "Progress:19.9% Speed(reviews/sec):1470. #Correct:174 #Tested:200 Testing Accuracy:87.0%\r",
      "Progress:20.0% Speed(reviews/sec):1475. #Correct:175 #Tested:201 Testing Accuracy:87.0%\r",
      "Progress:20.1% Speed(reviews/sec):1477. #Correct:176 #Tested:202 Testing Accuracy:87.1%\r",
      "Progress:20.2% Speed(reviews/sec):1481. #Correct:177 #Tested:203 Testing Accuracy:87.1%\r",
      "Progress:20.3% Speed(reviews/sec):1481. #Correct:178 #Tested:204 Testing Accuracy:87.2%\r",
      "Progress:20.4% Speed(reviews/sec):1484. #Correct:179 #Tested:205 Testing Accuracy:87.3%\r",
      "Progress:20.5% Speed(reviews/sec):1487. #Correct:180 #Tested:206 Testing Accuracy:87.3%\r",
      "Progress:20.6% Speed(reviews/sec):1490. #Correct:181 #Tested:207 Testing Accuracy:87.4%\r",
      "Progress:20.7% Speed(reviews/sec):1493. #Correct:182 #Tested:208 Testing Accuracy:87.5%\r",
      "Progress:20.8% Speed(reviews/sec):1497. #Correct:183 #Tested:209 Testing Accuracy:87.5%\r",
      "Progress:20.9% Speed(reviews/sec):1496. #Correct:183 #Tested:210 Testing Accuracy:87.1%\r",
      "Progress:21.0% Speed(reviews/sec):1499. #Correct:183 #Tested:211 Testing Accuracy:86.7%\r",
      "Progress:21.1% Speed(reviews/sec):1502. #Correct:184 #Tested:212 Testing Accuracy:86.7%\r",
      "Progress:21.2% Speed(reviews/sec):1506. #Correct:185 #Tested:213 Testing Accuracy:86.8%\r",
      "Progress:21.3% Speed(reviews/sec):1501. #Correct:186 #Tested:214 Testing Accuracy:86.9%\r",
      "Progress:21.4% Speed(reviews/sec):1506. #Correct:187 #Tested:215 Testing Accuracy:86.9%\r",
      "Progress:21.5% Speed(reviews/sec):1506. #Correct:188 #Tested:216 Testing Accuracy:87.0%\r",
      "Progress:21.6% Speed(reviews/sec):1509. #Correct:189 #Tested:217 Testing Accuracy:87.0%\r",
      "Progress:21.7% Speed(reviews/sec):1509. #Correct:190 #Tested:218 Testing Accuracy:87.1%\r",
      "Progress:21.8% Speed(reviews/sec):1509. #Correct:191 #Tested:219 Testing Accuracy:87.2%\r",
      "Progress:21.9% Speed(reviews/sec):1514. #Correct:192 #Tested:220 Testing Accuracy:87.2%\r",
      "Progress:22.0% Speed(reviews/sec):1512. #Correct:193 #Tested:221 Testing Accuracy:87.3%\r",
      "Progress:22.1% Speed(reviews/sec):1515. #Correct:194 #Tested:222 Testing Accuracy:87.3%\r",
      "Progress:22.2% Speed(reviews/sec):1497. #Correct:195 #Tested:223 Testing Accuracy:87.4%\r",
      "Progress:22.3% Speed(reviews/sec):1499. #Correct:196 #Tested:224 Testing Accuracy:87.5%\r",
      "Progress:22.4% Speed(reviews/sec):1494. #Correct:197 #Tested:225 Testing Accuracy:87.5%\r",
      "Progress:22.5% Speed(reviews/sec):1496. #Correct:198 #Tested:226 Testing Accuracy:87.6%\r",
      "Progress:22.6% Speed(reviews/sec):1493. #Correct:199 #Tested:227 Testing Accuracy:87.6%\r",
      "Progress:22.7% Speed(reviews/sec):1495. #Correct:200 #Tested:228 Testing Accuracy:87.7%\r",
      "Progress:22.8% Speed(reviews/sec):1497. #Correct:201 #Tested:229 Testing Accuracy:87.7%\r",
      "Progress:22.9% Speed(reviews/sec):1499. #Correct:202 #Tested:230 Testing Accuracy:87.8%\r",
      "Progress:23.0% Speed(reviews/sec):1497. #Correct:203 #Tested:231 Testing Accuracy:87.8%\r",
      "Progress:23.1% Speed(reviews/sec):1498. #Correct:204 #Tested:232 Testing Accuracy:87.9%\r",
      "Progress:23.2% Speed(reviews/sec):1503. #Correct:205 #Tested:233 Testing Accuracy:87.9%\r",
      "Progress:23.3% Speed(reviews/sec):1505. #Correct:206 #Tested:234 Testing Accuracy:88.0%\r",
      "Progress:23.4% Speed(reviews/sec):1501. #Correct:207 #Tested:235 Testing Accuracy:88.0%\r",
      "Progress:23.5% Speed(reviews/sec):1504. #Correct:208 #Tested:236 Testing Accuracy:88.1%\r",
      "Progress:23.6% Speed(reviews/sec):1502. #Correct:209 #Tested:237 Testing Accuracy:88.1%\r",
      "Progress:23.7% Speed(reviews/sec):1503. #Correct:210 #Tested:238 Testing Accuracy:88.2%\r",
      "Progress:23.8% Speed(reviews/sec):1506. #Correct:211 #Tested:239 Testing Accuracy:88.2%\r",
      "Progress:23.9% Speed(reviews/sec):1508. #Correct:212 #Tested:240 Testing Accuracy:88.3%\r",
      "Progress:24.0% Speed(reviews/sec):1506. #Correct:213 #Tested:241 Testing Accuracy:88.3%\r",
      "Progress:24.1% Speed(reviews/sec):1508. #Correct:213 #Tested:242 Testing Accuracy:88.0%\r",
      "Progress:24.2% Speed(reviews/sec):1509. #Correct:214 #Tested:243 Testing Accuracy:88.0%\r",
      "Progress:24.3% Speed(reviews/sec):1508. #Correct:215 #Tested:244 Testing Accuracy:88.1%\r",
      "Progress:24.4% Speed(reviews/sec):1510. #Correct:216 #Tested:245 Testing Accuracy:88.1%\r",
      "Progress:24.5% Speed(reviews/sec):1512. #Correct:217 #Tested:246 Testing Accuracy:88.2%\r",
      "Progress:24.6% Speed(reviews/sec):1516. #Correct:218 #Tested:247 Testing Accuracy:88.2%\r",
      "Progress:24.7% Speed(reviews/sec):1518. #Correct:219 #Tested:248 Testing Accuracy:88.3%\r",
      "Progress:24.8% Speed(reviews/sec):1522. #Correct:220 #Tested:249 Testing Accuracy:88.3%\r",
      "Progress:24.9% Speed(reviews/sec):1525. #Correct:220 #Tested:250 Testing Accuracy:88.0%\r",
      "Progress:25.0% Speed(reviews/sec):1528. #Correct:221 #Tested:251 Testing Accuracy:88.0%\r",
      "Progress:25.1% Speed(reviews/sec):1531. #Correct:222 #Tested:252 Testing Accuracy:88.0%\r",
      "Progress:25.2% Speed(reviews/sec):1534. #Correct:223 #Tested:253 Testing Accuracy:88.1%\r",
      "Progress:25.3% Speed(reviews/sec):1536. #Correct:224 #Tested:254 Testing Accuracy:88.1%\r",
      "Progress:25.4% Speed(reviews/sec):1540. #Correct:225 #Tested:255 Testing Accuracy:88.2%\r",
      "Progress:25.5% Speed(reviews/sec):1543. #Correct:226 #Tested:256 Testing Accuracy:88.2%\r",
      "Progress:25.6% Speed(reviews/sec):1547. #Correct:227 #Tested:257 Testing Accuracy:88.3%\r",
      "Progress:25.7% Speed(reviews/sec):1551. #Correct:228 #Tested:258 Testing Accuracy:88.3%\r",
      "Progress:25.8% Speed(reviews/sec):1542. #Correct:229 #Tested:259 Testing Accuracy:88.4%\r",
      "Progress:25.9% Speed(reviews/sec):1545. #Correct:229 #Tested:260 Testing Accuracy:88.0%\r",
      "Progress:26.0% Speed(reviews/sec):1548. #Correct:230 #Tested:261 Testing Accuracy:88.1%\r",
      "Progress:26.1% Speed(reviews/sec):1552. #Correct:231 #Tested:262 Testing Accuracy:88.1%\r",
      "Progress:26.2% Speed(reviews/sec):1552. #Correct:232 #Tested:263 Testing Accuracy:88.2%\r",
      "Progress:26.3% Speed(reviews/sec):1549. #Correct:233 #Tested:264 Testing Accuracy:88.2%\r",
      "Progress:26.4% Speed(reviews/sec):1523. #Correct:234 #Tested:265 Testing Accuracy:88.3%\r",
      "Progress:26.5% Speed(reviews/sec):1516. #Correct:235 #Tested:266 Testing Accuracy:88.3%\r",
      "Progress:26.6% Speed(reviews/sec):1518. #Correct:236 #Tested:267 Testing Accuracy:88.3%\r",
      "Progress:26.7% Speed(reviews/sec):1519. #Correct:236 #Tested:268 Testing Accuracy:88.0%\r",
      "Progress:26.8% Speed(reviews/sec):1521. #Correct:237 #Tested:269 Testing Accuracy:88.1%\r",
      "Progress:26.9% Speed(reviews/sec):1524. #Correct:237 #Tested:270 Testing Accuracy:87.7%\r",
      "Progress:27.0% Speed(reviews/sec):1525. #Correct:237 #Tested:271 Testing Accuracy:87.4%\r",
      "Progress:27.1% Speed(reviews/sec):1527. #Correct:238 #Tested:272 Testing Accuracy:87.5%\r",
      "Progress:27.2% Speed(reviews/sec):1526. #Correct:239 #Tested:273 Testing Accuracy:87.5%\r",
      "Progress:27.3% Speed(reviews/sec):1529. #Correct:240 #Tested:274 Testing Accuracy:87.5%\r",
      "Progress:27.4% Speed(reviews/sec):1529. #Correct:240 #Tested:275 Testing Accuracy:87.2%\r",
      "Progress:27.5% Speed(reviews/sec):1527. #Correct:241 #Tested:276 Testing Accuracy:87.3%\r",
      "Progress:27.6% Speed(reviews/sec):1529. #Correct:242 #Tested:277 Testing Accuracy:87.3%\r",
      "Progress:27.7% Speed(reviews/sec):1532. #Correct:243 #Tested:278 Testing Accuracy:87.4%\r",
      "Progress:27.8% Speed(reviews/sec):1535. #Correct:244 #Tested:279 Testing Accuracy:87.4%\r",
      "Progress:27.9% Speed(reviews/sec):1538. #Correct:245 #Tested:280 Testing Accuracy:87.5%\r",
      "Progress:28.0% Speed(reviews/sec):1540. #Correct:246 #Tested:281 Testing Accuracy:87.5%\r",
      "Progress:28.1% Speed(reviews/sec):1543. #Correct:247 #Tested:282 Testing Accuracy:87.5%\r",
      "Progress:28.2% Speed(reviews/sec):1547. #Correct:248 #Tested:283 Testing Accuracy:87.6%\r",
      "Progress:28.3% Speed(reviews/sec):1550. #Correct:249 #Tested:284 Testing Accuracy:87.6%\r",
      "Progress:28.4% Speed(reviews/sec):1544. #Correct:250 #Tested:285 Testing Accuracy:87.7%\r",
      "Progress:28.5% Speed(reviews/sec):1546. #Correct:251 #Tested:286 Testing Accuracy:87.7%\r",
      "Progress:28.6% Speed(reviews/sec):1549. #Correct:252 #Tested:287 Testing Accuracy:87.8%\r",
      "Progress:28.7% Speed(reviews/sec):1553. #Correct:253 #Tested:288 Testing Accuracy:87.8%\r",
      "Progress:28.8% Speed(reviews/sec):1556. #Correct:254 #Tested:289 Testing Accuracy:87.8%\r",
      "Progress:28.9% Speed(reviews/sec):1560. #Correct:255 #Tested:290 Testing Accuracy:87.9%\r",
      "Progress:29.0% Speed(reviews/sec):1562. #Correct:256 #Tested:291 Testing Accuracy:87.9%\r",
      "Progress:29.1% Speed(reviews/sec):1565. #Correct:257 #Tested:292 Testing Accuracy:88.0%\r",
      "Progress:29.2% Speed(reviews/sec):1568. #Correct:258 #Tested:293 Testing Accuracy:88.0%\r",
      "Progress:29.3% Speed(reviews/sec):1571. #Correct:259 #Tested:294 Testing Accuracy:88.0%\r",
      "Progress:29.4% Speed(reviews/sec):1555. #Correct:260 #Tested:295 Testing Accuracy:88.1%\r",
      "Progress:29.5% Speed(reviews/sec):1557. #Correct:261 #Tested:296 Testing Accuracy:88.1%\r",
      "Progress:29.6% Speed(reviews/sec):1558. #Correct:262 #Tested:297 Testing Accuracy:88.2%\r",
      "Progress:29.7% Speed(reviews/sec):1556. #Correct:262 #Tested:298 Testing Accuracy:87.9%\r",
      "Progress:29.8% Speed(reviews/sec):1557. #Correct:263 #Tested:299 Testing Accuracy:87.9%\r",
      "Progress:29.9% Speed(reviews/sec):1558. #Correct:264 #Tested:300 Testing Accuracy:88.0%\r",
      "Progress:30.0% Speed(reviews/sec):1559. #Correct:265 #Tested:301 Testing Accuracy:88.0%\r",
      "Progress:30.1% Speed(reviews/sec):1557. #Correct:266 #Tested:302 Testing Accuracy:88.0%\r",
      "Progress:30.2% Speed(reviews/sec):1551. #Correct:267 #Tested:303 Testing Accuracy:88.1%\r",
      "Progress:30.3% Speed(reviews/sec):1552. #Correct:268 #Tested:304 Testing Accuracy:88.1%\r",
      "Progress:30.4% Speed(reviews/sec):1556. #Correct:269 #Tested:305 Testing Accuracy:88.1%\r",
      "Progress:30.5% Speed(reviews/sec):1558. #Correct:269 #Tested:306 Testing Accuracy:87.9%\r",
      "Progress:30.6% Speed(reviews/sec):1552. #Correct:270 #Tested:307 Testing Accuracy:87.9%\r",
      "Progress:30.7% Speed(reviews/sec):1555. #Correct:270 #Tested:308 Testing Accuracy:87.6%\r",
      "Progress:30.8% Speed(reviews/sec):1556. #Correct:271 #Tested:309 Testing Accuracy:87.7%\r",
      "Progress:30.9% Speed(reviews/sec):1553. #Correct:271 #Tested:310 Testing Accuracy:87.4%\r",
      "Progress:31.0% Speed(reviews/sec):1556. #Correct:272 #Tested:311 Testing Accuracy:87.4%\r",
      "Progress:31.1% Speed(reviews/sec):1557. #Correct:273 #Tested:312 Testing Accuracy:87.5%\r",
      "Progress:31.2% Speed(reviews/sec):1560. #Correct:274 #Tested:313 Testing Accuracy:87.5%\r",
      "Progress:31.3% Speed(reviews/sec):1560. #Correct:275 #Tested:314 Testing Accuracy:87.5%\r",
      "Progress:31.4% Speed(reviews/sec):1563. #Correct:276 #Tested:315 Testing Accuracy:87.6%\r",
      "Progress:31.5% Speed(reviews/sec):1564. #Correct:277 #Tested:316 Testing Accuracy:87.6%\r",
      "Progress:31.6% Speed(reviews/sec):1567. #Correct:278 #Tested:317 Testing Accuracy:87.6%\r",
      "Progress:31.7% Speed(reviews/sec):1567. #Correct:278 #Tested:318 Testing Accuracy:87.4%\r",
      "Progress:31.8% Speed(reviews/sec):1569. #Correct:279 #Tested:319 Testing Accuracy:87.4%\r",
      "Progress:31.9% Speed(reviews/sec):1571. #Correct:280 #Tested:320 Testing Accuracy:87.5%\r",
      "Progress:32.0% Speed(reviews/sec):1574. #Correct:281 #Tested:321 Testing Accuracy:87.5%\r",
      "Progress:32.1% Speed(reviews/sec):1577. #Correct:281 #Tested:322 Testing Accuracy:87.2%\r",
      "Progress:32.2% Speed(reviews/sec):1580. #Correct:282 #Tested:323 Testing Accuracy:87.3%\r",
      "Progress:32.3% Speed(reviews/sec):1581. #Correct:283 #Tested:324 Testing Accuracy:87.3%\r",
      "Progress:32.4% Speed(reviews/sec):1580. #Correct:284 #Tested:325 Testing Accuracy:87.3%\r",
      "Progress:32.5% Speed(reviews/sec):1582. #Correct:285 #Tested:326 Testing Accuracy:87.4%\r",
      "Progress:32.6% Speed(reviews/sec):1584. #Correct:286 #Tested:327 Testing Accuracy:87.4%\r",
      "Progress:32.7% Speed(reviews/sec):1577. #Correct:287 #Tested:328 Testing Accuracy:87.5%\r",
      "Progress:32.8% Speed(reviews/sec):1576. #Correct:288 #Tested:329 Testing Accuracy:87.5%\r",
      "Progress:32.9% Speed(reviews/sec):1578. #Correct:289 #Tested:330 Testing Accuracy:87.5%\r",
      "Progress:33.0% Speed(reviews/sec):1580. #Correct:290 #Tested:331 Testing Accuracy:87.6%\r",
      "Progress:33.1% Speed(reviews/sec):1582. #Correct:291 #Tested:332 Testing Accuracy:87.6%\r",
      "Progress:33.2% Speed(reviews/sec):1582. #Correct:292 #Tested:333 Testing Accuracy:87.6%\r",
      "Progress:33.3% Speed(reviews/sec):1584. #Correct:293 #Tested:334 Testing Accuracy:87.7%\r",
      "Progress:33.4% Speed(reviews/sec):1580. #Correct:294 #Tested:335 Testing Accuracy:87.7%\r",
      "Progress:33.5% Speed(reviews/sec):1582. #Correct:295 #Tested:336 Testing Accuracy:87.7%\r",
      "Progress:33.6% Speed(reviews/sec):1584. #Correct:296 #Tested:337 Testing Accuracy:87.8%\r",
      "Progress:33.7% Speed(reviews/sec):1587. #Correct:297 #Tested:338 Testing Accuracy:87.8%\r",
      "Progress:33.8% Speed(reviews/sec):1588. #Correct:298 #Tested:339 Testing Accuracy:87.9%\r",
      "Progress:33.9% Speed(reviews/sec):1590. #Correct:299 #Tested:340 Testing Accuracy:87.9%\r",
      "Progress:34.0% Speed(reviews/sec):1589. #Correct:300 #Tested:341 Testing Accuracy:87.9%\r",
      "Progress:34.1% Speed(reviews/sec):1591. #Correct:301 #Tested:342 Testing Accuracy:88.0%\r",
      "Progress:34.2% Speed(reviews/sec):1586. #Correct:302 #Tested:343 Testing Accuracy:88.0%\r",
      "Progress:34.3% Speed(reviews/sec):1589. #Correct:303 #Tested:344 Testing Accuracy:88.0%\r",
      "Progress:34.4% Speed(reviews/sec):1588. #Correct:304 #Tested:345 Testing Accuracy:88.1%\r",
      "Progress:34.5% Speed(reviews/sec):1590. #Correct:305 #Tested:346 Testing Accuracy:88.1%\r",
      "Progress:34.6% Speed(reviews/sec):1592. #Correct:305 #Tested:347 Testing Accuracy:87.8%\r",
      "Progress:34.7% Speed(reviews/sec):1594. #Correct:306 #Tested:348 Testing Accuracy:87.9%\r",
      "Progress:34.8% Speed(reviews/sec):1595. #Correct:307 #Tested:349 Testing Accuracy:87.9%\r",
      "Progress:34.9% Speed(reviews/sec):1597. #Correct:308 #Tested:350 Testing Accuracy:88.0%\r",
      "Progress:35.0% Speed(reviews/sec):1599. #Correct:309 #Tested:351 Testing Accuracy:88.0%\r",
      "Progress:35.1% Speed(reviews/sec):1600. #Correct:310 #Tested:352 Testing Accuracy:88.0%\r",
      "Progress:35.2% Speed(reviews/sec):1601. #Correct:311 #Tested:353 Testing Accuracy:88.1%\r",
      "Progress:35.3% Speed(reviews/sec):1603. #Correct:311 #Tested:354 Testing Accuracy:87.8%\r",
      "Progress:35.4% Speed(reviews/sec):1605. #Correct:312 #Tested:355 Testing Accuracy:87.8%\r",
      "Progress:35.5% Speed(reviews/sec):1602. #Correct:313 #Tested:356 Testing Accuracy:87.9%\r",
      "Progress:35.6% Speed(reviews/sec):1603. #Correct:314 #Tested:357 Testing Accuracy:87.9%"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\r",
      "Progress:35.7% Speed(reviews/sec):1597. #Correct:314 #Tested:358 Testing Accuracy:87.7%\r",
      "Progress:35.8% Speed(reviews/sec):1598. #Correct:315 #Tested:359 Testing Accuracy:87.7%\r",
      "Progress:35.9% Speed(reviews/sec):1597. #Correct:315 #Tested:360 Testing Accuracy:87.5%\r",
      "Progress:36.0% Speed(reviews/sec):1597. #Correct:316 #Tested:361 Testing Accuracy:87.5%\r",
      "Progress:36.1% Speed(reviews/sec):1599. #Correct:317 #Tested:362 Testing Accuracy:87.5%\r",
      "Progress:36.2% Speed(reviews/sec):1596. #Correct:318 #Tested:363 Testing Accuracy:87.6%\r",
      "Progress:36.3% Speed(reviews/sec):1597. #Correct:319 #Tested:364 Testing Accuracy:87.6%\r",
      "Progress:36.4% Speed(reviews/sec):1599. #Correct:320 #Tested:365 Testing Accuracy:87.6%\r",
      "Progress:36.5% Speed(reviews/sec):1600. #Correct:321 #Tested:366 Testing Accuracy:87.7%\r",
      "Progress:36.6% Speed(reviews/sec):1601. #Correct:322 #Tested:367 Testing Accuracy:87.7%\r",
      "Progress:36.7% Speed(reviews/sec):1603. #Correct:323 #Tested:368 Testing Accuracy:87.7%\r",
      "Progress:36.8% Speed(reviews/sec):1604. #Correct:324 #Tested:369 Testing Accuracy:87.8%\r",
      "Progress:36.9% Speed(reviews/sec):1605. #Correct:325 #Tested:370 Testing Accuracy:87.8%\r",
      "Progress:37.0% Speed(reviews/sec):1606. #Correct:326 #Tested:371 Testing Accuracy:87.8%\r",
      "Progress:37.1% Speed(reviews/sec):1603. #Correct:326 #Tested:372 Testing Accuracy:87.6%\r",
      "Progress:37.2% Speed(reviews/sec):1604. #Correct:326 #Tested:373 Testing Accuracy:87.3%\r",
      "Progress:37.3% Speed(reviews/sec):1606. #Correct:327 #Tested:374 Testing Accuracy:87.4%\r",
      "Progress:37.4% Speed(reviews/sec):1606. #Correct:328 #Tested:375 Testing Accuracy:87.4%\r",
      "Progress:37.5% Speed(reviews/sec):1606. #Correct:328 #Tested:376 Testing Accuracy:87.2%\r",
      "Progress:37.6% Speed(reviews/sec):1609. #Correct:329 #Tested:377 Testing Accuracy:87.2%\r",
      "Progress:37.7% Speed(reviews/sec):1610. #Correct:330 #Tested:378 Testing Accuracy:87.3%\r",
      "Progress:37.8% Speed(reviews/sec):1612. #Correct:331 #Tested:379 Testing Accuracy:87.3%\r",
      "Progress:37.9% Speed(reviews/sec):1614. #Correct:332 #Tested:380 Testing Accuracy:87.3%\r",
      "Progress:38.0% Speed(reviews/sec):1611. #Correct:333 #Tested:381 Testing Accuracy:87.4%\r",
      "Progress:38.1% Speed(reviews/sec):1613. #Correct:334 #Tested:382 Testing Accuracy:87.4%\r",
      "Progress:38.2% Speed(reviews/sec):1613. #Correct:335 #Tested:383 Testing Accuracy:87.4%\r",
      "Progress:38.3% Speed(reviews/sec):1614. #Correct:336 #Tested:384 Testing Accuracy:87.5%\r",
      "Progress:38.4% Speed(reviews/sec):1615. #Correct:336 #Tested:385 Testing Accuracy:87.2%\r",
      "Progress:38.5% Speed(reviews/sec):1617. #Correct:337 #Tested:386 Testing Accuracy:87.3%\r",
      "Progress:38.6% Speed(reviews/sec):1617. #Correct:338 #Tested:387 Testing Accuracy:87.3%\r",
      "Progress:38.7% Speed(reviews/sec):1620. #Correct:338 #Tested:388 Testing Accuracy:87.1%\r",
      "Progress:38.8% Speed(reviews/sec):1621. #Correct:338 #Tested:389 Testing Accuracy:86.8%\r",
      "Progress:38.9% Speed(reviews/sec):1621. #Correct:339 #Tested:390 Testing Accuracy:86.9%\r",
      "Progress:39.0% Speed(reviews/sec):1622. #Correct:340 #Tested:391 Testing Accuracy:86.9%\r",
      "Progress:39.1% Speed(reviews/sec):1623. #Correct:340 #Tested:392 Testing Accuracy:86.7%\r",
      "Progress:39.2% Speed(reviews/sec):1625. #Correct:341 #Tested:393 Testing Accuracy:86.7%\r",
      "Progress:39.3% Speed(reviews/sec):1625. #Correct:342 #Tested:394 Testing Accuracy:86.8%\r",
      "Progress:39.4% Speed(reviews/sec):1625. #Correct:343 #Tested:395 Testing Accuracy:86.8%\r",
      "Progress:39.5% Speed(reviews/sec):1626. #Correct:344 #Tested:396 Testing Accuracy:86.8%\r",
      "Progress:39.6% Speed(reviews/sec):1627. #Correct:345 #Tested:397 Testing Accuracy:86.9%\r",
      "Progress:39.7% Speed(reviews/sec):1629. #Correct:346 #Tested:398 Testing Accuracy:86.9%\r",
      "Progress:39.8% Speed(reviews/sec):1628. #Correct:347 #Tested:399 Testing Accuracy:86.9%\r",
      "Progress:39.9% Speed(reviews/sec):1630. #Correct:348 #Tested:400 Testing Accuracy:87.0%\r",
      "Progress:40.0% Speed(reviews/sec):1628. #Correct:349 #Tested:401 Testing Accuracy:87.0%\r",
      "Progress:40.1% Speed(reviews/sec):1631. #Correct:350 #Tested:402 Testing Accuracy:87.0%\r",
      "Progress:40.2% Speed(reviews/sec):1631. #Correct:351 #Tested:403 Testing Accuracy:87.0%\r",
      "Progress:40.3% Speed(reviews/sec):1633. #Correct:351 #Tested:404 Testing Accuracy:86.8%\r",
      "Progress:40.4% Speed(reviews/sec):1635. #Correct:352 #Tested:405 Testing Accuracy:86.9%\r",
      "Progress:40.5% Speed(reviews/sec):1636. #Correct:353 #Tested:406 Testing Accuracy:86.9%\r",
      "Progress:40.6% Speed(reviews/sec):1637. #Correct:354 #Tested:407 Testing Accuracy:86.9%\r",
      "Progress:40.7% Speed(reviews/sec):1638. #Correct:355 #Tested:408 Testing Accuracy:87.0%\r",
      "Progress:40.8% Speed(reviews/sec):1639. #Correct:356 #Tested:409 Testing Accuracy:87.0%\r",
      "Progress:40.9% Speed(reviews/sec):1641. #Correct:357 #Tested:410 Testing Accuracy:87.0%\r",
      "Progress:41.0% Speed(reviews/sec):1604. #Correct:358 #Tested:411 Testing Accuracy:87.1%\r",
      "Progress:41.1% Speed(reviews/sec):1598. #Correct:359 #Tested:412 Testing Accuracy:87.1%\r",
      "Progress:41.2% Speed(reviews/sec):1598. #Correct:360 #Tested:413 Testing Accuracy:87.1%\r",
      "Progress:41.3% Speed(reviews/sec):1595. #Correct:361 #Tested:414 Testing Accuracy:87.1%\r",
      "Progress:41.4% Speed(reviews/sec):1597. #Correct:362 #Tested:415 Testing Accuracy:87.2%\r",
      "Progress:41.5% Speed(reviews/sec):1597. #Correct:363 #Tested:416 Testing Accuracy:87.2%\r",
      "Progress:41.6% Speed(reviews/sec):1576. #Correct:364 #Tested:417 Testing Accuracy:87.2%\r",
      "Progress:41.7% Speed(reviews/sec):1576. #Correct:365 #Tested:418 Testing Accuracy:87.3%\r",
      "Progress:41.8% Speed(reviews/sec):1577. #Correct:366 #Tested:419 Testing Accuracy:87.3%\r",
      "Progress:41.9% Speed(reviews/sec):1574. #Correct:367 #Tested:420 Testing Accuracy:87.3%\r",
      "Progress:42.0% Speed(reviews/sec):1571. #Correct:368 #Tested:421 Testing Accuracy:87.4%\r",
      "Progress:42.1% Speed(reviews/sec):1572. #Correct:369 #Tested:422 Testing Accuracy:87.4%\r",
      "Progress:42.2% Speed(reviews/sec):1573. #Correct:370 #Tested:423 Testing Accuracy:87.4%\r",
      "Progress:42.3% Speed(reviews/sec):1574. #Correct:371 #Tested:424 Testing Accuracy:87.5%\r",
      "Progress:42.4% Speed(reviews/sec):1575. #Correct:372 #Tested:425 Testing Accuracy:87.5%\r",
      "Progress:42.5% Speed(reviews/sec):1577. #Correct:373 #Tested:426 Testing Accuracy:87.5%\r",
      "Progress:42.6% Speed(reviews/sec):1575. #Correct:374 #Tested:427 Testing Accuracy:87.5%\r",
      "Progress:42.7% Speed(reviews/sec):1575. #Correct:375 #Tested:428 Testing Accuracy:87.6%\r",
      "Progress:42.8% Speed(reviews/sec):1576. #Correct:376 #Tested:429 Testing Accuracy:87.6%\r",
      "Progress:42.9% Speed(reviews/sec):1577. #Correct:377 #Tested:430 Testing Accuracy:87.6%\r",
      "Progress:43.0% Speed(reviews/sec):1578. #Correct:378 #Tested:431 Testing Accuracy:87.7%\r",
      "Progress:43.1% Speed(reviews/sec):1578. #Correct:379 #Tested:432 Testing Accuracy:87.7%\r",
      "Progress:43.2% Speed(reviews/sec):1575. #Correct:380 #Tested:433 Testing Accuracy:87.7%\r",
      "Progress:43.3% Speed(reviews/sec):1577. #Correct:381 #Tested:434 Testing Accuracy:87.7%\r",
      "Progress:43.4% Speed(reviews/sec):1577. #Correct:382 #Tested:435 Testing Accuracy:87.8%\r",
      "Progress:43.5% Speed(reviews/sec):1578. #Correct:383 #Tested:436 Testing Accuracy:87.8%\r",
      "Progress:43.6% Speed(reviews/sec):1580. #Correct:384 #Tested:437 Testing Accuracy:87.8%\r",
      "Progress:43.7% Speed(reviews/sec):1578. #Correct:385 #Tested:438 Testing Accuracy:87.8%\r",
      "Progress:43.8% Speed(reviews/sec):1579. #Correct:386 #Tested:439 Testing Accuracy:87.9%\r",
      "Progress:43.9% Speed(reviews/sec):1579. #Correct:387 #Tested:440 Testing Accuracy:87.9%\r",
      "Progress:44.0% Speed(reviews/sec):1578. #Correct:388 #Tested:441 Testing Accuracy:87.9%\r",
      "Progress:44.1% Speed(reviews/sec):1579. #Correct:388 #Tested:442 Testing Accuracy:87.7%\r",
      "Progress:44.2% Speed(reviews/sec):1581. #Correct:389 #Tested:443 Testing Accuracy:87.8%\r",
      "Progress:44.3% Speed(reviews/sec):1581. #Correct:390 #Tested:444 Testing Accuracy:87.8%\r",
      "Progress:44.4% Speed(reviews/sec):1577. #Correct:391 #Tested:445 Testing Accuracy:87.8%\r",
      "Progress:44.5% Speed(reviews/sec):1573. #Correct:392 #Tested:446 Testing Accuracy:87.8%\r",
      "Progress:44.6% Speed(reviews/sec):1559. #Correct:393 #Tested:447 Testing Accuracy:87.9%\r",
      "Progress:44.7% Speed(reviews/sec):1541. #Correct:394 #Tested:448 Testing Accuracy:87.9%\r",
      "Progress:44.8% Speed(reviews/sec):1532. #Correct:395 #Tested:449 Testing Accuracy:87.9%\r",
      "Progress:44.9% Speed(reviews/sec):1529. #Correct:396 #Tested:450 Testing Accuracy:88.0%\r",
      "Progress:45.0% Speed(reviews/sec):1531. #Correct:397 #Tested:451 Testing Accuracy:88.0%\r",
      "Progress:45.1% Speed(reviews/sec):1528. #Correct:398 #Tested:452 Testing Accuracy:88.0%\r",
      "Progress:45.2% Speed(reviews/sec):1529. #Correct:398 #Tested:453 Testing Accuracy:87.8%\r",
      "Progress:45.3% Speed(reviews/sec):1531. #Correct:399 #Tested:454 Testing Accuracy:87.8%\r",
      "Progress:45.4% Speed(reviews/sec):1534. #Correct:400 #Tested:455 Testing Accuracy:87.9%\r",
      "Progress:45.5% Speed(reviews/sec):1535. #Correct:401 #Tested:456 Testing Accuracy:87.9%\r",
      "Progress:45.6% Speed(reviews/sec):1537. #Correct:402 #Tested:457 Testing Accuracy:87.9%\r",
      "Progress:45.7% Speed(reviews/sec):1538. #Correct:403 #Tested:458 Testing Accuracy:87.9%\r",
      "Progress:45.8% Speed(reviews/sec):1539. #Correct:404 #Tested:459 Testing Accuracy:88.0%\r",
      "Progress:45.9% Speed(reviews/sec):1537. #Correct:404 #Tested:460 Testing Accuracy:87.8%\r",
      "Progress:46.0% Speed(reviews/sec):1539. #Correct:405 #Tested:461 Testing Accuracy:87.8%\r",
      "Progress:46.1% Speed(reviews/sec):1539. #Correct:406 #Tested:462 Testing Accuracy:87.8%\r",
      "Progress:46.2% Speed(reviews/sec):1535. #Correct:407 #Tested:463 Testing Accuracy:87.9%\r",
      "Progress:46.3% Speed(reviews/sec):1536. #Correct:407 #Tested:464 Testing Accuracy:87.7%\r",
      "Progress:46.4% Speed(reviews/sec):1538. #Correct:408 #Tested:465 Testing Accuracy:87.7%\r",
      "Progress:46.5% Speed(reviews/sec):1540. #Correct:408 #Tested:466 Testing Accuracy:87.5%\r",
      "Progress:46.6% Speed(reviews/sec):1541. #Correct:409 #Tested:467 Testing Accuracy:87.5%\r",
      "Progress:46.7% Speed(reviews/sec):1539. #Correct:410 #Tested:468 Testing Accuracy:87.6%\r",
      "Progress:46.8% Speed(reviews/sec):1537. #Correct:411 #Tested:469 Testing Accuracy:87.6%\r",
      "Progress:46.9% Speed(reviews/sec):1539. #Correct:411 #Tested:470 Testing Accuracy:87.4%\r",
      "Progress:47.0% Speed(reviews/sec):1540. #Correct:412 #Tested:471 Testing Accuracy:87.4%\r",
      "Progress:47.1% Speed(reviews/sec):1535. #Correct:413 #Tested:472 Testing Accuracy:87.5%\r",
      "Progress:47.2% Speed(reviews/sec):1531. #Correct:414 #Tested:473 Testing Accuracy:87.5%\r",
      "Progress:47.3% Speed(reviews/sec):1524. #Correct:415 #Tested:474 Testing Accuracy:87.5%\r",
      "Progress:47.4% Speed(reviews/sec):1524. #Correct:416 #Tested:475 Testing Accuracy:87.5%\r",
      "Progress:47.5% Speed(reviews/sec):1523. #Correct:417 #Tested:476 Testing Accuracy:87.6%\r",
      "Progress:47.6% Speed(reviews/sec):1514. #Correct:418 #Tested:477 Testing Accuracy:87.6%\r",
      "Progress:47.7% Speed(reviews/sec):1511. #Correct:419 #Tested:478 Testing Accuracy:87.6%\r",
      "Progress:47.8% Speed(reviews/sec):1508. #Correct:420 #Tested:479 Testing Accuracy:87.6%\r",
      "Progress:47.9% Speed(reviews/sec):1509. #Correct:421 #Tested:480 Testing Accuracy:87.7%\r",
      "Progress:48.0% Speed(reviews/sec):1510. #Correct:422 #Tested:481 Testing Accuracy:87.7%\r",
      "Progress:48.1% Speed(reviews/sec):1506. #Correct:423 #Tested:482 Testing Accuracy:87.7%\r",
      "Progress:48.2% Speed(reviews/sec):1507. #Correct:424 #Tested:483 Testing Accuracy:87.7%\r",
      "Progress:48.3% Speed(reviews/sec):1508. #Correct:425 #Tested:484 Testing Accuracy:87.8%\r",
      "Progress:48.4% Speed(reviews/sec):1508. #Correct:426 #Tested:485 Testing Accuracy:87.8%\r",
      "Progress:48.5% Speed(reviews/sec):1502. #Correct:427 #Tested:486 Testing Accuracy:87.8%\r",
      "Progress:48.6% Speed(reviews/sec):1491. #Correct:428 #Tested:487 Testing Accuracy:87.8%\r",
      "Progress:48.7% Speed(reviews/sec):1491. #Correct:428 #Tested:488 Testing Accuracy:87.7%\r",
      "Progress:48.8% Speed(reviews/sec):1492. #Correct:429 #Tested:489 Testing Accuracy:87.7%\r",
      "Progress:48.9% Speed(reviews/sec):1489. #Correct:429 #Tested:490 Testing Accuracy:87.5%\r",
      "Progress:49.0% Speed(reviews/sec):1490. #Correct:430 #Tested:491 Testing Accuracy:87.5%\r",
      "Progress:49.1% Speed(reviews/sec):1491. #Correct:431 #Tested:492 Testing Accuracy:87.6%\r",
      "Progress:49.2% Speed(reviews/sec):1491. #Correct:432 #Tested:493 Testing Accuracy:87.6%\r",
      "Progress:49.3% Speed(reviews/sec):1484. #Correct:433 #Tested:494 Testing Accuracy:87.6%\r",
      "Progress:49.4% Speed(reviews/sec):1485. #Correct:434 #Tested:495 Testing Accuracy:87.6%\r",
      "Progress:49.5% Speed(reviews/sec):1485. #Correct:435 #Tested:496 Testing Accuracy:87.7%\r",
      "Progress:49.6% Speed(reviews/sec):1486. #Correct:436 #Tested:497 Testing Accuracy:87.7%\r",
      "Progress:49.7% Speed(reviews/sec):1487. #Correct:436 #Tested:498 Testing Accuracy:87.5%\r",
      "Progress:49.8% Speed(reviews/sec):1486. #Correct:437 #Tested:499 Testing Accuracy:87.5%\r",
      "Progress:49.9% Speed(reviews/sec):1486. #Correct:438 #Tested:500 Testing Accuracy:87.6%\r",
      "Progress:50.0% Speed(reviews/sec):1488. #Correct:439 #Tested:501 Testing Accuracy:87.6%\r",
      "Progress:50.1% Speed(reviews/sec):1487. #Correct:440 #Tested:502 Testing Accuracy:87.6%\r",
      "Progress:50.2% Speed(reviews/sec):1486. #Correct:441 #Tested:503 Testing Accuracy:87.6%\r",
      "Progress:50.3% Speed(reviews/sec):1485. #Correct:442 #Tested:504 Testing Accuracy:87.6%\r",
      "Progress:50.4% Speed(reviews/sec):1482. #Correct:443 #Tested:505 Testing Accuracy:87.7%\r",
      "Progress:50.5% Speed(reviews/sec):1481. #Correct:444 #Tested:506 Testing Accuracy:87.7%\r",
      "Progress:50.6% Speed(reviews/sec):1472. #Correct:445 #Tested:507 Testing Accuracy:87.7%\r",
      "Progress:50.7% Speed(reviews/sec):1468. #Correct:446 #Tested:508 Testing Accuracy:87.7%\r",
      "Progress:50.8% Speed(reviews/sec):1469. #Correct:447 #Tested:509 Testing Accuracy:87.8%\r",
      "Progress:50.9% Speed(reviews/sec):1467. #Correct:448 #Tested:510 Testing Accuracy:87.8%\r",
      "Progress:51.0% Speed(reviews/sec):1462. #Correct:449 #Tested:511 Testing Accuracy:87.8%\r",
      "Progress:51.1% Speed(reviews/sec):1462. #Correct:450 #Tested:512 Testing Accuracy:87.8%\r",
      "Progress:51.2% Speed(reviews/sec):1462. #Correct:450 #Tested:513 Testing Accuracy:87.7%\r",
      "Progress:51.3% Speed(reviews/sec):1464. #Correct:451 #Tested:514 Testing Accuracy:87.7%\r",
      "Progress:51.4% Speed(reviews/sec):1464. #Correct:452 #Tested:515 Testing Accuracy:87.7%\r",
      "Progress:51.5% Speed(reviews/sec):1464. #Correct:453 #Tested:516 Testing Accuracy:87.7%\r",
      "Progress:51.6% Speed(reviews/sec):1464. #Correct:454 #Tested:517 Testing Accuracy:87.8%\r",
      "Progress:51.7% Speed(reviews/sec):1465. #Correct:455 #Tested:518 Testing Accuracy:87.8%\r",
      "Progress:51.8% Speed(reviews/sec):1465. #Correct:455 #Tested:519 Testing Accuracy:87.6%\r",
      "Progress:51.9% Speed(reviews/sec):1461. #Correct:456 #Tested:520 Testing Accuracy:87.6%\r",
      "Progress:52.0% Speed(reviews/sec):1460. #Correct:457 #Tested:521 Testing Accuracy:87.7%\r",
      "Progress:52.1% Speed(reviews/sec):1458. #Correct:458 #Tested:522 Testing Accuracy:87.7%\r",
      "Progress:52.2% Speed(reviews/sec):1458. #Correct:459 #Tested:523 Testing Accuracy:87.7%\r",
      "Progress:52.3% Speed(reviews/sec):1459. #Correct:460 #Tested:524 Testing Accuracy:87.7%\r",
      "Progress:52.4% Speed(reviews/sec):1460. #Correct:461 #Tested:525 Testing Accuracy:87.8%\r",
      "Progress:52.5% Speed(reviews/sec):1462. #Correct:462 #Tested:526 Testing Accuracy:87.8%\r",
      "Progress:52.6% Speed(reviews/sec):1463. #Correct:463 #Tested:527 Testing Accuracy:87.8%\r",
      "Progress:52.7% Speed(reviews/sec):1464. #Correct:464 #Tested:528 Testing Accuracy:87.8%\r",
      "Progress:52.8% Speed(reviews/sec):1463. #Correct:465 #Tested:529 Testing Accuracy:87.9%\r",
      "Progress:52.9% Speed(reviews/sec):1464. #Correct:466 #Tested:530 Testing Accuracy:87.9%\r",
      "Progress:53.0% Speed(reviews/sec):1462. #Correct:467 #Tested:531 Testing Accuracy:87.9%\r",
      "Progress:53.1% Speed(reviews/sec):1461. #Correct:467 #Tested:532 Testing Accuracy:87.7%\r",
      "Progress:53.2% Speed(reviews/sec):1459. #Correct:468 #Tested:533 Testing Accuracy:87.8%\r",
      "Progress:53.3% Speed(reviews/sec):1459. #Correct:469 #Tested:534 Testing Accuracy:87.8%\r",
      "Progress:53.4% Speed(reviews/sec):1460. #Correct:470 #Tested:535 Testing Accuracy:87.8%\r",
      "Progress:53.5% Speed(reviews/sec):1462. #Correct:471 #Tested:536 Testing Accuracy:87.8%\r",
      "Progress:53.6% Speed(reviews/sec):1463. #Correct:471 #Tested:537 Testing Accuracy:87.7%\r",
      "Progress:53.7% Speed(reviews/sec):1464. #Correct:472 #Tested:538 Testing Accuracy:87.7%\r",
      "Progress:53.8% Speed(reviews/sec):1465. #Correct:473 #Tested:539 Testing Accuracy:87.7%\r",
      "Progress:53.9% Speed(reviews/sec):1466. #Correct:474 #Tested:540 Testing Accuracy:87.7%\r",
      "Progress:54.0% Speed(reviews/sec):1467. #Correct:475 #Tested:541 Testing Accuracy:87.8%\r",
      "Progress:54.1% Speed(reviews/sec):1468. #Correct:476 #Tested:542 Testing Accuracy:87.8%\r",
      "Progress:54.2% Speed(reviews/sec):1468. #Correct:477 #Tested:543 Testing Accuracy:87.8%\r",
      "Progress:54.3% Speed(reviews/sec):1468. #Correct:478 #Tested:544 Testing Accuracy:87.8%\r",
      "Progress:54.4% Speed(reviews/sec):1465. #Correct:478 #Tested:545 Testing Accuracy:87.7%\r",
      "Progress:54.5% Speed(reviews/sec):1462. #Correct:479 #Tested:546 Testing Accuracy:87.7%\r",
      "Progress:54.6% Speed(reviews/sec):1458. #Correct:480 #Tested:547 Testing Accuracy:87.7%\r",
      "Progress:54.7% Speed(reviews/sec):1458. #Correct:480 #Tested:548 Testing Accuracy:87.5%\r",
      "Progress:54.8% Speed(reviews/sec):1460. #Correct:481 #Tested:549 Testing Accuracy:87.6%\r",
      "Progress:54.9% Speed(reviews/sec):1461. #Correct:482 #Tested:550 Testing Accuracy:87.6%\r",
      "Progress:55.0% Speed(reviews/sec):1462. #Correct:483 #Tested:551 Testing Accuracy:87.6%\r",
      "Progress:55.1% Speed(reviews/sec):1463. #Correct:483 #Tested:552 Testing Accuracy:87.5%\r",
      "Progress:55.2% Speed(reviews/sec):1464. #Correct:484 #Tested:553 Testing Accuracy:87.5%\r",
      "Progress:55.3% Speed(reviews/sec):1465. #Correct:485 #Tested:554 Testing Accuracy:87.5%\r",
      "Progress:55.4% Speed(reviews/sec):1456. #Correct:486 #Tested:555 Testing Accuracy:87.5%\r",
      "Progress:55.5% Speed(reviews/sec):1457. #Correct:487 #Tested:556 Testing Accuracy:87.5%\r",
      "Progress:55.6% Speed(reviews/sec):1457. #Correct:488 #Tested:557 Testing Accuracy:87.6%\r",
      "Progress:55.7% Speed(reviews/sec):1457. #Correct:489 #Tested:558 Testing Accuracy:87.6%\r",
      "Progress:55.8% Speed(reviews/sec):1456. #Correct:489 #Tested:559 Testing Accuracy:87.4%\r",
      "Progress:55.9% Speed(reviews/sec):1457. #Correct:490 #Tested:560 Testing Accuracy:87.5%\r",
      "Progress:56.0% Speed(reviews/sec):1458. #Correct:491 #Tested:561 Testing Accuracy:87.5%\r",
      "Progress:56.1% Speed(reviews/sec):1460. #Correct:492 #Tested:562 Testing Accuracy:87.5%\r",
      "Progress:56.2% Speed(reviews/sec):1460. #Correct:493 #Tested:563 Testing Accuracy:87.5%\r",
      "Progress:56.3% Speed(reviews/sec):1462. #Correct:494 #Tested:564 Testing Accuracy:87.5%\r",
      "Progress:56.4% Speed(reviews/sec):1461. #Correct:495 #Tested:565 Testing Accuracy:87.6%\r",
      "Progress:56.5% Speed(reviews/sec):1461. #Correct:496 #Tested:566 Testing Accuracy:87.6%\r",
      "Progress:56.6% Speed(reviews/sec):1459. #Correct:497 #Tested:567 Testing Accuracy:87.6%\r",
      "Progress:56.7% Speed(reviews/sec):1449. #Correct:498 #Tested:568 Testing Accuracy:87.6%\r",
      "Progress:56.8% Speed(reviews/sec):1450. #Correct:499 #Tested:569 Testing Accuracy:87.6%\r",
      "Progress:56.9% Speed(reviews/sec):1450. #Correct:500 #Tested:570 Testing Accuracy:87.7%\r",
      "Progress:57.0% Speed(reviews/sec):1448. #Correct:500 #Tested:571 Testing Accuracy:87.5%\r",
      "Progress:57.1% Speed(reviews/sec):1445. #Correct:501 #Tested:572 Testing Accuracy:87.5%\r",
      "Progress:57.2% Speed(reviews/sec):1442. #Correct:502 #Tested:573 Testing Accuracy:87.6%\r",
      "Progress:57.3% Speed(reviews/sec):1443. #Correct:503 #Tested:574 Testing Accuracy:87.6%\r",
      "Progress:57.4% Speed(reviews/sec):1442. #Correct:504 #Tested:575 Testing Accuracy:87.6%\r",
      "Progress:57.5% Speed(reviews/sec):1443. #Correct:505 #Tested:576 Testing Accuracy:87.6%\r",
      "Progress:57.6% Speed(reviews/sec):1444. #Correct:506 #Tested:577 Testing Accuracy:87.6%\r",
      "Progress:57.7% Speed(reviews/sec):1446. #Correct:507 #Tested:578 Testing Accuracy:87.7%\r",
      "Progress:57.8% Speed(reviews/sec):1445. #Correct:508 #Tested:579 Testing Accuracy:87.7%\r",
      "Progress:57.9% Speed(reviews/sec):1446. #Correct:509 #Tested:580 Testing Accuracy:87.7%\r",
      "Progress:58.0% Speed(reviews/sec):1446. #Correct:510 #Tested:581 Testing Accuracy:87.7%\r",
      "Progress:58.1% Speed(reviews/sec):1448. #Correct:511 #Tested:582 Testing Accuracy:87.8%\r",
      "Progress:58.2% Speed(reviews/sec):1449. #Correct:512 #Tested:583 Testing Accuracy:87.8%\r",
      "Progress:58.3% Speed(reviews/sec):1449. #Correct:513 #Tested:584 Testing Accuracy:87.8%\r",
      "Progress:58.4% Speed(reviews/sec):1449. #Correct:514 #Tested:585 Testing Accuracy:87.8%\r",
      "Progress:58.5% Speed(reviews/sec):1447. #Correct:515 #Tested:586 Testing Accuracy:87.8%\r",
      "Progress:58.6% Speed(reviews/sec):1448. #Correct:516 #Tested:587 Testing Accuracy:87.9%\r",
      "Progress:58.7% Speed(reviews/sec):1448. #Correct:517 #Tested:588 Testing Accuracy:87.9%\r",
      "Progress:58.8% Speed(reviews/sec):1445. #Correct:518 #Tested:589 Testing Accuracy:87.9%\r",
      "Progress:58.9% Speed(reviews/sec):1446. #Correct:519 #Tested:590 Testing Accuracy:87.9%\r",
      "Progress:59.0% Speed(reviews/sec):1447. #Correct:519 #Tested:591 Testing Accuracy:87.8%\r",
      "Progress:59.1% Speed(reviews/sec):1448. #Correct:520 #Tested:592 Testing Accuracy:87.8%\r",
      "Progress:59.2% Speed(reviews/sec):1447. #Correct:520 #Tested:593 Testing Accuracy:87.6%\r",
      "Progress:59.3% Speed(reviews/sec):1447. #Correct:521 #Tested:594 Testing Accuracy:87.7%\r",
      "Progress:59.4% Speed(reviews/sec):1444. #Correct:522 #Tested:595 Testing Accuracy:87.7%\r",
      "Progress:59.5% Speed(reviews/sec):1442. #Correct:523 #Tested:596 Testing Accuracy:87.7%\r",
      "Progress:59.6% Speed(reviews/sec):1444. #Correct:524 #Tested:597 Testing Accuracy:87.7%\r",
      "Progress:59.7% Speed(reviews/sec):1444. #Correct:525 #Tested:598 Testing Accuracy:87.7%\r",
      "Progress:59.8% Speed(reviews/sec):1445. #Correct:525 #Tested:599 Testing Accuracy:87.6%\r",
      "Progress:59.9% Speed(reviews/sec):1447. #Correct:526 #Tested:600 Testing Accuracy:87.6%\r",
      "Progress:60.0% Speed(reviews/sec):1447. #Correct:526 #Tested:601 Testing Accuracy:87.5%\r",
      "Progress:60.1% Speed(reviews/sec):1447. #Correct:527 #Tested:602 Testing Accuracy:87.5%\r",
      "Progress:60.2% Speed(reviews/sec):1447. #Correct:527 #Tested:603 Testing Accuracy:87.3%\r",
      "Progress:60.3% Speed(reviews/sec):1449. #Correct:528 #Tested:604 Testing Accuracy:87.4%\r",
      "Progress:60.4% Speed(reviews/sec):1449. #Correct:529 #Tested:605 Testing Accuracy:87.4%\r",
      "Progress:60.5% Speed(reviews/sec):1448. #Correct:529 #Tested:606 Testing Accuracy:87.2%\r",
      "Progress:60.6% Speed(reviews/sec):1448. #Correct:529 #Tested:607 Testing Accuracy:87.1%\r",
      "Progress:60.7% Speed(reviews/sec):1444. #Correct:530 #Tested:608 Testing Accuracy:87.1%\r",
      "Progress:60.8% Speed(reviews/sec):1445. #Correct:531 #Tested:609 Testing Accuracy:87.1%\r",
      "Progress:60.9% Speed(reviews/sec):1446. #Correct:532 #Tested:610 Testing Accuracy:87.2%\r",
      "Progress:61.0% Speed(reviews/sec):1443. #Correct:532 #Tested:611 Testing Accuracy:87.0%\r",
      "Progress:61.1% Speed(reviews/sec):1441. #Correct:533 #Tested:612 Testing Accuracy:87.0%\r",
      "Progress:61.2% Speed(reviews/sec):1441. #Correct:534 #Tested:613 Testing Accuracy:87.1%\r",
      "Progress:61.3% Speed(reviews/sec):1442. #Correct:535 #Tested:614 Testing Accuracy:87.1%\r",
      "Progress:61.4% Speed(reviews/sec):1443. #Correct:536 #Tested:615 Testing Accuracy:87.1%\r",
      "Progress:61.5% Speed(reviews/sec):1444. #Correct:536 #Tested:616 Testing Accuracy:87.0%\r",
      "Progress:61.6% Speed(reviews/sec):1442. #Correct:537 #Tested:617 Testing Accuracy:87.0%\r",
      "Progress:61.7% Speed(reviews/sec):1442. #Correct:538 #Tested:618 Testing Accuracy:87.0%\r",
      "Progress:61.8% Speed(reviews/sec):1441. #Correct:539 #Tested:619 Testing Accuracy:87.0%\r",
      "Progress:61.9% Speed(reviews/sec):1442. #Correct:539 #Tested:620 Testing Accuracy:86.9%\r",
      "Progress:62.0% Speed(reviews/sec):1441. #Correct:539 #Tested:621 Testing Accuracy:86.7%\r",
      "Progress:62.1% Speed(reviews/sec):1440. #Correct:540 #Tested:622 Testing Accuracy:86.8%\r",
      "Progress:62.2% Speed(reviews/sec):1440. #Correct:540 #Tested:623 Testing Accuracy:86.6%\r",
      "Progress:62.3% Speed(reviews/sec):1440. #Correct:540 #Tested:624 Testing Accuracy:86.5%\r",
      "Progress:62.4% Speed(reviews/sec):1441. #Correct:541 #Tested:625 Testing Accuracy:86.5%\r",
      "Progress:62.5% Speed(reviews/sec):1435. #Correct:542 #Tested:626 Testing Accuracy:86.5%\r",
      "Progress:62.6% Speed(reviews/sec):1433. #Correct:542 #Tested:627 Testing Accuracy:86.4%\r",
      "Progress:62.7% Speed(reviews/sec):1434. #Correct:543 #Tested:628 Testing Accuracy:86.4%\r",
      "Progress:62.8% Speed(reviews/sec):1434. #Correct:544 #Tested:629 Testing Accuracy:86.4%\r",
      "Progress:62.9% Speed(reviews/sec):1434. #Correct:545 #Tested:630 Testing Accuracy:86.5%\r",
      "Progress:63.0% Speed(reviews/sec):1434. #Correct:545 #Tested:631 Testing Accuracy:86.3%\r",
      "Progress:63.1% Speed(reviews/sec):1429. #Correct:546 #Tested:632 Testing Accuracy:86.3%\r",
      "Progress:63.2% Speed(reviews/sec):1430. #Correct:547 #Tested:633 Testing Accuracy:86.4%\r",
      "Progress:63.3% Speed(reviews/sec):1428. #Correct:547 #Tested:634 Testing Accuracy:86.2%\r",
      "Progress:63.4% Speed(reviews/sec):1425. #Correct:547 #Tested:635 Testing Accuracy:86.1%\r",
      "Progress:63.5% Speed(reviews/sec):1424. #Correct:547 #Tested:636 Testing Accuracy:86.0%\r",
      "Progress:63.6% Speed(reviews/sec):1425. #Correct:548 #Tested:637 Testing Accuracy:86.0%\r",
      "Progress:63.7% Speed(reviews/sec):1426. #Correct:549 #Tested:638 Testing Accuracy:86.0%\r",
      "Progress:63.8% Speed(reviews/sec):1427. #Correct:550 #Tested:639 Testing Accuracy:86.0%\r",
      "Progress:63.9% Speed(reviews/sec):1428. #Correct:550 #Tested:640 Testing Accuracy:85.9%\r",
      "Progress:64.0% Speed(reviews/sec):1430. #Correct:551 #Tested:641 Testing Accuracy:85.9%\r",
      "Progress:64.1% Speed(reviews/sec):1429. #Correct:551 #Tested:642 Testing Accuracy:85.8%\r",
      "Progress:64.2% Speed(reviews/sec):1430. #Correct:552 #Tested:643 Testing Accuracy:85.8%\r",
      "Progress:64.3% Speed(reviews/sec):1431. #Correct:553 #Tested:644 Testing Accuracy:85.8%\r",
      "Progress:64.4% Speed(reviews/sec):1430. #Correct:554 #Tested:645 Testing Accuracy:85.8%"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\r",
      "Progress:64.5% Speed(reviews/sec):1427. #Correct:555 #Tested:646 Testing Accuracy:85.9%\r",
      "Progress:64.6% Speed(reviews/sec):1425. #Correct:556 #Tested:647 Testing Accuracy:85.9%\r",
      "Progress:64.7% Speed(reviews/sec):1424. #Correct:557 #Tested:648 Testing Accuracy:85.9%\r",
      "Progress:64.8% Speed(reviews/sec):1425. #Correct:558 #Tested:649 Testing Accuracy:85.9%\r",
      "Progress:64.9% Speed(reviews/sec):1426. #Correct:559 #Tested:650 Testing Accuracy:86.0%\r",
      "Progress:65.0% Speed(reviews/sec):1427. #Correct:560 #Tested:651 Testing Accuracy:86.0%\r",
      "Progress:65.1% Speed(reviews/sec):1427. #Correct:561 #Tested:652 Testing Accuracy:86.0%\r",
      "Progress:65.2% Speed(reviews/sec):1429. #Correct:562 #Tested:653 Testing Accuracy:86.0%\r",
      "Progress:65.3% Speed(reviews/sec):1429. #Correct:562 #Tested:654 Testing Accuracy:85.9%\r",
      "Progress:65.4% Speed(reviews/sec):1430. #Correct:562 #Tested:655 Testing Accuracy:85.8%\r",
      "Progress:65.5% Speed(reviews/sec):1431. #Correct:563 #Tested:656 Testing Accuracy:85.8%\r",
      "Progress:65.6% Speed(reviews/sec):1432. #Correct:564 #Tested:657 Testing Accuracy:85.8%\r",
      "Progress:65.7% Speed(reviews/sec):1431. #Correct:565 #Tested:658 Testing Accuracy:85.8%\r",
      "Progress:65.8% Speed(reviews/sec):1432. #Correct:566 #Tested:659 Testing Accuracy:85.8%\r",
      "Progress:65.9% Speed(reviews/sec):1431. #Correct:567 #Tested:660 Testing Accuracy:85.9%\r",
      "Progress:66.0% Speed(reviews/sec):1432. #Correct:568 #Tested:661 Testing Accuracy:85.9%\r",
      "Progress:66.1% Speed(reviews/sec):1433. #Correct:569 #Tested:662 Testing Accuracy:85.9%\r",
      "Progress:66.2% Speed(reviews/sec):1433. #Correct:570 #Tested:663 Testing Accuracy:85.9%\r",
      "Progress:66.3% Speed(reviews/sec):1434. #Correct:571 #Tested:664 Testing Accuracy:85.9%\r",
      "Progress:66.4% Speed(reviews/sec):1435. #Correct:572 #Tested:665 Testing Accuracy:86.0%\r",
      "Progress:66.5% Speed(reviews/sec):1436. #Correct:573 #Tested:666 Testing Accuracy:86.0%\r",
      "Progress:66.6% Speed(reviews/sec):1437. #Correct:574 #Tested:667 Testing Accuracy:86.0%\r",
      "Progress:66.7% Speed(reviews/sec):1438. #Correct:575 #Tested:668 Testing Accuracy:86.0%\r",
      "Progress:66.8% Speed(reviews/sec):1439. #Correct:576 #Tested:669 Testing Accuracy:86.0%\r",
      "Progress:66.9% Speed(reviews/sec):1440. #Correct:576 #Tested:670 Testing Accuracy:85.9%\r",
      "Progress:67.0% Speed(reviews/sec):1441. #Correct:577 #Tested:671 Testing Accuracy:85.9%\r",
      "Progress:67.1% Speed(reviews/sec):1442. #Correct:578 #Tested:672 Testing Accuracy:86.0%\r",
      "Progress:67.2% Speed(reviews/sec):1444. #Correct:578 #Tested:673 Testing Accuracy:85.8%\r",
      "Progress:67.3% Speed(reviews/sec):1444. #Correct:579 #Tested:674 Testing Accuracy:85.9%\r",
      "Progress:67.4% Speed(reviews/sec):1444. #Correct:580 #Tested:675 Testing Accuracy:85.9%\r",
      "Progress:67.5% Speed(reviews/sec):1443. #Correct:580 #Tested:676 Testing Accuracy:85.7%\r",
      "Progress:67.6% Speed(reviews/sec):1443. #Correct:580 #Tested:677 Testing Accuracy:85.6%\r",
      "Progress:67.7% Speed(reviews/sec):1442. #Correct:580 #Tested:678 Testing Accuracy:85.5%\r",
      "Progress:67.8% Speed(reviews/sec):1443. #Correct:581 #Tested:679 Testing Accuracy:85.5%\r",
      "Progress:67.9% Speed(reviews/sec):1442. #Correct:582 #Tested:680 Testing Accuracy:85.5%\r",
      "Progress:68.0% Speed(reviews/sec):1442. #Correct:582 #Tested:681 Testing Accuracy:85.4%\r",
      "Progress:68.1% Speed(reviews/sec):1433. #Correct:583 #Tested:682 Testing Accuracy:85.4%\r",
      "Progress:68.2% Speed(reviews/sec):1429. #Correct:584 #Tested:683 Testing Accuracy:85.5%\r",
      "Progress:68.3% Speed(reviews/sec):1428. #Correct:585 #Tested:684 Testing Accuracy:85.5%\r",
      "Progress:68.4% Speed(reviews/sec):1429. #Correct:586 #Tested:685 Testing Accuracy:85.5%\r",
      "Progress:68.5% Speed(reviews/sec):1429. #Correct:587 #Tested:686 Testing Accuracy:85.5%\r",
      "Progress:68.6% Speed(reviews/sec):1430. #Correct:587 #Tested:687 Testing Accuracy:85.4%\r",
      "Progress:68.7% Speed(reviews/sec):1431. #Correct:588 #Tested:688 Testing Accuracy:85.4%\r",
      "Progress:68.8% Speed(reviews/sec):1429. #Correct:589 #Tested:689 Testing Accuracy:85.4%\r",
      "Progress:68.9% Speed(reviews/sec):1427. #Correct:590 #Tested:690 Testing Accuracy:85.5%\r",
      "Progress:69.0% Speed(reviews/sec):1424. #Correct:591 #Tested:691 Testing Accuracy:85.5%\r",
      "Progress:69.1% Speed(reviews/sec):1424. #Correct:592 #Tested:692 Testing Accuracy:85.5%\r",
      "Progress:69.2% Speed(reviews/sec):1425. #Correct:593 #Tested:693 Testing Accuracy:85.5%\r",
      "Progress:69.3% Speed(reviews/sec):1426. #Correct:594 #Tested:694 Testing Accuracy:85.5%\r",
      "Progress:69.4% Speed(reviews/sec):1426. #Correct:595 #Tested:695 Testing Accuracy:85.6%\r",
      "Progress:69.5% Speed(reviews/sec):1426. #Correct:596 #Tested:696 Testing Accuracy:85.6%\r",
      "Progress:69.6% Speed(reviews/sec):1428. #Correct:597 #Tested:697 Testing Accuracy:85.6%\r",
      "Progress:69.7% Speed(reviews/sec):1428. #Correct:598 #Tested:698 Testing Accuracy:85.6%\r",
      "Progress:69.8% Speed(reviews/sec):1429. #Correct:599 #Tested:699 Testing Accuracy:85.6%\r",
      "Progress:69.9% Speed(reviews/sec):1430. #Correct:600 #Tested:700 Testing Accuracy:85.7%\r",
      "Progress:70.0% Speed(reviews/sec):1427. #Correct:601 #Tested:701 Testing Accuracy:85.7%\r",
      "Progress:70.1% Speed(reviews/sec):1426. #Correct:602 #Tested:702 Testing Accuracy:85.7%\r",
      "Progress:70.2% Speed(reviews/sec):1426. #Correct:603 #Tested:703 Testing Accuracy:85.7%\r",
      "Progress:70.3% Speed(reviews/sec):1426. #Correct:604 #Tested:704 Testing Accuracy:85.7%\r",
      "Progress:70.4% Speed(reviews/sec):1426. #Correct:605 #Tested:705 Testing Accuracy:85.8%\r",
      "Progress:70.5% Speed(reviews/sec):1427. #Correct:606 #Tested:706 Testing Accuracy:85.8%\r",
      "Progress:70.6% Speed(reviews/sec):1428. #Correct:607 #Tested:707 Testing Accuracy:85.8%\r",
      "Progress:70.7% Speed(reviews/sec):1429. #Correct:608 #Tested:708 Testing Accuracy:85.8%\r",
      "Progress:70.8% Speed(reviews/sec):1429. #Correct:609 #Tested:709 Testing Accuracy:85.8%\r",
      "Progress:70.9% Speed(reviews/sec):1429. #Correct:610 #Tested:710 Testing Accuracy:85.9%\r",
      "Progress:71.0% Speed(reviews/sec):1429. #Correct:610 #Tested:711 Testing Accuracy:85.7%\r",
      "Progress:71.1% Speed(reviews/sec):1430. #Correct:611 #Tested:712 Testing Accuracy:85.8%\r",
      "Progress:71.2% Speed(reviews/sec):1432. #Correct:612 #Tested:713 Testing Accuracy:85.8%\r",
      "Progress:71.3% Speed(reviews/sec):1432. #Correct:613 #Tested:714 Testing Accuracy:85.8%\r",
      "Progress:71.4% Speed(reviews/sec):1430. #Correct:614 #Tested:715 Testing Accuracy:85.8%\r",
      "Progress:71.5% Speed(reviews/sec):1430. #Correct:615 #Tested:716 Testing Accuracy:85.8%\r",
      "Progress:71.6% Speed(reviews/sec):1427. #Correct:616 #Tested:717 Testing Accuracy:85.9%\r",
      "Progress:71.7% Speed(reviews/sec):1428. #Correct:617 #Tested:718 Testing Accuracy:85.9%\r",
      "Progress:71.8% Speed(reviews/sec):1429. #Correct:617 #Tested:719 Testing Accuracy:85.8%\r",
      "Progress:71.9% Speed(reviews/sec):1430. #Correct:618 #Tested:720 Testing Accuracy:85.8%\r",
      "Progress:72.0% Speed(reviews/sec):1431. #Correct:619 #Tested:721 Testing Accuracy:85.8%\r",
      "Progress:72.1% Speed(reviews/sec):1432. #Correct:620 #Tested:722 Testing Accuracy:85.8%\r",
      "Progress:72.2% Speed(reviews/sec):1432. #Correct:621 #Tested:723 Testing Accuracy:85.8%\r",
      "Progress:72.3% Speed(reviews/sec):1433. #Correct:621 #Tested:724 Testing Accuracy:85.7%\r",
      "Progress:72.4% Speed(reviews/sec):1434. #Correct:621 #Tested:725 Testing Accuracy:85.6%\r",
      "Progress:72.5% Speed(reviews/sec):1434. #Correct:622 #Tested:726 Testing Accuracy:85.6%\r",
      "Progress:72.6% Speed(reviews/sec):1433. #Correct:623 #Tested:727 Testing Accuracy:85.6%\r",
      "Progress:72.7% Speed(reviews/sec):1432. #Correct:624 #Tested:728 Testing Accuracy:85.7%\r",
      "Progress:72.8% Speed(reviews/sec):1434. #Correct:625 #Tested:729 Testing Accuracy:85.7%\r",
      "Progress:72.9% Speed(reviews/sec):1433. #Correct:626 #Tested:730 Testing Accuracy:85.7%\r",
      "Progress:73.0% Speed(reviews/sec):1430. #Correct:627 #Tested:731 Testing Accuracy:85.7%\r",
      "Progress:73.1% Speed(reviews/sec):1431. #Correct:628 #Tested:732 Testing Accuracy:85.7%\r",
      "Progress:73.2% Speed(reviews/sec):1431. #Correct:629 #Tested:733 Testing Accuracy:85.8%\r",
      "Progress:73.3% Speed(reviews/sec):1431. #Correct:630 #Tested:734 Testing Accuracy:85.8%\r",
      "Progress:73.4% Speed(reviews/sec):1430. #Correct:631 #Tested:735 Testing Accuracy:85.8%\r",
      "Progress:73.5% Speed(reviews/sec):1431. #Correct:632 #Tested:736 Testing Accuracy:85.8%\r",
      "Progress:73.6% Speed(reviews/sec):1432. #Correct:633 #Tested:737 Testing Accuracy:85.8%\r",
      "Progress:73.7% Speed(reviews/sec):1433. #Correct:633 #Tested:738 Testing Accuracy:85.7%\r",
      "Progress:73.8% Speed(reviews/sec):1433. #Correct:634 #Tested:739 Testing Accuracy:85.7%\r",
      "Progress:73.9% Speed(reviews/sec):1434. #Correct:635 #Tested:740 Testing Accuracy:85.8%\r",
      "Progress:74.0% Speed(reviews/sec):1435. #Correct:636 #Tested:741 Testing Accuracy:85.8%\r",
      "Progress:74.1% Speed(reviews/sec):1436. #Correct:637 #Tested:742 Testing Accuracy:85.8%\r",
      "Progress:74.2% Speed(reviews/sec):1437. #Correct:638 #Tested:743 Testing Accuracy:85.8%\r",
      "Progress:74.3% Speed(reviews/sec):1437. #Correct:638 #Tested:744 Testing Accuracy:85.7%\r",
      "Progress:74.4% Speed(reviews/sec):1438. #Correct:639 #Tested:745 Testing Accuracy:85.7%\r",
      "Progress:74.5% Speed(reviews/sec):1437. #Correct:639 #Tested:746 Testing Accuracy:85.6%\r",
      "Progress:74.6% Speed(reviews/sec):1437. #Correct:640 #Tested:747 Testing Accuracy:85.6%\r",
      "Progress:74.7% Speed(reviews/sec):1436. #Correct:641 #Tested:748 Testing Accuracy:85.6%\r",
      "Progress:74.8% Speed(reviews/sec):1436. #Correct:642 #Tested:749 Testing Accuracy:85.7%\r",
      "Progress:74.9% Speed(reviews/sec):1437. #Correct:643 #Tested:750 Testing Accuracy:85.7%\r",
      "Progress:75.0% Speed(reviews/sec):1438. #Correct:644 #Tested:751 Testing Accuracy:85.7%\r",
      "Progress:75.1% Speed(reviews/sec):1439. #Correct:645 #Tested:752 Testing Accuracy:85.7%\r",
      "Progress:75.2% Speed(reviews/sec):1439. #Correct:646 #Tested:753 Testing Accuracy:85.7%\r",
      "Progress:75.3% Speed(reviews/sec):1440. #Correct:647 #Tested:754 Testing Accuracy:85.8%\r",
      "Progress:75.4% Speed(reviews/sec):1439. #Correct:648 #Tested:755 Testing Accuracy:85.8%\r",
      "Progress:75.5% Speed(reviews/sec):1440. #Correct:649 #Tested:756 Testing Accuracy:85.8%\r",
      "Progress:75.6% Speed(reviews/sec):1441. #Correct:649 #Tested:757 Testing Accuracy:85.7%\r",
      "Progress:75.7% Speed(reviews/sec):1442. #Correct:650 #Tested:758 Testing Accuracy:85.7%\r",
      "Progress:75.8% Speed(reviews/sec):1442. #Correct:650 #Tested:759 Testing Accuracy:85.6%\r",
      "Progress:75.9% Speed(reviews/sec):1443. #Correct:651 #Tested:760 Testing Accuracy:85.6%\r",
      "Progress:76.0% Speed(reviews/sec):1443. #Correct:651 #Tested:761 Testing Accuracy:85.5%\r",
      "Progress:76.1% Speed(reviews/sec):1444. #Correct:652 #Tested:762 Testing Accuracy:85.5%\r",
      "Progress:76.2% Speed(reviews/sec):1444. #Correct:652 #Tested:763 Testing Accuracy:85.4%\r",
      "Progress:76.3% Speed(reviews/sec):1445. #Correct:653 #Tested:764 Testing Accuracy:85.4%\r",
      "Progress:76.4% Speed(reviews/sec):1445. #Correct:654 #Tested:765 Testing Accuracy:85.4%\r",
      "Progress:76.5% Speed(reviews/sec):1443. #Correct:654 #Tested:766 Testing Accuracy:85.3%\r",
      "Progress:76.6% Speed(reviews/sec):1443. #Correct:654 #Tested:767 Testing Accuracy:85.2%\r",
      "Progress:76.7% Speed(reviews/sec):1444. #Correct:655 #Tested:768 Testing Accuracy:85.2%\r",
      "Progress:76.8% Speed(reviews/sec):1444. #Correct:655 #Tested:769 Testing Accuracy:85.1%\r",
      "Progress:76.9% Speed(reviews/sec):1445. #Correct:656 #Tested:770 Testing Accuracy:85.1%\r",
      "Progress:77.0% Speed(reviews/sec):1444. #Correct:657 #Tested:771 Testing Accuracy:85.2%\r",
      "Progress:77.1% Speed(reviews/sec):1445. #Correct:657 #Tested:772 Testing Accuracy:85.1%\r",
      "Progress:77.2% Speed(reviews/sec):1446. #Correct:658 #Tested:773 Testing Accuracy:85.1%\r",
      "Progress:77.3% Speed(reviews/sec):1446. #Correct:659 #Tested:774 Testing Accuracy:85.1%\r",
      "Progress:77.4% Speed(reviews/sec):1445. #Correct:660 #Tested:775 Testing Accuracy:85.1%\r",
      "Progress:77.5% Speed(reviews/sec):1445. #Correct:661 #Tested:776 Testing Accuracy:85.1%\r",
      "Progress:77.6% Speed(reviews/sec):1445. #Correct:661 #Tested:777 Testing Accuracy:85.0%\r",
      "Progress:77.7% Speed(reviews/sec):1445. #Correct:661 #Tested:778 Testing Accuracy:84.9%\r",
      "Progress:77.8% Speed(reviews/sec):1443. #Correct:662 #Tested:779 Testing Accuracy:84.9%\r",
      "Progress:77.9% Speed(reviews/sec):1438. #Correct:663 #Tested:780 Testing Accuracy:85.0%\r",
      "Progress:78.0% Speed(reviews/sec):1435. #Correct:664 #Tested:781 Testing Accuracy:85.0%\r",
      "Progress:78.1% Speed(reviews/sec):1437. #Correct:664 #Tested:782 Testing Accuracy:84.9%\r",
      "Progress:78.2% Speed(reviews/sec):1437. #Correct:664 #Tested:783 Testing Accuracy:84.8%\r",
      "Progress:78.3% Speed(reviews/sec):1436. #Correct:665 #Tested:784 Testing Accuracy:84.8%\r",
      "Progress:78.4% Speed(reviews/sec):1436. #Correct:666 #Tested:785 Testing Accuracy:84.8%\r",
      "Progress:78.5% Speed(reviews/sec):1437. #Correct:666 #Tested:786 Testing Accuracy:84.7%\r",
      "Progress:78.6% Speed(reviews/sec):1437. #Correct:667 #Tested:787 Testing Accuracy:84.7%\r",
      "Progress:78.7% Speed(reviews/sec):1436. #Correct:668 #Tested:788 Testing Accuracy:84.7%\r",
      "Progress:78.8% Speed(reviews/sec):1437. #Correct:668 #Tested:789 Testing Accuracy:84.6%\r",
      "Progress:78.9% Speed(reviews/sec):1436. #Correct:669 #Tested:790 Testing Accuracy:84.6%\r",
      "Progress:79.0% Speed(reviews/sec):1436. #Correct:670 #Tested:791 Testing Accuracy:84.7%\r",
      "Progress:79.1% Speed(reviews/sec):1436. #Correct:671 #Tested:792 Testing Accuracy:84.7%\r",
      "Progress:79.2% Speed(reviews/sec):1435. #Correct:672 #Tested:793 Testing Accuracy:84.7%\r",
      "Progress:79.3% Speed(reviews/sec):1436. #Correct:673 #Tested:794 Testing Accuracy:84.7%\r",
      "Progress:79.4% Speed(reviews/sec):1436. #Correct:674 #Tested:795 Testing Accuracy:84.7%\r",
      "Progress:79.5% Speed(reviews/sec):1437. #Correct:675 #Tested:796 Testing Accuracy:84.7%\r",
      "Progress:79.6% Speed(reviews/sec):1438. #Correct:676 #Tested:797 Testing Accuracy:84.8%\r",
      "Progress:79.7% Speed(reviews/sec):1439. #Correct:677 #Tested:798 Testing Accuracy:84.8%\r",
      "Progress:79.8% Speed(reviews/sec):1440. #Correct:678 #Tested:799 Testing Accuracy:84.8%\r",
      "Progress:79.9% Speed(reviews/sec):1441. #Correct:679 #Tested:800 Testing Accuracy:84.8%\r",
      "Progress:80.0% Speed(reviews/sec):1442. #Correct:680 #Tested:801 Testing Accuracy:84.8%\r",
      "Progress:80.1% Speed(reviews/sec):1439. #Correct:681 #Tested:802 Testing Accuracy:84.9%\r",
      "Progress:80.2% Speed(reviews/sec):1440. #Correct:682 #Tested:803 Testing Accuracy:84.9%\r",
      "Progress:80.3% Speed(reviews/sec):1440. #Correct:683 #Tested:804 Testing Accuracy:84.9%\r",
      "Progress:80.4% Speed(reviews/sec):1441. #Correct:684 #Tested:805 Testing Accuracy:84.9%\r",
      "Progress:80.5% Speed(reviews/sec):1443. #Correct:684 #Tested:806 Testing Accuracy:84.8%\r",
      "Progress:80.6% Speed(reviews/sec):1443. #Correct:685 #Tested:807 Testing Accuracy:84.8%\r",
      "Progress:80.7% Speed(reviews/sec):1442. #Correct:685 #Tested:808 Testing Accuracy:84.7%\r",
      "Progress:80.8% Speed(reviews/sec):1441. #Correct:686 #Tested:809 Testing Accuracy:84.7%\r",
      "Progress:80.9% Speed(reviews/sec):1441. #Correct:687 #Tested:810 Testing Accuracy:84.8%\r",
      "Progress:81.0% Speed(reviews/sec):1441. #Correct:688 #Tested:811 Testing Accuracy:84.8%\r",
      "Progress:81.1% Speed(reviews/sec):1441. #Correct:689 #Tested:812 Testing Accuracy:84.8%\r",
      "Progress:81.2% Speed(reviews/sec):1442. #Correct:690 #Tested:813 Testing Accuracy:84.8%\r",
      "Progress:81.3% Speed(reviews/sec):1443. #Correct:691 #Tested:814 Testing Accuracy:84.8%\r",
      "Progress:81.4% Speed(reviews/sec):1440. #Correct:692 #Tested:815 Testing Accuracy:84.9%\r",
      "Progress:81.5% Speed(reviews/sec):1441. #Correct:693 #Tested:816 Testing Accuracy:84.9%\r",
      "Progress:81.6% Speed(reviews/sec):1442. #Correct:694 #Tested:817 Testing Accuracy:84.9%\r",
      "Progress:81.7% Speed(reviews/sec):1442. #Correct:695 #Tested:818 Testing Accuracy:84.9%\r",
      "Progress:81.8% Speed(reviews/sec):1442. #Correct:696 #Tested:819 Testing Accuracy:84.9%\r",
      "Progress:81.9% Speed(reviews/sec):1443. #Correct:697 #Tested:820 Testing Accuracy:85.0%\r",
      "Progress:82.0% Speed(reviews/sec):1441. #Correct:698 #Tested:821 Testing Accuracy:85.0%\r",
      "Progress:82.1% Speed(reviews/sec):1441. #Correct:698 #Tested:822 Testing Accuracy:84.9%\r",
      "Progress:82.2% Speed(reviews/sec):1440. #Correct:698 #Tested:823 Testing Accuracy:84.8%\r",
      "Progress:82.3% Speed(reviews/sec):1441. #Correct:699 #Tested:824 Testing Accuracy:84.8%\r",
      "Progress:82.4% Speed(reviews/sec):1441. #Correct:700 #Tested:825 Testing Accuracy:84.8%\r",
      "Progress:82.5% Speed(reviews/sec):1441. #Correct:700 #Tested:826 Testing Accuracy:84.7%\r",
      "Progress:82.6% Speed(reviews/sec):1438. #Correct:701 #Tested:827 Testing Accuracy:84.7%\r",
      "Progress:82.7% Speed(reviews/sec):1439. #Correct:702 #Tested:828 Testing Accuracy:84.7%\r",
      "Progress:82.8% Speed(reviews/sec):1439. #Correct:703 #Tested:829 Testing Accuracy:84.8%\r",
      "Progress:82.9% Speed(reviews/sec):1439. #Correct:704 #Tested:830 Testing Accuracy:84.8%\r",
      "Progress:83.0% Speed(reviews/sec):1440. #Correct:705 #Tested:831 Testing Accuracy:84.8%\r",
      "Progress:83.1% Speed(reviews/sec):1441. #Correct:706 #Tested:832 Testing Accuracy:84.8%\r",
      "Progress:83.2% Speed(reviews/sec):1441. #Correct:707 #Tested:833 Testing Accuracy:84.8%\r",
      "Progress:83.3% Speed(reviews/sec):1442. #Correct:708 #Tested:834 Testing Accuracy:84.8%\r",
      "Progress:83.4% Speed(reviews/sec):1442. #Correct:709 #Tested:835 Testing Accuracy:84.9%\r",
      "Progress:83.5% Speed(reviews/sec):1443. #Correct:710 #Tested:836 Testing Accuracy:84.9%\r",
      "Progress:83.6% Speed(reviews/sec):1443. #Correct:711 #Tested:837 Testing Accuracy:84.9%\r",
      "Progress:83.7% Speed(reviews/sec):1441. #Correct:712 #Tested:838 Testing Accuracy:84.9%\r",
      "Progress:83.8% Speed(reviews/sec):1438. #Correct:712 #Tested:839 Testing Accuracy:84.8%\r",
      "Progress:83.9% Speed(reviews/sec):1439. #Correct:713 #Tested:840 Testing Accuracy:84.8%\r",
      "Progress:84.0% Speed(reviews/sec):1440. #Correct:714 #Tested:841 Testing Accuracy:84.8%\r",
      "Progress:84.1% Speed(reviews/sec):1440. #Correct:715 #Tested:842 Testing Accuracy:84.9%\r",
      "Progress:84.2% Speed(reviews/sec):1441. #Correct:716 #Tested:843 Testing Accuracy:84.9%\r",
      "Progress:84.3% Speed(reviews/sec):1442. #Correct:717 #Tested:844 Testing Accuracy:84.9%\r",
      "Progress:84.4% Speed(reviews/sec):1442. #Correct:718 #Tested:845 Testing Accuracy:84.9%\r",
      "Progress:84.5% Speed(reviews/sec):1443. #Correct:719 #Tested:846 Testing Accuracy:84.9%\r",
      "Progress:84.6% Speed(reviews/sec):1443. #Correct:720 #Tested:847 Testing Accuracy:85.0%\r",
      "Progress:84.7% Speed(reviews/sec):1444. #Correct:720 #Tested:848 Testing Accuracy:84.9%\r",
      "Progress:84.8% Speed(reviews/sec):1445. #Correct:721 #Tested:849 Testing Accuracy:84.9%\r",
      "Progress:84.9% Speed(reviews/sec):1444. #Correct:722 #Tested:850 Testing Accuracy:84.9%\r",
      "Progress:85.0% Speed(reviews/sec):1440. #Correct:723 #Tested:851 Testing Accuracy:84.9%\r",
      "Progress:85.1% Speed(reviews/sec):1440. #Correct:724 #Tested:852 Testing Accuracy:84.9%\r",
      "Progress:85.2% Speed(reviews/sec):1440. #Correct:725 #Tested:853 Testing Accuracy:84.9%\r",
      "Progress:85.3% Speed(reviews/sec):1440. #Correct:726 #Tested:854 Testing Accuracy:85.0%\r",
      "Progress:85.4% Speed(reviews/sec):1440. #Correct:727 #Tested:855 Testing Accuracy:85.0%\r",
      "Progress:85.5% Speed(reviews/sec):1441. #Correct:728 #Tested:856 Testing Accuracy:85.0%\r",
      "Progress:85.6% Speed(reviews/sec):1441. #Correct:729 #Tested:857 Testing Accuracy:85.0%\r",
      "Progress:85.7% Speed(reviews/sec):1441. #Correct:729 #Tested:858 Testing Accuracy:84.9%\r",
      "Progress:85.8% Speed(reviews/sec):1440. #Correct:730 #Tested:859 Testing Accuracy:84.9%\r",
      "Progress:85.9% Speed(reviews/sec):1441. #Correct:731 #Tested:860 Testing Accuracy:85.0%\r",
      "Progress:86.0% Speed(reviews/sec):1442. #Correct:732 #Tested:861 Testing Accuracy:85.0%\r",
      "Progress:86.1% Speed(reviews/sec):1442. #Correct:733 #Tested:862 Testing Accuracy:85.0%\r",
      "Progress:86.2% Speed(reviews/sec):1442. #Correct:734 #Tested:863 Testing Accuracy:85.0%\r",
      "Progress:86.3% Speed(reviews/sec):1440. #Correct:735 #Tested:864 Testing Accuracy:85.0%\r",
      "Progress:86.4% Speed(reviews/sec):1441. #Correct:736 #Tested:865 Testing Accuracy:85.0%\r",
      "Progress:86.5% Speed(reviews/sec):1442. #Correct:737 #Tested:866 Testing Accuracy:85.1%\r",
      "Progress:86.6% Speed(reviews/sec):1442. #Correct:738 #Tested:867 Testing Accuracy:85.1%\r",
      "Progress:86.7% Speed(reviews/sec):1444. #Correct:739 #Tested:868 Testing Accuracy:85.1%\r",
      "Progress:86.8% Speed(reviews/sec):1444. #Correct:740 #Tested:869 Testing Accuracy:85.1%\r",
      "Progress:86.9% Speed(reviews/sec):1445. #Correct:741 #Tested:870 Testing Accuracy:85.1%\r",
      "Progress:87.0% Speed(reviews/sec):1446. #Correct:742 #Tested:871 Testing Accuracy:85.1%\r",
      "Progress:87.1% Speed(reviews/sec):1446. #Correct:743 #Tested:872 Testing Accuracy:85.2%\r",
      "Progress:87.2% Speed(reviews/sec):1446. #Correct:744 #Tested:873 Testing Accuracy:85.2%\r",
      "Progress:87.3% Speed(reviews/sec):1448. #Correct:745 #Tested:874 Testing Accuracy:85.2%\r",
      "Progress:87.4% Speed(reviews/sec):1448. #Correct:746 #Tested:875 Testing Accuracy:85.2%\r",
      "Progress:87.5% Speed(reviews/sec):1449. #Correct:747 #Tested:876 Testing Accuracy:85.2%\r",
      "Progress:87.6% Speed(reviews/sec):1450. #Correct:747 #Tested:877 Testing Accuracy:85.1%\r",
      "Progress:87.7% Speed(reviews/sec):1448. #Correct:747 #Tested:878 Testing Accuracy:85.0%\r",
      "Progress:87.8% Speed(reviews/sec):1449. #Correct:748 #Tested:879 Testing Accuracy:85.0%\r",
      "Progress:87.9% Speed(reviews/sec):1449. #Correct:749 #Tested:880 Testing Accuracy:85.1%\r",
      "Progress:88.0% Speed(reviews/sec):1450. #Correct:750 #Tested:881 Testing Accuracy:85.1%\r",
      "Progress:88.1% Speed(reviews/sec):1451. #Correct:751 #Tested:882 Testing Accuracy:85.1%\r",
      "Progress:88.2% Speed(reviews/sec):1451. #Correct:752 #Tested:883 Testing Accuracy:85.1%\r",
      "Progress:88.3% Speed(reviews/sec):1452. #Correct:753 #Tested:884 Testing Accuracy:85.1%\r",
      "Progress:88.4% Speed(reviews/sec):1453. #Correct:754 #Tested:885 Testing Accuracy:85.1%\r",
      "Progress:88.5% Speed(reviews/sec):1454. #Correct:755 #Tested:886 Testing Accuracy:85.2%\r",
      "Progress:88.6% Speed(reviews/sec):1455. #Correct:756 #Tested:887 Testing Accuracy:85.2%\r",
      "Progress:88.7% Speed(reviews/sec):1455. #Correct:757 #Tested:888 Testing Accuracy:85.2%\r",
      "Progress:88.8% Speed(reviews/sec):1456. #Correct:757 #Tested:889 Testing Accuracy:85.1%\r",
      "Progress:88.9% Speed(reviews/sec):1456. #Correct:757 #Tested:890 Testing Accuracy:85.0%\r",
      "Progress:89.0% Speed(reviews/sec):1456. #Correct:758 #Tested:891 Testing Accuracy:85.0%\r",
      "Progress:89.1% Speed(reviews/sec):1456. #Correct:759 #Tested:892 Testing Accuracy:85.0%\r",
      "Progress:89.2% Speed(reviews/sec):1457. #Correct:759 #Tested:893 Testing Accuracy:84.9%\r",
      "Progress:89.3% Speed(reviews/sec):1458. #Correct:760 #Tested:894 Testing Accuracy:85.0%\r",
      "Progress:89.4% Speed(reviews/sec):1459. #Correct:761 #Tested:895 Testing Accuracy:85.0%\r",
      "Progress:89.5% Speed(reviews/sec):1460. #Correct:762 #Tested:896 Testing Accuracy:85.0%\r",
      "Progress:89.6% Speed(reviews/sec):1461. #Correct:763 #Tested:897 Testing Accuracy:85.0%\r",
      "Progress:89.7% Speed(reviews/sec):1461. #Correct:764 #Tested:898 Testing Accuracy:85.0%\r",
      "Progress:89.8% Speed(reviews/sec):1460. #Correct:765 #Tested:899 Testing Accuracy:85.0%\r",
      "Progress:89.9% Speed(reviews/sec):1461. #Correct:766 #Tested:900 Testing Accuracy:85.1%\r",
      "Progress:90.0% Speed(reviews/sec):1461. #Correct:766 #Tested:901 Testing Accuracy:85.0%\r",
      "Progress:90.1% Speed(reviews/sec):1460. #Correct:767 #Tested:902 Testing Accuracy:85.0%\r",
      "Progress:90.2% Speed(reviews/sec):1460. #Correct:768 #Tested:903 Testing Accuracy:85.0%\r",
      "Progress:90.3% Speed(reviews/sec):1461. #Correct:769 #Tested:904 Testing Accuracy:85.0%\r",
      "Progress:90.4% Speed(reviews/sec):1462. #Correct:770 #Tested:905 Testing Accuracy:85.0%\r",
      "Progress:90.5% Speed(reviews/sec):1462. #Correct:771 #Tested:906 Testing Accuracy:85.0%\r",
      "Progress:90.6% Speed(reviews/sec):1463. #Correct:772 #Tested:907 Testing Accuracy:85.1%\r",
      "Progress:90.7% Speed(reviews/sec):1455. #Correct:773 #Tested:908 Testing Accuracy:85.1%\r",
      "Progress:90.8% Speed(reviews/sec):1455. #Correct:773 #Tested:909 Testing Accuracy:85.0%\r",
      "Progress:90.9% Speed(reviews/sec):1456. #Correct:773 #Tested:910 Testing Accuracy:84.9%\r",
      "Progress:91.0% Speed(reviews/sec):1456. #Correct:774 #Tested:911 Testing Accuracy:84.9%\r",
      "Progress:91.1% Speed(reviews/sec):1457. #Correct:774 #Tested:912 Testing Accuracy:84.8%\r",
      "Progress:91.2% Speed(reviews/sec):1457. #Correct:775 #Tested:913 Testing Accuracy:84.8%\r",
      "Progress:91.3% Speed(reviews/sec):1457. #Correct:775 #Tested:914 Testing Accuracy:84.7%\r",
      "Progress:91.4% Speed(reviews/sec):1457. #Correct:776 #Tested:915 Testing Accuracy:84.8%\r",
      "Progress:91.5% Speed(reviews/sec):1458. #Correct:776 #Tested:916 Testing Accuracy:84.7%\r",
      "Progress:91.6% Speed(reviews/sec):1458. #Correct:776 #Tested:917 Testing Accuracy:84.6%\r",
      "Progress:91.7% Speed(reviews/sec):1458. #Correct:777 #Tested:918 Testing Accuracy:84.6%\r",
      "Progress:91.8% Speed(reviews/sec):1459. #Correct:778 #Tested:919 Testing Accuracy:84.6%\r",
      "Progress:91.9% Speed(reviews/sec):1459. #Correct:779 #Tested:920 Testing Accuracy:84.6%\r",
      "Progress:92.0% Speed(reviews/sec):1456. #Correct:779 #Tested:921 Testing Accuracy:84.5%\r",
      "Progress:92.1% Speed(reviews/sec):1456. #Correct:780 #Tested:922 Testing Accuracy:84.5%\r",
      "Progress:92.2% Speed(reviews/sec):1457. #Correct:781 #Tested:923 Testing Accuracy:84.6%\r",
      "Progress:92.3% Speed(reviews/sec):1458. #Correct:781 #Tested:924 Testing Accuracy:84.5%\r",
      "Progress:92.4% Speed(reviews/sec):1459. #Correct:782 #Tested:925 Testing Accuracy:84.5%\r",
      "Progress:92.5% Speed(reviews/sec):1459. #Correct:783 #Tested:926 Testing Accuracy:84.5%\r",
      "Progress:92.6% Speed(reviews/sec):1460. #Correct:784 #Tested:927 Testing Accuracy:84.5%\r",
      "Progress:92.7% Speed(reviews/sec):1461. #Correct:785 #Tested:928 Testing Accuracy:84.5%\r",
      "Progress:92.8% Speed(reviews/sec):1462. #Correct:786 #Tested:929 Testing Accuracy:84.6%\r",
      "Progress:92.9% Speed(reviews/sec):1462. #Correct:787 #Tested:930 Testing Accuracy:84.6%\r",
      "Progress:93.0% Speed(reviews/sec):1463. #Correct:788 #Tested:931 Testing Accuracy:84.6%\r",
      "Progress:93.1% Speed(reviews/sec):1463. #Correct:788 #Tested:932 Testing Accuracy:84.5%\r",
      "Progress:93.2% Speed(reviews/sec):1461. #Correct:789 #Tested:933 Testing Accuracy:84.5%\r",
      "Progress:93.3% Speed(reviews/sec):1460. #Correct:790 #Tested:934 Testing Accuracy:84.5%\r",
      "Progress:93.4% Speed(reviews/sec):1460. #Correct:791 #Tested:935 Testing Accuracy:84.5%\r",
      "Progress:93.5% Speed(reviews/sec):1460. #Correct:792 #Tested:936 Testing Accuracy:84.6%\r",
      "Progress:93.6% Speed(reviews/sec):1461. #Correct:792 #Tested:937 Testing Accuracy:84.5%\r",
      "Progress:93.7% Speed(reviews/sec):1461. #Correct:793 #Tested:938 Testing Accuracy:84.5%\r",
      "Progress:93.8% Speed(reviews/sec):1462. #Correct:793 #Tested:939 Testing Accuracy:84.4%\r",
      "Progress:93.9% Speed(reviews/sec):1462. #Correct:794 #Tested:940 Testing Accuracy:84.4%\r",
      "Progress:94.0% Speed(reviews/sec):1463. #Correct:795 #Tested:941 Testing Accuracy:84.4%\r",
      "Progress:94.1% Speed(reviews/sec):1464. #Correct:795 #Tested:942 Testing Accuracy:84.3%\r",
      "Progress:94.2% Speed(reviews/sec):1465. #Correct:796 #Tested:943 Testing Accuracy:84.4%\r",
      "Progress:94.3% Speed(reviews/sec):1465. #Correct:797 #Tested:944 Testing Accuracy:84.4%\r",
      "Progress:94.4% Speed(reviews/sec):1463. #Correct:798 #Tested:945 Testing Accuracy:84.4%\r",
      "Progress:94.5% Speed(reviews/sec):1463. #Correct:798 #Tested:946 Testing Accuracy:84.3%\r",
      "Progress:94.6% Speed(reviews/sec):1462. #Correct:799 #Tested:947 Testing Accuracy:84.3%\r",
      "Progress:94.7% Speed(reviews/sec):1462. #Correct:800 #Tested:948 Testing Accuracy:84.3%\r",
      "Progress:94.8% Speed(reviews/sec):1463. #Correct:801 #Tested:949 Testing Accuracy:84.4%\r",
      "Progress:94.9% Speed(reviews/sec):1464. #Correct:802 #Tested:950 Testing Accuracy:84.4%\r",
      "Progress:95.0% Speed(reviews/sec):1464. #Correct:803 #Tested:951 Testing Accuracy:84.4%\r",
      "Progress:95.1% Speed(reviews/sec):1464. #Correct:804 #Tested:952 Testing Accuracy:84.4%\r",
      "Progress:95.2% Speed(reviews/sec):1464. #Correct:805 #Tested:953 Testing Accuracy:84.4%\r",
      "Progress:95.3% Speed(reviews/sec):1465. #Correct:806 #Tested:954 Testing Accuracy:84.4%\r",
      "Progress:95.4% Speed(reviews/sec):1464. #Correct:807 #Tested:955 Testing Accuracy:84.5%\r",
      "Progress:95.5% Speed(reviews/sec):1465. #Correct:808 #Tested:956 Testing Accuracy:84.5%\r",
      "Progress:95.6% Speed(reviews/sec):1465. #Correct:809 #Tested:957 Testing Accuracy:84.5%\r",
      "Progress:95.7% Speed(reviews/sec):1466. #Correct:810 #Tested:958 Testing Accuracy:84.5%\r",
      "Progress:95.8% Speed(reviews/sec):1467. #Correct:811 #Tested:959 Testing Accuracy:84.5%\r",
      "Progress:95.9% Speed(reviews/sec):1468. #Correct:812 #Tested:960 Testing Accuracy:84.5%\r",
      "Progress:96.0% Speed(reviews/sec):1467. #Correct:813 #Tested:961 Testing Accuracy:84.5%\r",
      "Progress:96.1% Speed(reviews/sec):1465. #Correct:814 #Tested:962 Testing Accuracy:84.6%\r",
      "Progress:96.2% Speed(reviews/sec):1466. #Correct:815 #Tested:963 Testing Accuracy:84.6%\r",
      "Progress:96.3% Speed(reviews/sec):1466. #Correct:816 #Tested:964 Testing Accuracy:84.6%\r",
      "Progress:96.4% Speed(reviews/sec):1467. #Correct:817 #Tested:965 Testing Accuracy:84.6%\r",
      "Progress:96.5% Speed(reviews/sec):1467. #Correct:818 #Tested:966 Testing Accuracy:84.6%\r",
      "Progress:96.6% Speed(reviews/sec):1468. #Correct:819 #Tested:967 Testing Accuracy:84.6%\r",
      "Progress:96.7% Speed(reviews/sec):1469. #Correct:820 #Tested:968 Testing Accuracy:84.7%\r",
      "Progress:96.8% Speed(reviews/sec):1469. #Correct:821 #Tested:969 Testing Accuracy:84.7%\r",
      "Progress:96.9% Speed(reviews/sec):1470. #Correct:822 #Tested:970 Testing Accuracy:84.7%\r",
      "Progress:97.0% Speed(reviews/sec):1470. #Correct:823 #Tested:971 Testing Accuracy:84.7%\r",
      "Progress:97.1% Speed(reviews/sec):1471. #Correct:824 #Tested:972 Testing Accuracy:84.7%\r",
      "Progress:97.2% Speed(reviews/sec):1472. #Correct:825 #Tested:973 Testing Accuracy:84.7%\r",
      "Progress:97.3% Speed(reviews/sec):1470. #Correct:826 #Tested:974 Testing Accuracy:84.8%\r",
      "Progress:97.4% Speed(reviews/sec):1469. #Correct:827 #Tested:975 Testing Accuracy:84.8%\r",
      "Progress:97.5% Speed(reviews/sec):1469. #Correct:828 #Tested:976 Testing Accuracy:84.8%\r",
      "Progress:97.6% Speed(reviews/sec):1470. #Correct:829 #Tested:977 Testing Accuracy:84.8%\r",
      "Progress:97.7% Speed(reviews/sec):1469. #Correct:829 #Tested:978 Testing Accuracy:84.7%\r",
      "Progress:97.8% Speed(reviews/sec):1469. #Correct:830 #Tested:979 Testing Accuracy:84.7%\r",
      "Progress:97.9% Speed(reviews/sec):1469. #Correct:831 #Tested:980 Testing Accuracy:84.7%\r",
      "Progress:98.0% Speed(reviews/sec):1470. #Correct:832 #Tested:981 Testing Accuracy:84.8%\r",
      "Progress:98.1% Speed(reviews/sec):1468. #Correct:833 #Tested:982 Testing Accuracy:84.8%\r",
      "Progress:98.2% Speed(reviews/sec):1469. #Correct:834 #Tested:983 Testing Accuracy:84.8%\r",
      "Progress:98.3% Speed(reviews/sec):1468. #Correct:835 #Tested:984 Testing Accuracy:84.8%\r",
      "Progress:98.4% Speed(reviews/sec):1468. #Correct:836 #Tested:985 Testing Accuracy:84.8%"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\r",
      "Progress:98.5% Speed(reviews/sec):1468. #Correct:837 #Tested:986 Testing Accuracy:84.8%\r",
      "Progress:98.6% Speed(reviews/sec):1469. #Correct:838 #Tested:987 Testing Accuracy:84.9%\r",
      "Progress:98.7% Speed(reviews/sec):1465. #Correct:839 #Tested:988 Testing Accuracy:84.9%\r",
      "Progress:98.8% Speed(reviews/sec):1463. #Correct:840 #Tested:989 Testing Accuracy:84.9%\r",
      "Progress:98.9% Speed(reviews/sec):1463. #Correct:841 #Tested:990 Testing Accuracy:84.9%\r",
      "Progress:99.0% Speed(reviews/sec):1463. #Correct:842 #Tested:991 Testing Accuracy:84.9%\r",
      "Progress:99.1% Speed(reviews/sec):1462. #Correct:842 #Tested:992 Testing Accuracy:84.8%\r",
      "Progress:99.2% Speed(reviews/sec):1463. #Correct:843 #Tested:993 Testing Accuracy:84.8%\r",
      "Progress:99.3% Speed(reviews/sec):1463. #Correct:844 #Tested:994 Testing Accuracy:84.9%\r",
      "Progress:99.4% Speed(reviews/sec):1462. #Correct:845 #Tested:995 Testing Accuracy:84.9%\r",
      "Progress:99.5% Speed(reviews/sec):1462. #Correct:846 #Tested:996 Testing Accuracy:84.9%\r",
      "Progress:99.6% Speed(reviews/sec):1463. #Correct:847 #Tested:997 Testing Accuracy:84.9%\r",
      "Progress:99.7% Speed(reviews/sec):1463. #Correct:848 #Tested:998 Testing Accuracy:84.9%\r",
      "Progress:99.8% Speed(reviews/sec):1464. #Correct:849 #Tested:999 Testing Accuracy:84.9%\r",
      "Progress:99.9% Speed(reviews/sec):1464. #Correct:850 #Tested:1000 Testing Accuracy:85.0%"
     ]
    }
   ],
   "source": [
    "mlp.test(reviews[-1000:],labels[-1000:])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 项目 5 到此结束。\n",
    "\n",
    "# 进一步降噪<a id='lesson_6'></a>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 66,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Image(filename='sentiment_network_sparse_2.png')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[('edie', 4.6913478822291435),\n",
       " ('paulie', 4.07753744390572),\n",
       " ('felix', 3.152736022363656),\n",
       " ('polanski', 2.8233610476132043),\n",
       " ('matthau', 2.80672172860924),\n",
       " ('victoria', 2.681021528714291),\n",
       " ('mildred', 2.6026896854443837),\n",
       " ('gandhi', 2.538973871058276),\n",
       " ('flawless', 2.451005098112319),\n",
       " ('superbly', 2.26002547857525),\n",
       " ('perfection', 2.159484249353372),\n",
       " ('astaire', 2.1400661634962708),\n",
       " ('captures', 2.038619547159581),\n",
       " ('voight', 2.030170492673053),\n",
       " ('wonderfully', 2.0218960560332353),\n",
       " ('powell', 1.978345424808467),\n",
       " ('brosnan', 1.9547990964725592),\n",
       " ('lily', 1.9203768470501485),\n",
       " ('bakshi', 1.9029851043382795),\n",
       " ('lincoln', 1.9014583864844796),\n",
       " ('refreshing', 1.8551812956655511),\n",
       " ('breathtaking', 1.8481124057791867),\n",
       " ('bourne', 1.8478489358790986),\n",
       " ('lemmon', 1.8458266904983307),\n",
       " ('delightful', 1.8002701588959635),\n",
       " ('flynn', 1.7996646487351682),\n",
       " ('andrews', 1.7764919970972666),\n",
       " ('homer', 1.7692866133759964),\n",
       " ('beautifully', 1.7626953362841438),\n",
       " ('soccer', 1.7578579175523736)]"
      ]
     },
     "execution_count": 89,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# words most frequently seen in a review with a \"POSITIVE\" label\n",
    "pos_neg_ratios.most_common(30)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[('boll', -4.07781526027089),\n",
       " ('uwe', -3.921875301871158),\n",
       " ('seagal', -3.320250105858192),\n",
       " ('unwatchable', -3.0269848170580955),\n",
       " ('stinker', -2.9876839403711624),\n",
       " ('mst', -2.775383321170797),\n",
       " ('incoherent', -2.7641396677532537),\n",
       " ('unfunny', -2.5545257844967644),\n",
       " ('waste', -2.4907515123361046),\n",
       " ('blah', -2.4475792789485005),\n",
       " ('horrid', -2.371577964480997),\n",
       " ('pointless', -2.345107387713634),\n",
       " ('atrocious', -2.3187369339642556),\n",
       " ('redeeming', -2.2667790015910296),\n",
       " ('prom', -2.2601040980178784),\n",
       " ('drivel', -2.247602958576693),\n",
       " ('lousy', -2.2118080125207054),\n",
       " ('worst', -2.1930856334332267),\n",
       " ('laughable', -2.172468615469592),\n",
       " ('awful', -2.1385076866397488),\n",
       " ('poorly', -2.132613384420701),\n",
       " ('wasting', -2.117815554561451),\n",
       " ('remotely', -2.111046881095167),\n",
       " ('existent', -2.0024805005437076),\n",
       " ('boredom', -1.9241486572738005),\n",
       " ('miserably', -1.921661093801999),\n",
       " ('sucks', -1.9166645809588516),\n",
       " ('uninspired', -1.9131499212248517),\n",
       " ('lame', -1.9117232884159072),\n",
       " ('insult', -1.908532376937626)]"
      ]
     },
     "execution_count": 68,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# words most frequently seen in a review with a \"NEGATIVE\" label\n",
    "list(reversed(pos_neg_ratios.most_common()))[0:30]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "\n",
       "    <div class=\"bk-root\">\n",
       "        <a href=\"https://bokeh.pydata.org\" target=\"_blank\" class=\"bk-logo bk-logo-small bk-logo-notebook\"></a>\n",
       "        <span id=\"1001\">Loading BokehJS ...</span>\n",
       "    </div>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/javascript": [
       "\n",
       "(function(root) {\n",
       "  function now() {\n",
       "    return new Date();\n",
       "  }\n",
       "\n",
       "  var force = true;\n",
       "\n",
       "  if (typeof (root._bokeh_onload_callbacks) === \"undefined\" || force === true) {\n",
       "    root._bokeh_onload_callbacks = [];\n",
       "    root._bokeh_is_loading = undefined;\n",
       "  }\n",
       "\n",
       "  var JS_MIME_TYPE = 'application/javascript';\n",
       "  var HTML_MIME_TYPE = 'text/html';\n",
       "  var EXEC_MIME_TYPE = 'application/vnd.bokehjs_exec.v0+json';\n",
       "  var CLASS_NAME = 'output_bokeh rendered_html';\n",
       "\n",
       "  /**\n",
       "   * Render data to the DOM node\n",
       "   */\n",
       "  function render(props, node) {\n",
       "    var script = document.createElement(\"script\");\n",
       "    node.appendChild(script);\n",
       "  }\n",
       "\n",
       "  /**\n",
       "   * Handle when an output is cleared or removed\n",
       "   */\n",
       "  function handleClearOutput(event, handle) {\n",
       "    var cell = handle.cell;\n",
       "\n",
       "    var id = cell.output_area._bokeh_element_id;\n",
       "    var server_id = cell.output_area._bokeh_server_id;\n",
       "    // Clean up Bokeh references\n",
       "    if (id != null && id in Bokeh.index) {\n",
       "      Bokeh.index[id].model.document.clear();\n",
       "      delete Bokeh.index[id];\n",
       "    }\n",
       "\n",
       "    if (server_id !== undefined) {\n",
       "      // Clean up Bokeh references\n",
       "      var cmd = \"from bokeh.io.state import curstate; print(curstate().uuid_to_server['\" + server_id + \"'].get_sessions()[0].document.roots[0]._id)\";\n",
       "      cell.notebook.kernel.execute(cmd, {\n",
       "        iopub: {\n",
       "          output: function(msg) {\n",
       "            var id = msg.content.text.trim();\n",
       "            if (id in Bokeh.index) {\n",
       "              Bokeh.index[id].model.document.clear();\n",
       "              delete Bokeh.index[id];\n",
       "            }\n",
       "          }\n",
       "        }\n",
       "      });\n",
       "      // Destroy server and session\n",
       "      var cmd = \"import bokeh.io.notebook as ion; ion.destroy_server('\" + server_id + \"')\";\n",
       "      cell.notebook.kernel.execute(cmd);\n",
       "    }\n",
       "  }\n",
       "\n",
       "  /**\n",
       "   * Handle when a new output is added\n",
       "   */\n",
       "  function handleAddOutput(event, handle) {\n",
       "    var output_area = handle.output_area;\n",
       "    var output = handle.output;\n",
       "\n",
       "    // limit handleAddOutput to display_data with EXEC_MIME_TYPE content only\n",
       "    if ((output.output_type != \"display_data\") || (!output.data.hasOwnProperty(EXEC_MIME_TYPE))) {\n",
       "      return\n",
       "    }\n",
       "\n",
       "    var toinsert = output_area.element.find(\".\" + CLASS_NAME.split(' ')[0]);\n",
       "\n",
       "    if (output.metadata[EXEC_MIME_TYPE][\"id\"] !== undefined) {\n",
       "      toinsert[toinsert.length - 1].firstChild.textContent = output.data[JS_MIME_TYPE];\n",
       "      // store reference to embed id on output_area\n",
       "      output_area._bokeh_element_id = output.metadata[EXEC_MIME_TYPE][\"id\"];\n",
       "    }\n",
       "    if (output.metadata[EXEC_MIME_TYPE][\"server_id\"] !== undefined) {\n",
       "      var bk_div = document.createElement(\"div\");\n",
       "      bk_div.innerHTML = output.data[HTML_MIME_TYPE];\n",
       "      var script_attrs = bk_div.children[0].attributes;\n",
       "      for (var i = 0; i < script_attrs.length; i++) {\n",
       "        toinsert[toinsert.length - 1].firstChild.setAttribute(script_attrs[i].name, script_attrs[i].value);\n",
       "      }\n",
       "      // store reference to server id on output_area\n",
       "      output_area._bokeh_server_id = output.metadata[EXEC_MIME_TYPE][\"server_id\"];\n",
       "    }\n",
       "  }\n",
       "\n",
       "  function register_renderer(events, OutputArea) {\n",
       "\n",
       "    function append_mime(data, metadata, element) {\n",
       "      // create a DOM node to render to\n",
       "      var toinsert = this.create_output_subarea(\n",
       "        metadata,\n",
       "        CLASS_NAME,\n",
       "        EXEC_MIME_TYPE\n",
       "      );\n",
       "      this.keyboard_manager.register_events(toinsert);\n",
       "      // Render to node\n",
       "      var props = {data: data, metadata: metadata[EXEC_MIME_TYPE]};\n",
       "      render(props, toinsert[toinsert.length - 1]);\n",
       "      element.append(toinsert);\n",
       "      return toinsert\n",
       "    }\n",
       "\n",
       "    /* Handle when an output is cleared or removed */\n",
       "    events.on('clear_output.CodeCell', handleClearOutput);\n",
       "    events.on('delete.Cell', handleClearOutput);\n",
       "\n",
       "    /* Handle when a new output is added */\n",
       "    events.on('output_added.OutputArea', handleAddOutput);\n",
       "\n",
       "    /**\n",
       "     * Register the mime type and append_mime function with output_area\n",
       "     */\n",
       "    OutputArea.prototype.register_mime_type(EXEC_MIME_TYPE, append_mime, {\n",
       "      /* Is output safe? */\n",
       "      safe: true,\n",
       "      /* Index of renderer in `output_area.display_order` */\n",
       "      index: 0\n",
       "    });\n",
       "  }\n",
       "\n",
       "  // register the mime type if in Jupyter Notebook environment and previously unregistered\n",
       "  if (root.Jupyter !== undefined) {\n",
       "    var events = require('base/js/events');\n",
       "    var OutputArea = require('notebook/js/outputarea').OutputArea;\n",
       "\n",
       "    if (OutputArea.prototype.mime_types().indexOf(EXEC_MIME_TYPE) == -1) {\n",
       "      register_renderer(events, OutputArea);\n",
       "    }\n",
       "  }\n",
       "\n",
       "  \n",
       "  if (typeof (root._bokeh_timeout) === \"undefined\" || force === true) {\n",
       "    root._bokeh_timeout = Date.now() + 5000;\n",
       "    root._bokeh_failed_load = false;\n",
       "  }\n",
       "\n",
       "  var NB_LOAD_WARNING = {'data': {'text/html':\n",
       "     \"<div style='background-color: #fdd'>\\n\"+\n",
       "     \"<p>\\n\"+\n",
       "     \"BokehJS does not appear to have successfully loaded. If loading BokehJS from CDN, this \\n\"+\n",
       "     \"may be due to a slow or bad network connection. Possible fixes:\\n\"+\n",
       "     \"</p>\\n\"+\n",
       "     \"<ul>\\n\"+\n",
       "     \"<li>re-rerun `output_notebook()` to attempt to load from CDN again, or</li>\\n\"+\n",
       "     \"<li>use INLINE resources instead, as so:</li>\\n\"+\n",
       "     \"</ul>\\n\"+\n",
       "     \"<code>\\n\"+\n",
       "     \"from bokeh.resources import INLINE\\n\"+\n",
       "     \"output_notebook(resources=INLINE)\\n\"+\n",
       "     \"</code>\\n\"+\n",
       "     \"</div>\"}};\n",
       "\n",
       "  function display_loaded() {\n",
       "    var el = document.getElementById(\"1001\");\n",
       "    if (el != null) {\n",
       "      el.textContent = \"BokehJS is loading...\";\n",
       "    }\n",
       "    if (root.Bokeh !== undefined) {\n",
       "      if (el != null) {\n",
       "        el.textContent = \"BokehJS \" + root.Bokeh.version + \" successfully loaded.\";\n",
       "      }\n",
       "    } else if (Date.now() < root._bokeh_timeout) {\n",
       "      setTimeout(display_loaded, 100)\n",
       "    }\n",
       "  }\n",
       "\n",
       "\n",
       "  function run_callbacks() {\n",
       "    try {\n",
       "      root._bokeh_onload_callbacks.forEach(function(callback) { callback() });\n",
       "    }\n",
       "    finally {\n",
       "      delete root._bokeh_onload_callbacks\n",
       "    }\n",
       "    console.info(\"Bokeh: all callbacks have finished\");\n",
       "  }\n",
       "\n",
       "  function load_libs(js_urls, callback) {\n",
       "    root._bokeh_onload_callbacks.push(callback);\n",
       "    if (root._bokeh_is_loading > 0) {\n",
       "      console.log(\"Bokeh: BokehJS is being loaded, scheduling callback at\", now());\n",
       "      return null;\n",
       "    }\n",
       "    if (js_urls == null || js_urls.length === 0) {\n",
       "      run_callbacks();\n",
       "      return null;\n",
       "    }\n",
       "    console.log(\"Bokeh: BokehJS not loaded, scheduling load and callback at\", now());\n",
       "    root._bokeh_is_loading = js_urls.length;\n",
       "    for (var i = 0; i < js_urls.length; i++) {\n",
       "      var url = js_urls[i];\n",
       "      var s = document.createElement('script');\n",
       "      s.src = url;\n",
       "      s.async = false;\n",
       "      s.onreadystatechange = s.onload = function() {\n",
       "        root._bokeh_is_loading--;\n",
       "        if (root._bokeh_is_loading === 0) {\n",
       "          console.log(\"Bokeh: all BokehJS libraries loaded\");\n",
       "          run_callbacks()\n",
       "        }\n",
       "      };\n",
       "      s.onerror = function() {\n",
       "        console.warn(\"failed to load library \" + url);\n",
       "      };\n",
       "      console.log(\"Bokeh: injecting script tag for BokehJS library: \", url);\n",
       "      document.getElementsByTagName(\"head\")[0].appendChild(s);\n",
       "    }\n",
       "  };var element = document.getElementById(\"1001\");\n",
       "  if (element == null) {\n",
       "    console.log(\"Bokeh: ERROR: autoload.js configured with elementid '1001' but no matching script tag was found. \")\n",
       "    return false;\n",
       "  }\n",
       "\n",
       "  var js_urls = [\"https://cdn.pydata.org/bokeh/release/bokeh-1.0.4.min.js\", \"https://cdn.pydata.org/bokeh/release/bokeh-widgets-1.0.4.min.js\", \"https://cdn.pydata.org/bokeh/release/bokeh-tables-1.0.4.min.js\", \"https://cdn.pydata.org/bokeh/release/bokeh-gl-1.0.4.min.js\"];\n",
       "\n",
       "  var inline_js = [\n",
       "    function(Bokeh) {\n",
       "      Bokeh.set_log_level(\"info\");\n",
       "    },\n",
       "    \n",
       "    function(Bokeh) {\n",
       "      \n",
       "    },\n",
       "    function(Bokeh) {\n",
       "      console.log(\"Bokeh: injecting CSS: https://cdn.pydata.org/bokeh/release/bokeh-1.0.4.min.css\");\n",
       "      Bokeh.embed.inject_css(\"https://cdn.pydata.org/bokeh/release/bokeh-1.0.4.min.css\");\n",
       "      console.log(\"Bokeh: injecting CSS: https://cdn.pydata.org/bokeh/release/bokeh-widgets-1.0.4.min.css\");\n",
       "      Bokeh.embed.inject_css(\"https://cdn.pydata.org/bokeh/release/bokeh-widgets-1.0.4.min.css\");\n",
       "      console.log(\"Bokeh: injecting CSS: https://cdn.pydata.org/bokeh/release/bokeh-tables-1.0.4.min.css\");\n",
       "      Bokeh.embed.inject_css(\"https://cdn.pydata.org/bokeh/release/bokeh-tables-1.0.4.min.css\");\n",
       "    }\n",
       "  ];\n",
       "\n",
       "  function run_inline_js() {\n",
       "    \n",
       "    if ((root.Bokeh !== undefined) || (force === true)) {\n",
       "      for (var i = 0; i < inline_js.length; i++) {\n",
       "        inline_js[i].call(root, root.Bokeh);\n",
       "      }if (force === true) {\n",
       "        display_loaded();\n",
       "      }} else if (Date.now() < root._bokeh_timeout) {\n",
       "      setTimeout(run_inline_js, 100);\n",
       "    } else if (!root._bokeh_failed_load) {\n",
       "      console.log(\"Bokeh: BokehJS failed to load within specified timeout.\");\n",
       "      root._bokeh_failed_load = true;\n",
       "    } else if (force !== true) {\n",
       "      var cell = $(document.getElementById(\"1001\")).parents('.cell').data().cell;\n",
       "      cell.output_area.append_execute_result(NB_LOAD_WARNING)\n",
       "    }\n",
       "\n",
       "  }\n",
       "\n",
       "  if (root._bokeh_is_loading === 0) {\n",
       "    console.log(\"Bokeh: BokehJS loaded, going straight to plotting\");\n",
       "    run_inline_js();\n",
       "  } else {\n",
       "    load_libs(js_urls, function() {\n",
       "      console.log(\"Bokeh: BokehJS plotting callback run at\", now());\n",
       "      run_inline_js();\n",
       "    });\n",
       "  }\n",
       "}(window));"
      ],
      "application/vnd.bokehjs_load.v0+json": "\n(function(root) {\n  function now() {\n    return new Date();\n  }\n\n  var force = true;\n\n  if (typeof (root._bokeh_onload_callbacks) === \"undefined\" || force === true) {\n    root._bokeh_onload_callbacks = [];\n    root._bokeh_is_loading = undefined;\n  }\n\n  \n\n  \n  if (typeof (root._bokeh_timeout) === \"undefined\" || force === true) {\n    root._bokeh_timeout = Date.now() + 5000;\n    root._bokeh_failed_load = false;\n  }\n\n  var NB_LOAD_WARNING = {'data': {'text/html':\n     \"<div style='background-color: #fdd'>\\n\"+\n     \"<p>\\n\"+\n     \"BokehJS does not appear to have successfully loaded. If loading BokehJS from CDN, this \\n\"+\n     \"may be due to a slow or bad network connection. Possible fixes:\\n\"+\n     \"</p>\\n\"+\n     \"<ul>\\n\"+\n     \"<li>re-rerun `output_notebook()` to attempt to load from CDN again, or</li>\\n\"+\n     \"<li>use INLINE resources instead, as so:</li>\\n\"+\n     \"</ul>\\n\"+\n     \"<code>\\n\"+\n     \"from bokeh.resources import INLINE\\n\"+\n     \"output_notebook(resources=INLINE)\\n\"+\n     \"</code>\\n\"+\n     \"</div>\"}};\n\n  function display_loaded() {\n    var el = document.getElementById(\"1001\");\n    if (el != null) {\n      el.textContent = \"BokehJS is loading...\";\n    }\n    if (root.Bokeh !== undefined) {\n      if (el != null) {\n        el.textContent = \"BokehJS \" + root.Bokeh.version + \" successfully loaded.\";\n      }\n    } else if (Date.now() < root._bokeh_timeout) {\n      setTimeout(display_loaded, 100)\n    }\n  }\n\n\n  function run_callbacks() {\n    try {\n      root._bokeh_onload_callbacks.forEach(function(callback) { callback() });\n    }\n    finally {\n      delete root._bokeh_onload_callbacks\n    }\n    console.info(\"Bokeh: all callbacks have finished\");\n  }\n\n  function load_libs(js_urls, callback) {\n    root._bokeh_onload_callbacks.push(callback);\n    if (root._bokeh_is_loading > 0) {\n      console.log(\"Bokeh: BokehJS is being loaded, scheduling callback at\", now());\n      return null;\n    }\n    if (js_urls == null || js_urls.length === 0) {\n      run_callbacks();\n      return null;\n    }\n    console.log(\"Bokeh: BokehJS not loaded, scheduling load and callback at\", now());\n    root._bokeh_is_loading = js_urls.length;\n    for (var i = 0; i < js_urls.length; i++) {\n      var url = js_urls[i];\n      var s = document.createElement('script');\n      s.src = url;\n      s.async = false;\n      s.onreadystatechange = s.onload = function() {\n        root._bokeh_is_loading--;\n        if (root._bokeh_is_loading === 0) {\n          console.log(\"Bokeh: all BokehJS libraries loaded\");\n          run_callbacks()\n        }\n      };\n      s.onerror = function() {\n        console.warn(\"failed to load library \" + url);\n      };\n      console.log(\"Bokeh: injecting script tag for BokehJS library: \", url);\n      document.getElementsByTagName(\"head\")[0].appendChild(s);\n    }\n  };var element = document.getElementById(\"1001\");\n  if (element == null) {\n    console.log(\"Bokeh: ERROR: autoload.js configured with elementid '1001' but no matching script tag was found. \")\n    return false;\n  }\n\n  var js_urls = [\"https://cdn.pydata.org/bokeh/release/bokeh-1.0.4.min.js\", \"https://cdn.pydata.org/bokeh/release/bokeh-widgets-1.0.4.min.js\", \"https://cdn.pydata.org/bokeh/release/bokeh-tables-1.0.4.min.js\", \"https://cdn.pydata.org/bokeh/release/bokeh-gl-1.0.4.min.js\"];\n\n  var inline_js = [\n    function(Bokeh) {\n      Bokeh.set_log_level(\"info\");\n    },\n    \n    function(Bokeh) {\n      \n    },\n    function(Bokeh) {\n      console.log(\"Bokeh: injecting CSS: https://cdn.pydata.org/bokeh/release/bokeh-1.0.4.min.css\");\n      Bokeh.embed.inject_css(\"https://cdn.pydata.org/bokeh/release/bokeh-1.0.4.min.css\");\n      console.log(\"Bokeh: injecting CSS: https://cdn.pydata.org/bokeh/release/bokeh-widgets-1.0.4.min.css\");\n      Bokeh.embed.inject_css(\"https://cdn.pydata.org/bokeh/release/bokeh-widgets-1.0.4.min.css\");\n      console.log(\"Bokeh: injecting CSS: https://cdn.pydata.org/bokeh/release/bokeh-tables-1.0.4.min.css\");\n      Bokeh.embed.inject_css(\"https://cdn.pydata.org/bokeh/release/bokeh-tables-1.0.4.min.css\");\n    }\n  ];\n\n  function run_inline_js() {\n    \n    if ((root.Bokeh !== undefined) || (force === true)) {\n      for (var i = 0; i < inline_js.length; i++) {\n        inline_js[i].call(root, root.Bokeh);\n      }if (force === true) {\n        display_loaded();\n      }} else if (Date.now() < root._bokeh_timeout) {\n      setTimeout(run_inline_js, 100);\n    } else if (!root._bokeh_failed_load) {\n      console.log(\"Bokeh: BokehJS failed to load within specified timeout.\");\n      root._bokeh_failed_load = true;\n    } else if (force !== true) {\n      var cell = $(document.getElementById(\"1001\")).parents('.cell').data().cell;\n      cell.output_area.append_execute_result(NB_LOAD_WARNING)\n    }\n\n  }\n\n  if (root._bokeh_is_loading === 0) {\n    console.log(\"Bokeh: BokehJS loaded, going straight to plotting\");\n    run_inline_js();\n  } else {\n    load_libs(js_urls, function() {\n      console.log(\"Bokeh: BokehJS plotting callback run at\", now());\n      run_inline_js();\n    });\n  }\n}(window));"
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from bokeh.models import ColumnDataSource, LabelSet\n",
    "from bokeh.plotting import figure, show, output_file\n",
    "from bokeh.io import output_notebook\n",
    "import matplotlib\n",
    "output_notebook()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/anaconda3/lib/python3.7/site-packages/ipykernel_launcher.py:1: DeprecationWarning: The normed argument is ignored when density is provided. In future passing both will result in an error.\n",
      "  \"\"\"Entry point for launching an IPython kernel.\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "\n",
       "\n",
       "\n",
       "\n",
       "\n",
       "\n",
       "  <div class=\"bk-root\" id=\"3a1f9a0a-4649-4f2c-bb42-9f38375932fd\" data-root-id=\"1003\"></div>\n"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/javascript": [
       "(function(root) {\n",
       "  function embed_document(root) {\n",
       "    \n",
       "  var docs_json = {\"44f3b510-a40e-4687-9e39-ce3af534f83f\":{\"roots\":{\"references\":[{\"attributes\":{\"below\":[{\"id\":\"1013\",\"type\":\"LinearAxis\"}],\"left\":[{\"id\":\"1018\",\"type\":\"LinearAxis\"}],\"renderers\":[{\"id\":\"1013\",\"type\":\"LinearAxis\"},{\"id\":\"1017\",\"type\":\"Grid\"},{\"id\":\"1018\",\"type\":\"LinearAxis\"},{\"id\":\"1022\",\"type\":\"Grid\"},{\"id\":\"1035\",\"type\":\"GlyphRenderer\"}],\"title\":{\"id\":\"1002\",\"type\":\"Title\"},\"toolbar\":{\"id\":\"1027\",\"type\":\"Toolbar\"},\"toolbar_location\":\"above\",\"x_range\":{\"id\":\"1005\",\"type\":\"DataRange1d\"},\"x_scale\":{\"id\":\"1009\",\"type\":\"LinearScale\"},\"y_range\":{\"id\":\"1007\",\"type\":\"DataRange1d\"},\"y_scale\":{\"id\":\"1011\",\"type\":\"LinearScale\"}},\"id\":\"1003\",\"subtype\":\"Figure\",\"type\":\"Plot\"},{\"attributes\":{\"data_source\":{\"id\":\"1032\",\"type\":\"ColumnDataSource\"},\"glyph\":{\"id\":\"1033\",\"type\":\"Quad\"},\"hover_glyph\":null,\"muted_glyph\":null,\"nonselection_glyph\":{\"id\":\"1034\",\"type\":\"Quad\"},\"selection_glyph\":null,\"view\":{\"id\":\"1036\",\"type\":\"CDSView\"}},\"id\":\"1035\",\"type\":\"GlyphRenderer\"},{\"attributes\":{},\"id\":\"1023\",\"type\":\"PanTool\"},{\"attributes\":{\"bottom\":{\"value\":0},\"fill_alpha\":{\"value\":0.1},\"fill_color\":{\"value\":\"#1f77b4\"},\"left\":{\"field\":\"left\"},\"line_alpha\":{\"value\":0.1},\"line_color\":{\"value\":\"#1f77b4\"},\"right\":{\"field\":\"right\"},\"top\":{\"field\":\"top\"}},\"id\":\"1034\",\"type\":\"Quad\"},{\"attributes\":{},\"id\":\"1024\",\"type\":\"WheelZoomTool\"},{\"attributes\":{},\"id\":\"1025\",\"type\":\"ResetTool\"},{\"attributes\":{},\"id\":\"1026\",\"type\":\"SaveTool\"},{\"attributes\":{\"plot\":null,\"text\":\"Word Positive/Negative Affinity Distribution\"},\"id\":\"1002\",\"type\":\"Title\"},{\"attributes\":{},\"id\":\"1042\",\"type\":\"Selection\"},{\"attributes\":{\"active_drag\":\"auto\",\"active_inspect\":\"auto\",\"active_multi\":null,\"active_scroll\":\"auto\",\"active_tap\":\"auto\",\"tools\":[{\"id\":\"1023\",\"type\":\"PanTool\"},{\"id\":\"1024\",\"type\":\"WheelZoomTool\"},{\"id\":\"1025\",\"type\":\"ResetTool\"},{\"id\":\"1026\",\"type\":\"SaveTool\"}]},\"id\":\"1027\",\"type\":\"Toolbar\"},{\"attributes\":{\"callback\":null},\"id\":\"1005\",\"type\":\"DataRange1d\"},{\"attributes\":{\"source\":{\"id\":\"1032\",\"type\":\"ColumnDataSource\"}},\"id\":\"1036\",\"type\":\"CDSView\"},{\"attributes\":{\"callback\":null},\"id\":\"1007\",\"type\":\"DataRange1d\"},{\"attributes\":{},\"id\":\"1038\",\"type\":\"BasicTickFormatter\"},{\"attributes\":{},\"id\":\"1009\",\"type\":\"LinearScale\"},{\"attributes\":{},\"id\":\"1011\",\"type\":\"LinearScale\"},{\"attributes\":{},\"id\":\"1040\",\"type\":\"BasicTickFormatter\"},{\"attributes\":{\"formatter\":{\"id\":\"1038\",\"type\":\"BasicTickFormatter\"},\"plot\":{\"id\":\"1003\",\"subtype\":\"Figure\",\"type\":\"Plot\"},\"ticker\":{\"id\":\"1014\",\"type\":\"BasicTicker\"}},\"id\":\"1013\",\"type\":\"LinearAxis\"},{\"attributes\":{\"bottom\":{\"value\":0},\"fill_color\":{\"value\":\"#1f77b4\"},\"left\":{\"field\":\"left\"},\"line_color\":{\"value\":\"#555555\"},\"right\":{\"field\":\"right\"},\"top\":{\"field\":\"top\"}},\"id\":\"1033\",\"type\":\"Quad\"},{\"attributes\":{},\"id\":\"1043\",\"type\":\"UnionRenderers\"},{\"attributes\":{},\"id\":\"1014\",\"type\":\"BasicTicker\"},{\"attributes\":{\"plot\":{\"id\":\"1003\",\"subtype\":\"Figure\",\"type\":\"Plot\"},\"ticker\":{\"id\":\"1014\",\"type\":\"BasicTicker\"}},\"id\":\"1017\",\"type\":\"Grid\"},{\"attributes\":{\"callback\":null,\"data\":{\"left\":{\"__ndarray__\":\"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\",\"dtype\":\"float64\",\"shape\":[100]},\"right\":{\"__ndarray__\":\"mR7n78XrD8AeS15ELjgPwKJ31ZiWhA7AJ6RM7f7QDcCs0MNBZx0NwDH9OpbPaQzAtimy6je2C8A6Vik/oAILwL+CoJMITwrARK8X6HCbCcDJ24482ecIwE4IBpFBNAjA0jR95amAB8BYYfQ5Es0GwNyNa456GQbAYbri4uJlBcDm5lk3S7IEwGoT0Yuz/gPA8D9I4BtLA8B0bL80hJcCwPmYNons4wHAfsWt3VQwAcAD8iQyvXwAwBA9OA1Lkv+/GJYmthsr/r8i7xRf7MP8vyxIAwi9XPu/NqHxsI31+b9A+t9ZXo74v0hTzgIvJ/e/Uqy8q/+/9b9cBatU0Fj0v2Zemf2g8fK/cLeHpnGK8b94EHZPQiPwvwTTyPAleO2/GIWlQsep6r8sN4KUaNvnv0DpXuYJDeW/UJs7OKs+4r/ImjAUmeDev/D+6bfbQ9m/GGOjWx6n07+Ajrn+wRTMv9BWLEZH28C/gHx8NjKHpr+AYriruF6mP0BQe+Mo0cA/AIgInKMKzD/g30oqD6LTP7B7kYbMPtk/kBfY4onb3j+wWY+fIzziP6Cnsk2CCuU/kPXV++DY5z94Q/mpP6fqP2iRHFiede0/qO8fg/4h8D+gljHaLYnxP5g9QzFd8PI/jORUiIxX9D+Ei2bfu771P3gyeDbrJfc/cNmJjRqN+D9ogJvkSfT5P1wnrTt5W/s/VM6+kqjC/D9IddDp1yn+P0Ac4kAHkf8/nOH5Sxt8AEAWtYL3si8BQJKIC6NK4wFADFyUTuKWAkCILx36eUoDQAQDpqUR/gNAftYuUamxBED6qbf8QGUFQHR9QKjYGAZA8FDJU3DMBkBsJFL/B4AHQOb32qqfMwhAYstjVjfnCEDcnuwBz5oJQFhyda1mTgpA0kX+WP4BC0BOGYcElrULQMrsD7AtaQxARMCYW8UcDUDAkyEHXdANQDpnqrL0gw5AtjozXow3D0AwDrwJJOsPQNZwotpdTxBAlNpmsCmpEEBSRCuG9QIRQA6u71vBXBFAzBe0MY22EUCKgXgHWRASQEjrPN0kahJABlUBs/DDEkA=\",\"dtype\":\"float64\",\"shape\":[100]},\"top\":{\"__ndarray__\":\"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\",\"dtype\":\"float64\",\"shape\":[100]}},\"selected\":{\"id\":\"1042\",\"type\":\"Selection\"},\"selection_policy\":{\"id\":\"1043\",\"type\":\"UnionRenderers\"}},\"id\":\"1032\",\"type\":\"ColumnDataSource\"},{\"attributes\":{\"formatter\":{\"id\":\"1040\",\"type\":\"BasicTickFormatter\"},\"plot\":{\"id\":\"1003\",\"subtype\":\"Figure\",\"type\":\"Plot\"},\"ticker\":{\"id\":\"1019\",\"type\":\"BasicTicker\"}},\"id\":\"1018\",\"type\":\"LinearAxis\"},{\"attributes\":{},\"id\":\"1019\",\"type\":\"BasicTicker\"},{\"attributes\":{\"dimension\":1,\"plot\":{\"id\":\"1003\",\"subtype\":\"Figure\",\"type\":\"Plot\"},\"ticker\":{\"id\":\"1019\",\"type\":\"BasicTicker\"}},\"id\":\"1022\",\"type\":\"Grid\"}],\"root_ids\":[\"1003\"]},\"title\":\"Bokeh Application\",\"version\":\"1.0.4\"}};\n",
       "  var render_items = [{\"docid\":\"44f3b510-a40e-4687-9e39-ce3af534f83f\",\"roots\":{\"1003\":\"3a1f9a0a-4649-4f2c-bb42-9f38375932fd\"}}];\n",
       "  root.Bokeh.embed.embed_items_notebook(docs_json, render_items);\n",
       "\n",
       "  }\n",
       "  if (root.Bokeh !== undefined) {\n",
       "    embed_document(root);\n",
       "  } else {\n",
       "    var attempts = 0;\n",
       "    var timer = setInterval(function(root) {\n",
       "      if (root.Bokeh !== undefined) {\n",
       "        embed_document(root);\n",
       "        clearInterval(timer);\n",
       "      }\n",
       "      attempts++;\n",
       "      if (attempts > 100) {\n",
       "        console.log(\"Bokeh: ERROR: Unable to run BokehJS code because BokehJS library is missing\");\n",
       "        clearInterval(timer);\n",
       "      }\n",
       "    }, 10, root)\n",
       "  }\n",
       "})(window);"
      ],
      "application/vnd.bokehjs_exec.v0+json": ""
     },
     "metadata": {
      "application/vnd.bokehjs_exec.v0+json": {
       "id": "1003"
      }
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "hist, edges = np.histogram(list(map(lambda x:x[1],pos_neg_ratios.most_common())), density=True, bins=100, normed=True)\n",
    "\n",
    "p = figure(tools=\"pan,wheel_zoom,reset,save\",\n",
    "           toolbar_location=\"above\",\n",
    "           title=\"Word Positive/Negative Affinity Distribution\")\n",
    "p.quad(top=hist, bottom=0, left=edges[:-1], right=edges[1:], line_color=\"#555555\")\n",
    "show(p)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [],
   "source": [
    "frequency_frequency = Counter()\n",
    "\n",
    "for word, cnt in total_counts.most_common():\n",
    "    frequency_frequency[cnt] += 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/anaconda3/lib/python3.7/site-packages/ipykernel_launcher.py:1: DeprecationWarning: The normed argument is ignored when density is provided. In future passing both will result in an error.\n",
      "  \"\"\"Entry point for launching an IPython kernel.\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "\n",
       "\n",
       "\n",
       "\n",
       "\n",
       "\n",
       "  <div class=\"bk-root\" id=\"911a4e07-1466-4663-aa30-8061f92635ac\" data-root-id=\"1087\"></div>\n"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/javascript": [
       "(function(root) {\n",
       "  function embed_document(root) {\n",
       "    \n",
       "  var docs_json = {\"81272b43-f2fc-4532-b10c-41ddf6f17662\":{\"roots\":{\"references\":[{\"attributes\":{\"below\":[{\"id\":\"1097\",\"type\":\"LinearAxis\"}],\"left\":[{\"id\":\"1102\",\"type\":\"LinearAxis\"}],\"renderers\":[{\"id\":\"1097\",\"type\":\"LinearAxis\"},{\"id\":\"1101\",\"type\":\"Grid\"},{\"id\":\"1102\",\"type\":\"LinearAxis\"},{\"id\":\"1106\",\"type\":\"Grid\"},{\"id\":\"1119\",\"type\":\"GlyphRenderer\"}],\"title\":{\"id\":\"1086\",\"type\":\"Title\"},\"toolbar\":{\"id\":\"1111\",\"type\":\"Toolbar\"},\"toolbar_location\":\"above\",\"x_range\":{\"id\":\"1089\",\"type\":\"DataRange1d\"},\"x_scale\":{\"id\":\"1093\",\"type\":\"LinearScale\"},\"y_range\":{\"id\":\"1091\",\"type\":\"DataRange1d\"},\"y_scale\":{\"id\":\"1095\",\"type\":\"LinearScale\"}},\"id\":\"1087\",\"subtype\":\"Figure\",\"type\":\"Plot\"},{\"attributes\":{\"source\":{\"id\":\"1116\",\"type\":\"ColumnDataSource\"}},\"id\":\"1120\",\"type\":\"CDSView\"},{\"attributes\":{\"formatter\":{\"id\":\"1129\",\"type\":\"BasicTickFormatter\"},\"plot\":{\"id\":\"1087\",\"subtype\":\"Figure\",\"type\":\"Plot\"},\"ticker\":{\"id\":\"1098\",\"type\":\"BasicTicker\"}},\"id\":\"1097\",\"type\":\"LinearAxis\"},{\"attributes\":{\"callback\":null,\"data\":{\"left\":{\"__ndarray__\":\"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\",\"dtype\":\"float64\",\"shape\":[100]},\"right\":{\"__ndarray__\":\"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\",\"dtype\":\"float64\",\"shape\":[100]},\"top\":{\"__ndarray__\":\"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\",\"dtype\":\"float64\",\"shape\":[100]}},\"selected\":{\"id\":\"1133\",\"type\":\"Selection\"},\"selection_policy\":{\"id\":\"1134\",\"type\":\"UnionRenderers\"}},\"id\":\"1116\",\"type\":\"ColumnDataSource\"},{\"attributes\":{},\"id\":\"1095\",\"type\":\"LinearScale\"},{\"attributes\":{\"formatter\":{\"id\":\"1131\",\"type\":\"BasicTickFormatter\"},\"plot\":{\"id\":\"1087\",\"subtype\":\"Figure\",\"type\":\"Plot\"},\"ticker\":{\"id\":\"1103\",\"type\":\"BasicTicker\"}},\"id\":\"1102\",\"type\":\"LinearAxis\"},{\"attributes\":{},\"id\":\"1129\",\"type\":\"BasicTickFormatter\"},{\"attributes\":{},\"id\":\"1093\",\"type\":\"LinearScale\"},{\"attributes\":{},\"id\":\"1107\",\"type\":\"PanTool\"},{\"attributes\":{\"plot\":{\"id\":\"1087\",\"subtype\":\"Figure\",\"type\":\"Plot\"},\"ticker\":{\"id\":\"1098\",\"type\":\"BasicTicker\"}},\"id\":\"1101\",\"type\":\"Grid\"},{\"attributes\":{},\"id\":\"1108\",\"type\":\"WheelZoomTool\"},{\"attributes\":{},\"id\":\"1131\",\"type\":\"BasicTickFormatter\"},{\"attributes\":{},\"id\":\"1098\",\"type\":\"BasicTicker\"},{\"attributes\":{\"bottom\":{\"value\":0},\"fill_alpha\":{\"value\":0.1},\"fill_color\":{\"value\":\"#1f77b4\"},\"left\":{\"field\":\"left\"},\"line_alpha\":{\"value\":0.1},\"line_color\":{\"value\":\"#1f77b4\"},\"right\":{\"field\":\"right\"},\"top\":{\"field\":\"top\"}},\"id\":\"1118\",\"type\":\"Quad\"},{\"attributes\":{},\"id\":\"1103\",\"type\":\"BasicTicker\"},{\"attributes\":{\"active_drag\":\"auto\",\"active_inspect\":\"auto\",\"active_multi\":null,\"active_scroll\":\"auto\",\"active_tap\":\"auto\",\"tools\":[{\"id\":\"1107\",\"type\":\"PanTool\"},{\"id\":\"1108\",\"type\":\"WheelZoomTool\"},{\"id\":\"1109\",\"type\":\"ResetTool\"},{\"id\":\"1110\",\"type\":\"SaveTool\"}]},\"id\":\"1111\",\"type\":\"Toolbar\"},{\"attributes\":{\"bottom\":{\"value\":0},\"fill_color\":{\"value\":\"#1f77b4\"},\"left\":{\"field\":\"left\"},\"line_color\":{\"value\":\"#555555\"},\"right\":{\"field\":\"right\"},\"top\":{\"field\":\"top\"}},\"id\":\"1117\",\"type\":\"Quad\"},{\"attributes\":{\"callback\":null},\"id\":\"1089\",\"type\":\"DataRange1d\"},{\"attributes\":{},\"id\":\"1133\",\"type\":\"Selection\"},{\"attributes\":{\"data_source\":{\"id\":\"1116\",\"type\":\"ColumnDataSource\"},\"glyph\":{\"id\":\"1117\",\"type\":\"Quad\"},\"hover_glyph\":null,\"muted_glyph\":null,\"nonselection_glyph\":{\"id\":\"1118\",\"type\":\"Quad\"},\"selection_glyph\":null,\"view\":{\"id\":\"1120\",\"type\":\"CDSView\"}},\"id\":\"1119\",\"type\":\"GlyphRenderer\"},{\"attributes\":{\"dimension\":1,\"plot\":{\"id\":\"1087\",\"subtype\":\"Figure\",\"type\":\"Plot\"},\"ticker\":{\"id\":\"1103\",\"type\":\"BasicTicker\"}},\"id\":\"1106\",\"type\":\"Grid\"},{\"attributes\":{},\"id\":\"1134\",\"type\":\"UnionRenderers\"},{\"attributes\":{\"callback\":null},\"id\":\"1091\",\"type\":\"DataRange1d\"},{\"attributes\":{},\"id\":\"1110\",\"type\":\"SaveTool\"},{\"attributes\":{\"plot\":null,\"text\":\"The frequency distribution of the words in our corpus\"},\"id\":\"1086\",\"type\":\"Title\"},{\"attributes\":{},\"id\":\"1109\",\"type\":\"ResetTool\"}],\"root_ids\":[\"1087\"]},\"title\":\"Bokeh Application\",\"version\":\"1.0.4\"}};\n",
       "  var render_items = [{\"docid\":\"81272b43-f2fc-4532-b10c-41ddf6f17662\",\"roots\":{\"1087\":\"911a4e07-1466-4663-aa30-8061f92635ac\"}}];\n",
       "  root.Bokeh.embed.embed_items_notebook(docs_json, render_items);\n",
       "\n",
       "  }\n",
       "  if (root.Bokeh !== undefined) {\n",
       "    embed_document(root);\n",
       "  } else {\n",
       "    var attempts = 0;\n",
       "    var timer = setInterval(function(root) {\n",
       "      if (root.Bokeh !== undefined) {\n",
       "        embed_document(root);\n",
       "        clearInterval(timer);\n",
       "      }\n",
       "      attempts++;\n",
       "      if (attempts > 100) {\n",
       "        console.log(\"Bokeh: ERROR: Unable to run BokehJS code because BokehJS library is missing\");\n",
       "        clearInterval(timer);\n",
       "      }\n",
       "    }, 10, root)\n",
       "  }\n",
       "})(window);"
      ],
      "application/vnd.bokehjs_exec.v0+json": ""
     },
     "metadata": {
      "application/vnd.bokehjs_exec.v0+json": {
       "id": "1087"
      }
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "hist, edges = np.histogram(list(map(lambda x:x[1],frequency_frequency.most_common())), density=True, bins=100, normed=True)\n",
    "\n",
    "p = figure(tools=\"pan,wheel_zoom,reset,save\",\n",
    "           toolbar_location=\"above\",\n",
    "           title=\"The frequency distribution of the words in our corpus\")\n",
    "p.quad(top=hist, bottom=0, left=edges[:-1], right=edges[1:], line_color=\"#555555\")\n",
    "show(p)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 项目 6：有策略地删减词汇，进行降噪<a id='project_6'></a>\n",
    "\n",
    "**待办：** 进一步减小词汇表的噪声，以便提高 `SentimentNetwork` 的性能，具体操作如下：\n",
    "* 将在之前项目里创建的类 `SentimentNetwork` 复制到下方代码框中。\n",
    "* 修改  `pre_process_data`；\n",
    ">* 新增两个参数：`min_count` 和 `polarity_cutoff`\n",
    ">* 计算在评论中出现的单词的正面评价/负面评价比。（你可以利用在 notebook 其它地方写下的代码，但跟之前那些辅助代码一样，我们要把代码放到本类中。）\n",
    ">* Andrew 的解决方案只计算了出现次数不少于 50 次的单词的正面评价/负面评价比，这样能防止网络赋予罕见单词太多的感情色彩，如果你喜欢，也可以在你的解决方案里添加这一步。\n",
    ">* 修改本函数，使单词只有在词汇表中出现次数大于 `min_count` 时才得以添加到词汇表中。\n",
    ">* 修改本函数，使单词只有在正面评价/负面评价比的绝对值不小于 `polarity_cutoff`时 才得以添加到词汇表中。\n",
    "* 修改 `__init__`：\n",
    ">* 还是新增那两个参数 (`min_count` 和 `polarity_cutoff`) 并在调用 `pre_process_data` 时使用它们。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [],
   "source": [
    "import time\n",
    "import sys\n",
    "import numpy as np\n",
    "\n",
    "# Encapsulate our neural network in a class\n",
    "class SentimentNetwork:\n",
    "    ## New for Project 6: added min_count and polarity_cutoff parameters\n",
    "    def __init__(self, reviews,labels,min_count = 10,polarity_cutoff = 0.1,hidden_nodes = 10, learning_rate = 0.1):\n",
    "        \"\"\"Create a SentimenNetwork with the given settings\n",
    "        Args:\n",
    "            reviews(list) - List of reviews used for training\n",
    "            labels(list) - List of POSITIVE/NEGATIVE labels associated with the given reviews\n",
    "            min_count(int) - Words should only be added to the vocabulary \n",
    "                             if they occur more than this many times\n",
    "            polarity_cutoff(float) - The absolute value of a word's positive-to-negative\n",
    "                                     ratio must be at least this big to be considered.\n",
    "            hidden_nodes(int) - Number of nodes to create in the hidden layer\n",
    "            learning_rate(float) - Learning rate to use while training\n",
    "        \n",
    "        \"\"\"\n",
    "        # Assign a seed to our random number generator to ensure we get\n",
    "        # reproducable results during development \n",
    "        np.random.seed(1)\n",
    "\n",
    "        # process the reviews and their associated labels so that everything\n",
    "        # is ready for training\n",
    "        ## New for Project 6: added min_count and polarity_cutoff arguments to pre_process_data call\n",
    "        self.pre_process_data(reviews, labels, polarity_cutoff, min_count)\n",
    "        \n",
    "        # Build the network to have the number of hidden nodes and the learning rate that\n",
    "        # were passed into this initializer. Make the same number of input nodes as\n",
    "        # there are vocabulary words and create a single output node.\n",
    "        self.init_network(len(self.review_vocab),hidden_nodes, 1, learning_rate)\n",
    "\n",
    "    ## New for Project 6: added min_count and polarity_cutoff parameters\n",
    "    def pre_process_data(self, reviews, labels, polarity_cutoff, min_count):\n",
    "        \n",
    "        ## ----------------------------------------\n",
    "        ## New for Project 6: Calculate positive-to-negative ratios for words before\n",
    "        #                     building vocabulary\n",
    "        #\n",
    "        positive_counts = Counter()\n",
    "        negative_counts = Counter()\n",
    "        total_counts = Counter()\n",
    "\n",
    "        for i in range(len(reviews)):\n",
    "            if(labels[i] == 'POSITIVE'):\n",
    "                for word in reviews[i].split(\" \"):\n",
    "                    positive_counts[word] += 1\n",
    "                    total_counts[word] += 1\n",
    "            else:\n",
    "                for word in reviews[i].split(\" \"):\n",
    "                    negative_counts[word] += 1\n",
    "                    total_counts[word] += 1\n",
    "\n",
    "        pos_neg_ratios = Counter()\n",
    "\n",
    "        for term,cnt in list(total_counts.most_common()):\n",
    "            if(cnt >= 50):\n",
    "                pos_neg_ratio = positive_counts[term] / float(negative_counts[term]+1)\n",
    "                pos_neg_ratios[term] = pos_neg_ratio\n",
    "\n",
    "        for word,ratio in pos_neg_ratios.most_common():\n",
    "            if(ratio > 1):\n",
    "                pos_neg_ratios[word] = np.log(ratio)\n",
    "            else:\n",
    "                pos_neg_ratios[word] = -np.log((1 / (ratio + 0.01)))\n",
    "        #\n",
    "        ## end New for Project 6\n",
    "        ## ----------------------------------------\n",
    "\n",
    "        # populate review_vocab with all of the words in the given reviews\n",
    "        review_vocab = set()\n",
    "        for review in reviews:\n",
    "            for word in review.split(\" \"):\n",
    "                ## New for Project 6: only add words that occur at least min_count times\n",
    "                #                     and for words with pos/neg ratios, only add words\n",
    "                #                     that meet the polarity_cutoff\n",
    "                if(total_counts[word] > min_count):\n",
    "                    if(word in pos_neg_ratios.keys()):\n",
    "                        if((pos_neg_ratios[word] >= polarity_cutoff) or (pos_neg_ratios[word] <= -polarity_cutoff)):\n",
    "                            review_vocab.add(word)\n",
    "                    else:\n",
    "                        review_vocab.add(word)\n",
    "\n",
    "        # Convert the vocabulary set to a list so we can access words via indices\n",
    "        self.review_vocab = list(review_vocab)\n",
    "        \n",
    "        # populate label_vocab with all of the words in the given labels.\n",
    "        label_vocab = set()\n",
    "        for label in labels:\n",
    "            label_vocab.add(label)\n",
    "        \n",
    "        # Convert the label vocabulary set to a list so we can access labels via indices\n",
    "        self.label_vocab = list(label_vocab)\n",
    "        \n",
    "        # Store the sizes of the review and label vocabularies.\n",
    "        self.review_vocab_size = len(self.review_vocab)\n",
    "        self.label_vocab_size = len(self.label_vocab)\n",
    "        \n",
    "        # Create a dictionary of words in the vocabulary mapped to index positions\n",
    "        self.word2index = {}\n",
    "        for i, word in enumerate(self.review_vocab):\n",
    "            self.word2index[word] = i\n",
    "        \n",
    "        # Create a dictionary of labels mapped to index positions\n",
    "        self.label2index = {}\n",
    "        for i, label in enumerate(self.label_vocab):\n",
    "            self.label2index[label] = i\n",
    "\n",
    "    def init_network(self, input_nodes, hidden_nodes, output_nodes, learning_rate):\n",
    "        # Set number of nodes in input, hidden and output layers.\n",
    "        self.input_nodes = input_nodes\n",
    "        self.hidden_nodes = hidden_nodes\n",
    "        self.output_nodes = output_nodes\n",
    "\n",
    "        # Store the learning rate\n",
    "        self.learning_rate = learning_rate\n",
    "\n",
    "        # Initialize weights\n",
    "\n",
    "        # These are the weights between the input layer and the hidden layer.\n",
    "        self.weights_0_1 = np.zeros((self.input_nodes,self.hidden_nodes))\n",
    "\n",
    "        # These are the weights between the hidden layer and the output layer.\n",
    "        self.weights_1_2 = np.random.normal(0.0, self.output_nodes**-0.5, \n",
    "                                                (self.hidden_nodes, self.output_nodes))\n",
    "        \n",
    "        ## New for Project 5: Removed self.layer_0; added self.layer_1\n",
    "        # The input layer, a two-dimensional matrix with shape 1 x hidden_nodes\n",
    "        self.layer_1 = np.zeros((1,hidden_nodes))\n",
    "    \n",
    "    ## New for Project 5: Removed update_input_layer function\n",
    "    \n",
    "    def get_target_for_label(self,label):\n",
    "        if(label == 'POSITIVE'):\n",
    "            return 1\n",
    "        else:\n",
    "            return 0\n",
    "        \n",
    "    def sigmoid(self,x):\n",
    "        return 1 / (1 + np.exp(-x))\n",
    "    \n",
    "    def sigmoid_output_2_derivative(self,output):\n",
    "        return output * (1 - output)\n",
    "    \n",
    "    ## New for Project 5: changed name of first parameter form 'training_reviews' \n",
    "    #                     to 'training_reviews_raw'\n",
    "    def train(self, training_reviews_raw, training_labels):\n",
    "\n",
    "        ## New for Project 5: pre-process training reviews so we can deal \n",
    "        #                     directly with the indices of non-zero inputs\n",
    "        training_reviews = list()\n",
    "        for review in training_reviews_raw:\n",
    "            indices = set()\n",
    "            for word in review.split(\" \"):\n",
    "                if(word in self.word2index.keys()):\n",
    "                    indices.add(self.word2index[word])\n",
    "            training_reviews.append(list(indices))\n",
    "\n",
    "        # make sure out we have a matching number of reviews and labels\n",
    "        assert(len(training_reviews) == len(training_labels))\n",
    "        \n",
    "        # Keep track of correct predictions to display accuracy during training \n",
    "        correct_so_far = 0\n",
    "\n",
    "        # Remember when we started for printing time statistics\n",
    "        start = time.time()\n",
    "        \n",
    "        # loop through all the given reviews and run a forward and backward pass,\n",
    "        # updating weights for every item\n",
    "        for i in range(len(training_reviews)):\n",
    "            \n",
    "            # Get the next review and its correct label\n",
    "            review = training_reviews[i]\n",
    "            label = training_labels[i]\n",
    "            \n",
    "            #### Implement the forward pass here ####\n",
    "            ### Forward pass ###\n",
    "\n",
    "            ## New for Project 5: Removed call to 'update_input_layer' function\n",
    "            #                     because 'layer_0' is no longer used\n",
    "\n",
    "            # Hidden layer\n",
    "            ## New for Project 5: Add in only the weights for non-zero items\n",
    "            self.layer_1 *= 0\n",
    "            for index in review:\n",
    "                self.layer_1 += self.weights_0_1[index]\n",
    "\n",
    "            # Output layer\n",
    "            ## New for Project 5: changed to use 'self.layer_1' instead of 'local layer_1'\n",
    "            layer_2 = self.sigmoid(self.layer_1.dot(self.weights_1_2))            \n",
    "            \n",
    "            #### Implement the backward pass here ####\n",
    "            ### Backward pass ###\n",
    "\n",
    "            # Output error\n",
    "            layer_2_error = layer_2 - self.get_target_for_label(label) # Output layer error is the difference between desired target and actual output.\n",
    "            layer_2_delta = layer_2_error * self.sigmoid_output_2_derivative(layer_2)\n",
    "\n",
    "            # Backpropagated error\n",
    "            layer_1_error = layer_2_delta.dot(self.weights_1_2.T) # errors propagated to the hidden layer\n",
    "            layer_1_delta = layer_1_error # hidden layer gradients - no nonlinearity so it's the same as the error\n",
    "\n",
    "            # Update the weights\n",
    "            ## New for Project 5: changed to use 'self.layer_1' instead of local 'layer_1'\n",
    "            self.weights_1_2 -= self.layer_1.T.dot(layer_2_delta) * self.learning_rate # update hidden-to-output weights with gradient descent step\n",
    "            \n",
    "            ## New for Project 5: Only update the weights that were used in the forward pass\n",
    "            for index in review:\n",
    "                self.weights_0_1[index] -= layer_1_delta[0] * self.learning_rate # update input-to-hidden weights with gradient descent step\n",
    "\n",
    "            # Keep track of correct predictions.\n",
    "            if(layer_2 >= 0.5 and label == 'POSITIVE'):\n",
    "                correct_so_far += 1\n",
    "            elif(layer_2 < 0.5 and label == 'NEGATIVE'):\n",
    "                correct_so_far += 1\n",
    "            \n",
    "            # For debug purposes, print out our prediction accuracy and speed \n",
    "            # throughout the training process. \n",
    "            elapsed_time = float(time.time() - start)\n",
    "            reviews_per_second = i / elapsed_time if elapsed_time > 0 else 0\n",
    "            \n",
    "            if i % 500 == 0:\n",
    "                sys.stdout.write(\"\\rProgress:\" + str(100 * i/float(len(training_reviews)))[:4] \\\n",
    "                                 + \"% Speed(reviews/sec):\" + str(reviews_per_second)[0:5] \\\n",
    "                                 + \" #Correct:\" + str(correct_so_far) + \" #Trained:\" + str(i+1) \\\n",
    "                                 + \" Training Accuracy:\" + str(correct_so_far * 100 / float(i+1))[:4] + \"%\")\n",
    "                if(i % 2500 == 0):\n",
    "                    print(\"\")\n",
    "\n",
    "    def test(self, testing_reviews, testing_labels):\n",
    "        \"\"\"\n",
    "        Attempts to predict the labels for the given testing_reviews,\n",
    "        and uses the test_labels to calculate the accuracy of those predictions.\n",
    "        \"\"\"\n",
    "        \n",
    "        # keep track of how many correct predictions we make\n",
    "        correct = 0\n",
    "\n",
    "        # we'll time how many predictions per second we make\n",
    "        start = time.time()\n",
    "\n",
    "        # Loop through each of the given reviews and call run to predict\n",
    "        # its label. \n",
    "        for i in range(len(testing_reviews)):\n",
    "            pred = self.run(testing_reviews[i])\n",
    "            if(pred == testing_labels[i]):\n",
    "                correct += 1\n",
    "            \n",
    "            # For debug purposes, print out our prediction accuracy and speed \n",
    "            # throughout the prediction process. \n",
    "\n",
    "            elapsed_time = float(time.time() - start)\n",
    "            reviews_per_second = i / elapsed_time if elapsed_time > 0 else 0\n",
    "            \n",
    "            sys.stdout.write(\"\\rProgress:\" + str(100 * i/float(len(testing_reviews)))[:4] \\\n",
    "                             + \"% Speed(reviews/sec):\" + str(reviews_per_second)[0:5] \\\n",
    "                             + \" #Correct:\" + str(correct) + \" #Tested:\" + str(i+1) \\\n",
    "                             + \" Testing Accuracy:\" + str(correct * 100 / float(i+1))[:4] + \"%\")\n",
    "    \n",
    "    def run(self, review):\n",
    "        \"\"\"\n",
    "        Returns a POSITIVE or NEGATIVE prediction for the given review.\n",
    "        \"\"\"\n",
    "        # Run a forward pass through the network, like in the \"train\" function.\n",
    "        \n",
    "        ## New for Project 5: Removed call to update_input_layer function\n",
    "        #                     because layer_0 is no longer used\n",
    "\n",
    "        # Hidden layer\n",
    "        ## New for Project 5: Identify the indices used in the review and then add\n",
    "        #                     just those weights to layer_1 \n",
    "        self.layer_1 *= 0\n",
    "        unique_indices = set()\n",
    "        for word in review.lower().split(\" \"):\n",
    "            if word in self.word2index.keys():\n",
    "                unique_indices.add(self.word2index[word])\n",
    "        for index in unique_indices:\n",
    "            self.layer_1 += self.weights_0_1[index]\n",
    "        \n",
    "        # Output layer\n",
    "        ## New for Project 5: changed to use self.layer_1 instead of local layer_1\n",
    "        layer_2 = self.sigmoid(self.layer_1.dot(self.weights_1_2))\n",
    "         \n",
    "        # Return POSITIVE for values above greater-than-or-equal-to 0.5 in the output layer;\n",
    "        # return NEGATIVE for other values\n",
    "        if(layer_2[0] >= 0.5):\n",
    "            return \"POSITIVE\"\n",
    "        else:\n",
    "            return \"NEGATIVE\""
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "运行下方代码框，用较小的 polarity cutoff 参数值来训练网络。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Progress:0.0% Speed(reviews/sec):0.0 #Correct:1 #Trained:1 Training Accuracy:100.%\n",
      "Progress:10.4% Speed(reviews/sec):1369. #Correct:1994 #Trained:2501 Training Accuracy:79.7%\n",
      "Progress:20.8% Speed(reviews/sec):1346. #Correct:4063 #Trained:5001 Training Accuracy:81.2%\n",
      "Progress:31.2% Speed(reviews/sec):1348. #Correct:6176 #Trained:7501 Training Accuracy:82.3%\n",
      "Progress:41.6% Speed(reviews/sec):1340. #Correct:8336 #Trained:10001 Training Accuracy:83.3%\n",
      "Progress:52.0% Speed(reviews/sec):1340. #Correct:10501 #Trained:12501 Training Accuracy:84.0%\n",
      "Progress:62.5% Speed(reviews/sec):1342. #Correct:12641 #Trained:15001 Training Accuracy:84.2%\n",
      "Progress:72.9% Speed(reviews/sec):1340. #Correct:14782 #Trained:17501 Training Accuracy:84.4%\n",
      "Progress:83.3% Speed(reviews/sec):1337. #Correct:16954 #Trained:20001 Training Accuracy:84.7%\n",
      "Progress:93.7% Speed(reviews/sec):1333. #Correct:19143 #Trained:22501 Training Accuracy:85.0%\n",
      "Progress:97.9% Speed(reviews/sec):1334. #Correct:20005 #Trained:23501 Training Accuracy:85.1%"
     ]
    }
   ],
   "source": [
    "mlp = SentimentNetwork(reviews[:-1000],labels[:-1000],min_count=20,polarity_cutoff=0.05,learning_rate=0.01)\n",
    "mlp.train(reviews[:-1000],labels[:-1000])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "运行下方代码框以测试网络性能，结果应为"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\r",
      "Progress:0.0% Speed(reviews/sec):0.0 #Correct:1 #Tested:1 Testing Accuracy:100.%\r",
      "Progress:0.1% Speed(reviews/sec):804.4 #Correct:1 #Tested:2 Testing Accuracy:50.0%\r",
      "Progress:0.2% Speed(reviews/sec):1377. #Correct:2 #Tested:3 Testing Accuracy:66.6%\r",
      "Progress:0.3% Speed(reviews/sec):1592. #Correct:3 #Tested:4 Testing Accuracy:75.0%\r",
      "Progress:0.4% Speed(reviews/sec):1911. #Correct:4 #Tested:5 Testing Accuracy:80.0%\r",
      "Progress:0.5% Speed(reviews/sec):1947. #Correct:5 #Tested:6 Testing Accuracy:83.3%\r",
      "Progress:0.6% Speed(reviews/sec):1960. #Correct:6 #Tested:7 Testing Accuracy:85.7%\r",
      "Progress:0.7% Speed(reviews/sec):2029. #Correct:7 #Tested:8 Testing Accuracy:87.5%\r",
      "Progress:0.8% Speed(reviews/sec):2119. #Correct:8 #Tested:9 Testing Accuracy:88.8%\r",
      "Progress:0.9% Speed(reviews/sec):1735. #Correct:9 #Tested:10 Testing Accuracy:90.0%\r",
      "Progress:1.0% Speed(reviews/sec):1750. #Correct:10 #Tested:11 Testing Accuracy:90.9%\r",
      "Progress:1.1% Speed(reviews/sec):1799. #Correct:11 #Tested:12 Testing Accuracy:91.6%\r",
      "Progress:1.2% Speed(reviews/sec):1770. #Correct:12 #Tested:13 Testing Accuracy:92.3%\r",
      "Progress:1.3% Speed(reviews/sec):1662. #Correct:13 #Tested:14 Testing Accuracy:92.8%\r",
      "Progress:1.4% Speed(reviews/sec):1429. #Correct:14 #Tested:15 Testing Accuracy:93.3%\r",
      "Progress:1.5% Speed(reviews/sec):1412. #Correct:15 #Tested:16 Testing Accuracy:93.7%\r",
      "Progress:1.6% Speed(reviews/sec):1250. #Correct:16 #Tested:17 Testing Accuracy:94.1%\r",
      "Progress:1.7% Speed(reviews/sec):1208. #Correct:17 #Tested:18 Testing Accuracy:94.4%\r",
      "Progress:1.8% Speed(reviews/sec):1235. #Correct:18 #Tested:19 Testing Accuracy:94.7%\r",
      "Progress:1.9% Speed(reviews/sec):1225. #Correct:18 #Tested:20 Testing Accuracy:90.0%\r",
      "Progress:2.0% Speed(reviews/sec):1152. #Correct:19 #Tested:21 Testing Accuracy:90.4%\r",
      "Progress:2.1% Speed(reviews/sec):1165. #Correct:20 #Tested:22 Testing Accuracy:90.9%\r",
      "Progress:2.2% Speed(reviews/sec):1183. #Correct:21 #Tested:23 Testing Accuracy:91.3%\r",
      "Progress:2.3% Speed(reviews/sec):1187. #Correct:22 #Tested:24 Testing Accuracy:91.6%\r",
      "Progress:2.4% Speed(reviews/sec):1164. #Correct:23 #Tested:25 Testing Accuracy:92.0%\r",
      "Progress:2.5% Speed(reviews/sec):1188. #Correct:24 #Tested:26 Testing Accuracy:92.3%\r",
      "Progress:2.6% Speed(reviews/sec):1219. #Correct:25 #Tested:27 Testing Accuracy:92.5%\r",
      "Progress:2.7% Speed(reviews/sec):1189. #Correct:26 #Tested:28 Testing Accuracy:92.8%\r",
      "Progress:2.8% Speed(reviews/sec):1197. #Correct:27 #Tested:29 Testing Accuracy:93.1%\r",
      "Progress:2.9% Speed(reviews/sec):1210. #Correct:28 #Tested:30 Testing Accuracy:93.3%\r",
      "Progress:3.0% Speed(reviews/sec):1181. #Correct:28 #Tested:31 Testing Accuracy:90.3%\r",
      "Progress:3.1% Speed(reviews/sec):1190. #Correct:29 #Tested:32 Testing Accuracy:90.6%\r",
      "Progress:3.2% Speed(reviews/sec):1202. #Correct:30 #Tested:33 Testing Accuracy:90.9%\r",
      "Progress:3.3% Speed(reviews/sec):1221. #Correct:31 #Tested:34 Testing Accuracy:91.1%\r",
      "Progress:3.4% Speed(reviews/sec):1229. #Correct:32 #Tested:35 Testing Accuracy:91.4%\r",
      "Progress:3.5% Speed(reviews/sec):1248. #Correct:33 #Tested:36 Testing Accuracy:91.6%\r",
      "Progress:3.6% Speed(reviews/sec):1249. #Correct:34 #Tested:37 Testing Accuracy:91.8%\r",
      "Progress:3.7% Speed(reviews/sec):1275. #Correct:35 #Tested:38 Testing Accuracy:92.1%\r",
      "Progress:3.8% Speed(reviews/sec):1281. #Correct:36 #Tested:39 Testing Accuracy:92.3%\r",
      "Progress:3.9% Speed(reviews/sec):1265. #Correct:37 #Tested:40 Testing Accuracy:92.5%\r",
      "Progress:4.0% Speed(reviews/sec):1270. #Correct:38 #Tested:41 Testing Accuracy:92.6%\r",
      "Progress:4.1% Speed(reviews/sec):1284. #Correct:39 #Tested:42 Testing Accuracy:92.8%\r",
      "Progress:4.2% Speed(reviews/sec):1269. #Correct:40 #Tested:43 Testing Accuracy:93.0%\r",
      "Progress:4.3% Speed(reviews/sec):1260. #Correct:41 #Tested:44 Testing Accuracy:93.1%\r",
      "Progress:4.4% Speed(reviews/sec):1251. #Correct:42 #Tested:45 Testing Accuracy:93.3%\r",
      "Progress:4.5% Speed(reviews/sec):1222. #Correct:42 #Tested:46 Testing Accuracy:91.3%\r",
      "Progress:4.6% Speed(reviews/sec):1217. #Correct:43 #Tested:47 Testing Accuracy:91.4%\r",
      "Progress:4.7% Speed(reviews/sec):1236. #Correct:44 #Tested:48 Testing Accuracy:91.6%\r",
      "Progress:4.8% Speed(reviews/sec):1240. #Correct:45 #Tested:49 Testing Accuracy:91.8%\r",
      "Progress:4.9% Speed(reviews/sec):1253. #Correct:46 #Tested:50 Testing Accuracy:92.0%\r",
      "Progress:5.0% Speed(reviews/sec):1268. #Correct:47 #Tested:51 Testing Accuracy:92.1%\r",
      "Progress:5.1% Speed(reviews/sec):1285. #Correct:48 #Tested:52 Testing Accuracy:92.3%\r",
      "Progress:5.2% Speed(reviews/sec):1272. #Correct:49 #Tested:53 Testing Accuracy:92.4%\r",
      "Progress:5.3% Speed(reviews/sec):1280. #Correct:50 #Tested:54 Testing Accuracy:92.5%\r",
      "Progress:5.4% Speed(reviews/sec):1289. #Correct:51 #Tested:55 Testing Accuracy:92.7%\r",
      "Progress:5.5% Speed(reviews/sec):1304. #Correct:52 #Tested:56 Testing Accuracy:92.8%\r",
      "Progress:5.6% Speed(reviews/sec):1268. #Correct:53 #Tested:57 Testing Accuracy:92.9%\r",
      "Progress:5.7% Speed(reviews/sec):1276. #Correct:54 #Tested:58 Testing Accuracy:93.1%\r",
      "Progress:5.8% Speed(reviews/sec):1290. #Correct:55 #Tested:59 Testing Accuracy:93.2%\r",
      "Progress:5.9% Speed(reviews/sec):1304. #Correct:56 #Tested:60 Testing Accuracy:93.3%\r",
      "Progress:6.0% Speed(reviews/sec):1295. #Correct:57 #Tested:61 Testing Accuracy:93.4%\r",
      "Progress:6.1% Speed(reviews/sec):1299. #Correct:58 #Tested:62 Testing Accuracy:93.5%\r",
      "Progress:6.2% Speed(reviews/sec):1315. #Correct:59 #Tested:63 Testing Accuracy:93.6%\r",
      "Progress:6.3% Speed(reviews/sec):1328. #Correct:60 #Tested:64 Testing Accuracy:93.7%\r",
      "Progress:6.4% Speed(reviews/sec):1326. #Correct:60 #Tested:65 Testing Accuracy:92.3%\r",
      "Progress:6.5% Speed(reviews/sec):1335. #Correct:60 #Tested:66 Testing Accuracy:90.9%\r",
      "Progress:6.6% Speed(reviews/sec):1347. #Correct:61 #Tested:67 Testing Accuracy:91.0%\r",
      "Progress:6.7% Speed(reviews/sec):1350. #Correct:62 #Tested:68 Testing Accuracy:91.1%\r",
      "Progress:6.8% Speed(reviews/sec):1354. #Correct:63 #Tested:69 Testing Accuracy:91.3%\r",
      "Progress:6.9% Speed(reviews/sec):1357. #Correct:64 #Tested:70 Testing Accuracy:91.4%\r",
      "Progress:7.0% Speed(reviews/sec):1360. #Correct:64 #Tested:71 Testing Accuracy:90.1%\r",
      "Progress:7.1% Speed(reviews/sec):1373. #Correct:65 #Tested:72 Testing Accuracy:90.2%\r",
      "Progress:7.2% Speed(reviews/sec):1383. #Correct:66 #Tested:73 Testing Accuracy:90.4%\r",
      "Progress:7.3% Speed(reviews/sec):1366. #Correct:67 #Tested:74 Testing Accuracy:90.5%\r",
      "Progress:7.4% Speed(reviews/sec):1375. #Correct:67 #Tested:75 Testing Accuracy:89.3%\r",
      "Progress:7.5% Speed(reviews/sec):1383. #Correct:68 #Tested:76 Testing Accuracy:89.4%\r",
      "Progress:7.6% Speed(reviews/sec):1394. #Correct:69 #Tested:77 Testing Accuracy:89.6%\r",
      "Progress:7.7% Speed(reviews/sec):1404. #Correct:70 #Tested:78 Testing Accuracy:89.7%\r",
      "Progress:7.8% Speed(reviews/sec):1413. #Correct:71 #Tested:79 Testing Accuracy:89.8%\r",
      "Progress:7.9% Speed(reviews/sec):1421. #Correct:72 #Tested:80 Testing Accuracy:90.0%\r",
      "Progress:8.0% Speed(reviews/sec):1431. #Correct:73 #Tested:81 Testing Accuracy:90.1%\r",
      "Progress:8.1% Speed(reviews/sec):1415. #Correct:73 #Tested:82 Testing Accuracy:89.0%\r",
      "Progress:8.2% Speed(reviews/sec):1415. #Correct:74 #Tested:83 Testing Accuracy:89.1%\r",
      "Progress:8.3% Speed(reviews/sec):1318. #Correct:75 #Tested:84 Testing Accuracy:89.2%\r",
      "Progress:8.4% Speed(reviews/sec):1328. #Correct:76 #Tested:85 Testing Accuracy:89.4%\r",
      "Progress:8.5% Speed(reviews/sec):1334. #Correct:77 #Tested:86 Testing Accuracy:89.5%\r",
      "Progress:8.6% Speed(reviews/sec):1336. #Correct:78 #Tested:87 Testing Accuracy:89.6%\r",
      "Progress:8.7% Speed(reviews/sec):1346. #Correct:78 #Tested:88 Testing Accuracy:88.6%\r",
      "Progress:8.8% Speed(reviews/sec):1354. #Correct:79 #Tested:89 Testing Accuracy:88.7%\r",
      "Progress:8.9% Speed(reviews/sec):1343. #Correct:80 #Tested:90 Testing Accuracy:88.8%\r",
      "Progress:9.0% Speed(reviews/sec):1343. #Correct:81 #Tested:91 Testing Accuracy:89.0%\r",
      "Progress:9.1% Speed(reviews/sec):1334. #Correct:82 #Tested:92 Testing Accuracy:89.1%\r",
      "Progress:9.2% Speed(reviews/sec):1341. #Correct:83 #Tested:93 Testing Accuracy:89.2%\r",
      "Progress:9.3% Speed(reviews/sec):1347. #Correct:84 #Tested:94 Testing Accuracy:89.3%\r",
      "Progress:9.4% Speed(reviews/sec):1343. #Correct:85 #Tested:95 Testing Accuracy:89.4%\r",
      "Progress:9.5% Speed(reviews/sec):1346. #Correct:86 #Tested:96 Testing Accuracy:89.5%\r",
      "Progress:9.6% Speed(reviews/sec):1351. #Correct:87 #Tested:97 Testing Accuracy:89.6%\r",
      "Progress:9.7% Speed(reviews/sec):1347. #Correct:88 #Tested:98 Testing Accuracy:89.7%\r",
      "Progress:9.8% Speed(reviews/sec):1355. #Correct:89 #Tested:99 Testing Accuracy:89.8%\r",
      "Progress:9.9% Speed(reviews/sec):1363. #Correct:90 #Tested:100 Testing Accuracy:90.0%\r",
      "Progress:10.0% Speed(reviews/sec):1369. #Correct:91 #Tested:101 Testing Accuracy:90.0%\r",
      "Progress:10.1% Speed(reviews/sec):1364. #Correct:92 #Tested:102 Testing Accuracy:90.1%\r",
      "Progress:10.2% Speed(reviews/sec):1368. #Correct:93 #Tested:103 Testing Accuracy:90.2%\r",
      "Progress:10.3% Speed(reviews/sec):1366. #Correct:93 #Tested:104 Testing Accuracy:89.4%\r",
      "Progress:10.4% Speed(reviews/sec):1369. #Correct:94 #Tested:105 Testing Accuracy:89.5%\r",
      "Progress:10.5% Speed(reviews/sec):1378. #Correct:95 #Tested:106 Testing Accuracy:89.6%\r",
      "Progress:10.6% Speed(reviews/sec):1380. #Correct:96 #Tested:107 Testing Accuracy:89.7%\r",
      "Progress:10.7% Speed(reviews/sec):1385. #Correct:97 #Tested:108 Testing Accuracy:89.8%\r",
      "Progress:10.8% Speed(reviews/sec):1393. #Correct:98 #Tested:109 Testing Accuracy:89.9%\r",
      "Progress:10.9% Speed(reviews/sec):1392. #Correct:98 #Tested:110 Testing Accuracy:89.0%\r",
      "Progress:11.0% Speed(reviews/sec):1397. #Correct:99 #Tested:111 Testing Accuracy:89.1%\r",
      "Progress:11.1% Speed(reviews/sec):1402. #Correct:100 #Tested:112 Testing Accuracy:89.2%\r",
      "Progress:11.2% Speed(reviews/sec):1407. #Correct:101 #Tested:113 Testing Accuracy:89.3%\r",
      "Progress:11.3% Speed(reviews/sec):1408. #Correct:102 #Tested:114 Testing Accuracy:89.4%\r",
      "Progress:11.4% Speed(reviews/sec):1412. #Correct:103 #Tested:115 Testing Accuracy:89.5%\r",
      "Progress:11.5% Speed(reviews/sec):1419. #Correct:103 #Tested:116 Testing Accuracy:88.7%\r",
      "Progress:11.6% Speed(reviews/sec):1425. #Correct:104 #Tested:117 Testing Accuracy:88.8%\r",
      "Progress:11.7% Speed(reviews/sec):1426. #Correct:105 #Tested:118 Testing Accuracy:88.9%\r",
      "Progress:11.8% Speed(reviews/sec):1418. #Correct:106 #Tested:119 Testing Accuracy:89.0%\r",
      "Progress:11.9% Speed(reviews/sec):1421. #Correct:107 #Tested:120 Testing Accuracy:89.1%\r",
      "Progress:12.0% Speed(reviews/sec):1426. #Correct:108 #Tested:121 Testing Accuracy:89.2%\r",
      "Progress:12.1% Speed(reviews/sec):1430. #Correct:109 #Tested:122 Testing Accuracy:89.3%\r",
      "Progress:12.2% Speed(reviews/sec):1437. #Correct:110 #Tested:123 Testing Accuracy:89.4%\r",
      "Progress:12.3% Speed(reviews/sec):1435. #Correct:111 #Tested:124 Testing Accuracy:89.5%\r",
      "Progress:12.4% Speed(reviews/sec):1438. #Correct:112 #Tested:125 Testing Accuracy:89.6%\r",
      "Progress:12.5% Speed(reviews/sec):1429. #Correct:113 #Tested:126 Testing Accuracy:89.6%\r",
      "Progress:12.6% Speed(reviews/sec):1437. #Correct:114 #Tested:127 Testing Accuracy:89.7%\r",
      "Progress:12.7% Speed(reviews/sec):1439. #Correct:115 #Tested:128 Testing Accuracy:89.8%\r",
      "Progress:12.8% Speed(reviews/sec):1445. #Correct:116 #Tested:129 Testing Accuracy:89.9%\r",
      "Progress:12.9% Speed(reviews/sec):1451. #Correct:117 #Tested:130 Testing Accuracy:90.0%\r",
      "Progress:13.0% Speed(reviews/sec):1456. #Correct:117 #Tested:131 Testing Accuracy:89.3%\r",
      "Progress:13.1% Speed(reviews/sec):1463. #Correct:118 #Tested:132 Testing Accuracy:89.3%\r",
      "Progress:13.2% Speed(reviews/sec):1469. #Correct:119 #Tested:133 Testing Accuracy:89.4%\r",
      "Progress:13.3% Speed(reviews/sec):1475. #Correct:120 #Tested:134 Testing Accuracy:89.5%\r",
      "Progress:13.4% Speed(reviews/sec):1476. #Correct:121 #Tested:135 Testing Accuracy:89.6%\r",
      "Progress:13.5% Speed(reviews/sec):1477. #Correct:121 #Tested:136 Testing Accuracy:88.9%\r",
      "Progress:13.6% Speed(reviews/sec):1483. #Correct:122 #Tested:137 Testing Accuracy:89.0%\r",
      "Progress:13.7% Speed(reviews/sec):1489. #Correct:123 #Tested:138 Testing Accuracy:89.1%\r",
      "Progress:13.8% Speed(reviews/sec):1493. #Correct:124 #Tested:139 Testing Accuracy:89.2%\r",
      "Progress:13.9% Speed(reviews/sec):1475. #Correct:125 #Tested:140 Testing Accuracy:89.2%\r",
      "Progress:14.0% Speed(reviews/sec):1483. #Correct:126 #Tested:141 Testing Accuracy:89.3%\r",
      "Progress:14.1% Speed(reviews/sec):1481. #Correct:127 #Tested:142 Testing Accuracy:89.4%\r",
      "Progress:14.2% Speed(reviews/sec):1486. #Correct:128 #Tested:143 Testing Accuracy:89.5%\r",
      "Progress:14.3% Speed(reviews/sec):1492. #Correct:129 #Tested:144 Testing Accuracy:89.5%\r",
      "Progress:14.4% Speed(reviews/sec):1493. #Correct:130 #Tested:145 Testing Accuracy:89.6%\r",
      "Progress:14.5% Speed(reviews/sec):1493. #Correct:131 #Tested:146 Testing Accuracy:89.7%\r",
      "Progress:14.6% Speed(reviews/sec):1492. #Correct:132 #Tested:147 Testing Accuracy:89.7%\r",
      "Progress:14.7% Speed(reviews/sec):1497. #Correct:133 #Tested:148 Testing Accuracy:89.8%\r",
      "Progress:14.8% Speed(reviews/sec):1501. #Correct:134 #Tested:149 Testing Accuracy:89.9%\r",
      "Progress:14.9% Speed(reviews/sec):1506. #Correct:135 #Tested:150 Testing Accuracy:90.0%\r",
      "Progress:15.0% Speed(reviews/sec):1509. #Correct:136 #Tested:151 Testing Accuracy:90.0%\r",
      "Progress:15.1% Speed(reviews/sec):1514. #Correct:137 #Tested:152 Testing Accuracy:90.1%\r",
      "Progress:15.2% Speed(reviews/sec):1507. #Correct:138 #Tested:153 Testing Accuracy:90.1%\r",
      "Progress:15.3% Speed(reviews/sec):1503. #Correct:139 #Tested:154 Testing Accuracy:90.2%\r",
      "Progress:15.4% Speed(reviews/sec):1506. #Correct:140 #Tested:155 Testing Accuracy:90.3%\r",
      "Progress:15.5% Speed(reviews/sec):1510. #Correct:141 #Tested:156 Testing Accuracy:90.3%\r",
      "Progress:15.6% Speed(reviews/sec):1509. #Correct:142 #Tested:157 Testing Accuracy:90.4%\r",
      "Progress:15.7% Speed(reviews/sec):1498. #Correct:142 #Tested:158 Testing Accuracy:89.8%\r",
      "Progress:15.8% Speed(reviews/sec):1503. #Correct:143 #Tested:159 Testing Accuracy:89.9%\r",
      "Progress:15.9% Speed(reviews/sec):1507. #Correct:144 #Tested:160 Testing Accuracy:90.0%\r",
      "Progress:16.0% Speed(reviews/sec):1508. #Correct:144 #Tested:161 Testing Accuracy:89.4%\r",
      "Progress:16.1% Speed(reviews/sec):1510. #Correct:144 #Tested:162 Testing Accuracy:88.8%\r",
      "Progress:16.2% Speed(reviews/sec):1515. #Correct:145 #Tested:163 Testing Accuracy:88.9%\r",
      "Progress:16.3% Speed(reviews/sec):1519. #Correct:146 #Tested:164 Testing Accuracy:89.0%\r",
      "Progress:16.4% Speed(reviews/sec):1524. #Correct:147 #Tested:165 Testing Accuracy:89.0%\r",
      "Progress:16.5% Speed(reviews/sec):1524. #Correct:148 #Tested:166 Testing Accuracy:89.1%\r",
      "Progress:16.6% Speed(reviews/sec):1525. #Correct:149 #Tested:167 Testing Accuracy:89.2%\r",
      "Progress:16.7% Speed(reviews/sec):1526. #Correct:150 #Tested:168 Testing Accuracy:89.2%\r",
      "Progress:16.8% Speed(reviews/sec):1531. #Correct:151 #Tested:169 Testing Accuracy:89.3%\r",
      "Progress:16.9% Speed(reviews/sec):1533. #Correct:152 #Tested:170 Testing Accuracy:89.4%\r",
      "Progress:17.0% Speed(reviews/sec):1533. #Correct:152 #Tested:171 Testing Accuracy:88.8%\r",
      "Progress:17.1% Speed(reviews/sec):1536. #Correct:153 #Tested:172 Testing Accuracy:88.9%\r",
      "Progress:17.2% Speed(reviews/sec):1534. #Correct:154 #Tested:173 Testing Accuracy:89.0%\r",
      "Progress:17.3% Speed(reviews/sec):1516. #Correct:154 #Tested:174 Testing Accuracy:88.5%\r",
      "Progress:17.4% Speed(reviews/sec):1509. #Correct:155 #Tested:175 Testing Accuracy:88.5%\r",
      "Progress:17.5% Speed(reviews/sec):1510. #Correct:156 #Tested:176 Testing Accuracy:88.6%\r",
      "Progress:17.6% Speed(reviews/sec):1513. #Correct:157 #Tested:177 Testing Accuracy:88.7%\r",
      "Progress:17.7% Speed(reviews/sec):1507. #Correct:158 #Tested:178 Testing Accuracy:88.7%\r",
      "Progress:17.8% Speed(reviews/sec):1512. #Correct:159 #Tested:179 Testing Accuracy:88.8%\r",
      "Progress:17.9% Speed(reviews/sec):1515. #Correct:160 #Tested:180 Testing Accuracy:88.8%\r",
      "Progress:18.0% Speed(reviews/sec):1513. #Correct:161 #Tested:181 Testing Accuracy:88.9%\r",
      "Progress:18.1% Speed(reviews/sec):1510. #Correct:162 #Tested:182 Testing Accuracy:89.0%\r",
      "Progress:18.2% Speed(reviews/sec):1513. #Correct:163 #Tested:183 Testing Accuracy:89.0%\r",
      "Progress:18.3% Speed(reviews/sec):1516. #Correct:164 #Tested:184 Testing Accuracy:89.1%\r",
      "Progress:18.4% Speed(reviews/sec):1515. #Correct:165 #Tested:185 Testing Accuracy:89.1%\r",
      "Progress:18.5% Speed(reviews/sec):1517. #Correct:166 #Tested:186 Testing Accuracy:89.2%\r",
      "Progress:18.6% Speed(reviews/sec):1522. #Correct:166 #Tested:187 Testing Accuracy:88.7%\r",
      "Progress:18.7% Speed(reviews/sec):1524. #Correct:167 #Tested:188 Testing Accuracy:88.8%\r",
      "Progress:18.8% Speed(reviews/sec):1526. #Correct:168 #Tested:189 Testing Accuracy:88.8%\r",
      "Progress:18.9% Speed(reviews/sec):1527. #Correct:169 #Tested:190 Testing Accuracy:88.9%\r",
      "Progress:19.0% Speed(reviews/sec):1518. #Correct:170 #Tested:191 Testing Accuracy:89.0%\r",
      "Progress:19.1% Speed(reviews/sec):1521. #Correct:171 #Tested:192 Testing Accuracy:89.0%\r",
      "Progress:19.2% Speed(reviews/sec):1525. #Correct:171 #Tested:193 Testing Accuracy:88.6%\r",
      "Progress:19.3% Speed(reviews/sec):1524. #Correct:172 #Tested:194 Testing Accuracy:88.6%\r",
      "Progress:19.4% Speed(reviews/sec):1529. #Correct:173 #Tested:195 Testing Accuracy:88.7%\r",
      "Progress:19.5% Speed(reviews/sec):1530. #Correct:173 #Tested:196 Testing Accuracy:88.2%\r",
      "Progress:19.6% Speed(reviews/sec):1532. #Correct:174 #Tested:197 Testing Accuracy:88.3%\r",
      "Progress:19.7% Speed(reviews/sec):1527. #Correct:175 #Tested:198 Testing Accuracy:88.3%\r",
      "Progress:19.8% Speed(reviews/sec):1532. #Correct:176 #Tested:199 Testing Accuracy:88.4%\r",
      "Progress:19.9% Speed(reviews/sec):1533. #Correct:177 #Tested:200 Testing Accuracy:88.5%\r",
      "Progress:20.0% Speed(reviews/sec):1534. #Correct:178 #Tested:201 Testing Accuracy:88.5%\r",
      "Progress:20.1% Speed(reviews/sec):1536. #Correct:179 #Tested:202 Testing Accuracy:88.6%\r",
      "Progress:20.2% Speed(reviews/sec):1540. #Correct:180 #Tested:203 Testing Accuracy:88.6%\r",
      "Progress:20.3% Speed(reviews/sec):1542. #Correct:181 #Tested:204 Testing Accuracy:88.7%\r",
      "Progress:20.4% Speed(reviews/sec):1546. #Correct:182 #Tested:205 Testing Accuracy:88.7%\r",
      "Progress:20.5% Speed(reviews/sec):1549. #Correct:183 #Tested:206 Testing Accuracy:88.8%\r",
      "Progress:20.6% Speed(reviews/sec):1551. #Correct:184 #Tested:207 Testing Accuracy:88.8%\r",
      "Progress:20.7% Speed(reviews/sec):1553. #Correct:185 #Tested:208 Testing Accuracy:88.9%\r",
      "Progress:20.8% Speed(reviews/sec):1556. #Correct:186 #Tested:209 Testing Accuracy:88.9%\r",
      "Progress:20.9% Speed(reviews/sec):1555. #Correct:186 #Tested:210 Testing Accuracy:88.5%\r",
      "Progress:21.0% Speed(reviews/sec):1543. #Correct:186 #Tested:211 Testing Accuracy:88.1%\r",
      "Progress:21.1% Speed(reviews/sec):1547. #Correct:187 #Tested:212 Testing Accuracy:88.2%\r",
      "Progress:21.2% Speed(reviews/sec):1551. #Correct:188 #Tested:213 Testing Accuracy:88.2%\r",
      "Progress:21.3% Speed(reviews/sec):1544. #Correct:189 #Tested:214 Testing Accuracy:88.3%\r",
      "Progress:21.4% Speed(reviews/sec):1549. #Correct:190 #Tested:215 Testing Accuracy:88.3%\r",
      "Progress:21.5% Speed(reviews/sec):1548. #Correct:191 #Tested:216 Testing Accuracy:88.4%\r",
      "Progress:21.6% Speed(reviews/sec):1551. #Correct:192 #Tested:217 Testing Accuracy:88.4%\r",
      "Progress:21.7% Speed(reviews/sec):1552. #Correct:193 #Tested:218 Testing Accuracy:88.5%\r",
      "Progress:21.8% Speed(reviews/sec):1552. #Correct:194 #Tested:219 Testing Accuracy:88.5%\r",
      "Progress:21.9% Speed(reviews/sec):1556. #Correct:195 #Tested:220 Testing Accuracy:88.6%\r",
      "Progress:22.0% Speed(reviews/sec):1556. #Correct:196 #Tested:221 Testing Accuracy:88.6%\r",
      "Progress:22.1% Speed(reviews/sec):1559. #Correct:197 #Tested:222 Testing Accuracy:88.7%\r",
      "Progress:22.2% Speed(reviews/sec):1561. #Correct:198 #Tested:223 Testing Accuracy:88.7%\r",
      "Progress:22.3% Speed(reviews/sec):1563. #Correct:199 #Tested:224 Testing Accuracy:88.8%\r",
      "Progress:22.4% Speed(reviews/sec):1554. #Correct:200 #Tested:225 Testing Accuracy:88.8%\r",
      "Progress:22.5% Speed(reviews/sec):1555. #Correct:201 #Tested:226 Testing Accuracy:88.9%\r",
      "Progress:22.6% Speed(reviews/sec):1548. #Correct:202 #Tested:227 Testing Accuracy:88.9%\r",
      "Progress:22.7% Speed(reviews/sec):1547. #Correct:203 #Tested:228 Testing Accuracy:89.0%\r",
      "Progress:22.8% Speed(reviews/sec):1549. #Correct:204 #Tested:229 Testing Accuracy:89.0%\r",
      "Progress:22.9% Speed(reviews/sec):1549. #Correct:205 #Tested:230 Testing Accuracy:89.1%\r",
      "Progress:23.0% Speed(reviews/sec):1546. #Correct:206 #Tested:231 Testing Accuracy:89.1%\r",
      "Progress:23.1% Speed(reviews/sec):1523. #Correct:207 #Tested:232 Testing Accuracy:89.2%\r",
      "Progress:23.2% Speed(reviews/sec):1527. #Correct:208 #Tested:233 Testing Accuracy:89.2%\r",
      "Progress:23.3% Speed(reviews/sec):1529. #Correct:209 #Tested:234 Testing Accuracy:89.3%\r",
      "Progress:23.4% Speed(reviews/sec):1525. #Correct:210 #Tested:235 Testing Accuracy:89.3%\r",
      "Progress:23.5% Speed(reviews/sec):1528. #Correct:211 #Tested:236 Testing Accuracy:89.4%\r",
      "Progress:23.6% Speed(reviews/sec):1526. #Correct:212 #Tested:237 Testing Accuracy:89.4%\r",
      "Progress:23.7% Speed(reviews/sec):1527. #Correct:213 #Tested:238 Testing Accuracy:89.4%\r",
      "Progress:23.8% Speed(reviews/sec):1531. #Correct:214 #Tested:239 Testing Accuracy:89.5%\r",
      "Progress:23.9% Speed(reviews/sec):1533. #Correct:215 #Tested:240 Testing Accuracy:89.5%\r",
      "Progress:24.0% Speed(reviews/sec):1533. #Correct:216 #Tested:241 Testing Accuracy:89.6%\r",
      "Progress:24.1% Speed(reviews/sec):1534. #Correct:216 #Tested:242 Testing Accuracy:89.2%\r",
      "Progress:24.2% Speed(reviews/sec):1536. #Correct:217 #Tested:243 Testing Accuracy:89.3%\r",
      "Progress:24.3% Speed(reviews/sec):1534. #Correct:218 #Tested:244 Testing Accuracy:89.3%\r",
      "Progress:24.4% Speed(reviews/sec):1534. #Correct:219 #Tested:245 Testing Accuracy:89.3%\r",
      "Progress:24.5% Speed(reviews/sec):1537. #Correct:220 #Tested:246 Testing Accuracy:89.4%\r",
      "Progress:24.6% Speed(reviews/sec):1540. #Correct:221 #Tested:247 Testing Accuracy:89.4%\r",
      "Progress:24.7% Speed(reviews/sec):1541. #Correct:222 #Tested:248 Testing Accuracy:89.5%\r",
      "Progress:24.8% Speed(reviews/sec):1546. #Correct:223 #Tested:249 Testing Accuracy:89.5%\r",
      "Progress:24.9% Speed(reviews/sec):1548. #Correct:223 #Tested:250 Testing Accuracy:89.2%\r",
      "Progress:25.0% Speed(reviews/sec):1551. #Correct:224 #Tested:251 Testing Accuracy:89.2%\r",
      "Progress:25.1% Speed(reviews/sec):1555. #Correct:225 #Tested:252 Testing Accuracy:89.2%\r",
      "Progress:25.2% Speed(reviews/sec):1557. #Correct:226 #Tested:253 Testing Accuracy:89.3%\r",
      "Progress:25.3% Speed(reviews/sec):1560. #Correct:227 #Tested:254 Testing Accuracy:89.3%\r",
      "Progress:25.4% Speed(reviews/sec):1563. #Correct:228 #Tested:255 Testing Accuracy:89.4%\r",
      "Progress:25.5% Speed(reviews/sec):1566. #Correct:229 #Tested:256 Testing Accuracy:89.4%\r",
      "Progress:25.6% Speed(reviews/sec):1571. #Correct:230 #Tested:257 Testing Accuracy:89.4%\r",
      "Progress:25.7% Speed(reviews/sec):1574. #Correct:231 #Tested:258 Testing Accuracy:89.5%\r",
      "Progress:25.8% Speed(reviews/sec):1578. #Correct:232 #Tested:259 Testing Accuracy:89.5%\r",
      "Progress:25.9% Speed(reviews/sec):1582. #Correct:232 #Tested:260 Testing Accuracy:89.2%\r",
      "Progress:26.0% Speed(reviews/sec):1585. #Correct:233 #Tested:261 Testing Accuracy:89.2%\r",
      "Progress:26.1% Speed(reviews/sec):1590. #Correct:234 #Tested:262 Testing Accuracy:89.3%\r",
      "Progress:26.2% Speed(reviews/sec):1588. #Correct:235 #Tested:263 Testing Accuracy:89.3%\r",
      "Progress:26.3% Speed(reviews/sec):1586. #Correct:236 #Tested:264 Testing Accuracy:89.3%\r",
      "Progress:26.4% Speed(reviews/sec):1587. #Correct:237 #Tested:265 Testing Accuracy:89.4%\r",
      "Progress:26.5% Speed(reviews/sec):1579. #Correct:238 #Tested:266 Testing Accuracy:89.4%\r",
      "Progress:26.6% Speed(reviews/sec):1580. #Correct:239 #Tested:267 Testing Accuracy:89.5%\r",
      "Progress:26.7% Speed(reviews/sec):1557. #Correct:239 #Tested:268 Testing Accuracy:89.1%\r",
      "Progress:26.8% Speed(reviews/sec):1559. #Correct:240 #Tested:269 Testing Accuracy:89.2%\r",
      "Progress:26.9% Speed(reviews/sec):1562. #Correct:240 #Tested:270 Testing Accuracy:88.8%\r",
      "Progress:27.0% Speed(reviews/sec):1562. #Correct:240 #Tested:271 Testing Accuracy:88.5%\r",
      "Progress:27.1% Speed(reviews/sec):1564. #Correct:241 #Tested:272 Testing Accuracy:88.6%\r",
      "Progress:27.2% Speed(reviews/sec):1563. #Correct:242 #Tested:273 Testing Accuracy:88.6%\r",
      "Progress:27.3% Speed(reviews/sec):1566. #Correct:243 #Tested:274 Testing Accuracy:88.6%\r",
      "Progress:27.4% Speed(reviews/sec):1566. #Correct:244 #Tested:275 Testing Accuracy:88.7%\r",
      "Progress:27.5% Speed(reviews/sec):1563. #Correct:245 #Tested:276 Testing Accuracy:88.7%\r",
      "Progress:27.6% Speed(reviews/sec):1566. #Correct:246 #Tested:277 Testing Accuracy:88.8%\r",
      "Progress:27.7% Speed(reviews/sec):1568. #Correct:247 #Tested:278 Testing Accuracy:88.8%\r",
      "Progress:27.8% Speed(reviews/sec):1571. #Correct:248 #Tested:279 Testing Accuracy:88.8%\r",
      "Progress:27.9% Speed(reviews/sec):1574. #Correct:249 #Tested:280 Testing Accuracy:88.9%\r",
      "Progress:28.0% Speed(reviews/sec):1577. #Correct:250 #Tested:281 Testing Accuracy:88.9%\r",
      "Progress:28.1% Speed(reviews/sec):1579. #Correct:251 #Tested:282 Testing Accuracy:89.0%\r",
      "Progress:28.2% Speed(reviews/sec):1583. #Correct:252 #Tested:283 Testing Accuracy:89.0%\r",
      "Progress:28.3% Speed(reviews/sec):1586. #Correct:253 #Tested:284 Testing Accuracy:89.0%\r",
      "Progress:28.4% Speed(reviews/sec):1580. #Correct:254 #Tested:285 Testing Accuracy:89.1%\r",
      "Progress:28.5% Speed(reviews/sec):1582. #Correct:255 #Tested:286 Testing Accuracy:89.1%\r",
      "Progress:28.6% Speed(reviews/sec):1585. #Correct:256 #Tested:287 Testing Accuracy:89.1%\r",
      "Progress:28.7% Speed(reviews/sec):1589. #Correct:257 #Tested:288 Testing Accuracy:89.2%\r",
      "Progress:28.8% Speed(reviews/sec):1592. #Correct:258 #Tested:289 Testing Accuracy:89.2%\r",
      "Progress:28.9% Speed(reviews/sec):1595. #Correct:259 #Tested:290 Testing Accuracy:89.3%\r",
      "Progress:29.0% Speed(reviews/sec):1597. #Correct:260 #Tested:291 Testing Accuracy:89.3%\r",
      "Progress:29.1% Speed(reviews/sec):1599. #Correct:261 #Tested:292 Testing Accuracy:89.3%\r",
      "Progress:29.2% Speed(reviews/sec):1602. #Correct:262 #Tested:293 Testing Accuracy:89.4%\r",
      "Progress:29.3% Speed(reviews/sec):1588. #Correct:263 #Tested:294 Testing Accuracy:89.4%\r",
      "Progress:29.4% Speed(reviews/sec):1590. #Correct:264 #Tested:295 Testing Accuracy:89.4%\r",
      "Progress:29.5% Speed(reviews/sec):1594. #Correct:265 #Tested:296 Testing Accuracy:89.5%\r",
      "Progress:29.6% Speed(reviews/sec):1598. #Correct:266 #Tested:297 Testing Accuracy:89.5%\r",
      "Progress:29.7% Speed(reviews/sec):1597. #Correct:266 #Tested:298 Testing Accuracy:89.2%\r",
      "Progress:29.8% Speed(reviews/sec):1598. #Correct:267 #Tested:299 Testing Accuracy:89.2%\r",
      "Progress:29.9% Speed(reviews/sec):1595. #Correct:268 #Tested:300 Testing Accuracy:89.3%\r",
      "Progress:30.0% Speed(reviews/sec):1596. #Correct:269 #Tested:301 Testing Accuracy:89.3%\r",
      "Progress:30.1% Speed(reviews/sec):1595. #Correct:269 #Tested:302 Testing Accuracy:89.0%\r",
      "Progress:30.2% Speed(reviews/sec):1590. #Correct:270 #Tested:303 Testing Accuracy:89.1%\r",
      "Progress:30.3% Speed(reviews/sec):1591. #Correct:271 #Tested:304 Testing Accuracy:89.1%\r",
      "Progress:30.4% Speed(reviews/sec):1594. #Correct:272 #Tested:305 Testing Accuracy:89.1%\r",
      "Progress:30.5% Speed(reviews/sec):1597. #Correct:272 #Tested:306 Testing Accuracy:88.8%\r",
      "Progress:30.6% Speed(reviews/sec):1591. #Correct:273 #Tested:307 Testing Accuracy:88.9%\r",
      "Progress:30.7% Speed(reviews/sec):1594. #Correct:273 #Tested:308 Testing Accuracy:88.6%\r",
      "Progress:30.8% Speed(reviews/sec):1595. #Correct:274 #Tested:309 Testing Accuracy:88.6%\r",
      "Progress:30.9% Speed(reviews/sec):1593. #Correct:274 #Tested:310 Testing Accuracy:88.3%\r",
      "Progress:31.0% Speed(reviews/sec):1597. #Correct:275 #Tested:311 Testing Accuracy:88.4%\r",
      "Progress:31.1% Speed(reviews/sec):1598. #Correct:276 #Tested:312 Testing Accuracy:88.4%\r",
      "Progress:31.2% Speed(reviews/sec):1601. #Correct:277 #Tested:313 Testing Accuracy:88.4%\r",
      "Progress:31.3% Speed(reviews/sec):1602. #Correct:278 #Tested:314 Testing Accuracy:88.5%\r",
      "Progress:31.4% Speed(reviews/sec):1605. #Correct:279 #Tested:315 Testing Accuracy:88.5%\r",
      "Progress:31.5% Speed(reviews/sec):1606. #Correct:280 #Tested:316 Testing Accuracy:88.6%\r",
      "Progress:31.6% Speed(reviews/sec):1608. #Correct:281 #Tested:317 Testing Accuracy:88.6%\r",
      "Progress:31.7% Speed(reviews/sec):1609. #Correct:281 #Tested:318 Testing Accuracy:88.3%\r",
      "Progress:31.8% Speed(reviews/sec):1612. #Correct:282 #Tested:319 Testing Accuracy:88.4%\r",
      "Progress:31.9% Speed(reviews/sec):1614. #Correct:283 #Tested:320 Testing Accuracy:88.4%\r",
      "Progress:32.0% Speed(reviews/sec):1617. #Correct:284 #Tested:321 Testing Accuracy:88.4%\r",
      "Progress:32.1% Speed(reviews/sec):1621. #Correct:284 #Tested:322 Testing Accuracy:88.1%\r",
      "Progress:32.2% Speed(reviews/sec):1624. #Correct:285 #Tested:323 Testing Accuracy:88.2%\r",
      "Progress:32.3% Speed(reviews/sec):1625. #Correct:286 #Tested:324 Testing Accuracy:88.2%\r",
      "Progress:32.4% Speed(reviews/sec):1624. #Correct:287 #Tested:325 Testing Accuracy:88.3%\r",
      "Progress:32.5% Speed(reviews/sec):1623. #Correct:288 #Tested:326 Testing Accuracy:88.3%\r",
      "Progress:32.6% Speed(reviews/sec):1625. #Correct:289 #Tested:327 Testing Accuracy:88.3%\r",
      "Progress:32.7% Speed(reviews/sec):1625. #Correct:290 #Tested:328 Testing Accuracy:88.4%\r",
      "Progress:32.8% Speed(reviews/sec):1618. #Correct:291 #Tested:329 Testing Accuracy:88.4%\r",
      "Progress:32.9% Speed(reviews/sec):1620. #Correct:292 #Tested:330 Testing Accuracy:88.4%\r",
      "Progress:33.0% Speed(reviews/sec):1622. #Correct:293 #Tested:331 Testing Accuracy:88.5%\r",
      "Progress:33.1% Speed(reviews/sec):1624. #Correct:294 #Tested:332 Testing Accuracy:88.5%\r",
      "Progress:33.2% Speed(reviews/sec):1623. #Correct:295 #Tested:333 Testing Accuracy:88.5%\r",
      "Progress:33.3% Speed(reviews/sec):1625. #Correct:296 #Tested:334 Testing Accuracy:88.6%\r",
      "Progress:33.4% Speed(reviews/sec):1622. #Correct:297 #Tested:335 Testing Accuracy:88.6%\r",
      "Progress:33.5% Speed(reviews/sec):1624. #Correct:298 #Tested:336 Testing Accuracy:88.6%\r",
      "Progress:33.6% Speed(reviews/sec):1626. #Correct:299 #Tested:337 Testing Accuracy:88.7%\r",
      "Progress:33.7% Speed(reviews/sec):1629. #Correct:300 #Tested:338 Testing Accuracy:88.7%\r",
      "Progress:33.8% Speed(reviews/sec):1630. #Correct:301 #Tested:339 Testing Accuracy:88.7%\r",
      "Progress:33.9% Speed(reviews/sec):1631. #Correct:302 #Tested:340 Testing Accuracy:88.8%\r",
      "Progress:34.0% Speed(reviews/sec):1630. #Correct:303 #Tested:341 Testing Accuracy:88.8%\r",
      "Progress:34.1% Speed(reviews/sec):1632. #Correct:304 #Tested:342 Testing Accuracy:88.8%\r",
      "Progress:34.2% Speed(reviews/sec):1609. #Correct:305 #Tested:343 Testing Accuracy:88.9%\r",
      "Progress:34.3% Speed(reviews/sec):1611. #Correct:306 #Tested:344 Testing Accuracy:88.9%\r",
      "Progress:34.4% Speed(reviews/sec):1610. #Correct:307 #Tested:345 Testing Accuracy:88.9%\r",
      "Progress:34.5% Speed(reviews/sec):1613. #Correct:308 #Tested:346 Testing Accuracy:89.0%\r",
      "Progress:34.6% Speed(reviews/sec):1615. #Correct:309 #Tested:347 Testing Accuracy:89.0%\r",
      "Progress:34.7% Speed(reviews/sec):1617. #Correct:310 #Tested:348 Testing Accuracy:89.0%\r",
      "Progress:34.8% Speed(reviews/sec):1614. #Correct:311 #Tested:349 Testing Accuracy:89.1%\r",
      "Progress:34.9% Speed(reviews/sec):1615. #Correct:312 #Tested:350 Testing Accuracy:89.1%\r",
      "Progress:35.0% Speed(reviews/sec):1617. #Correct:313 #Tested:351 Testing Accuracy:89.1%\r",
      "Progress:35.1% Speed(reviews/sec):1617. #Correct:314 #Tested:352 Testing Accuracy:89.2%\r",
      "Progress:35.2% Speed(reviews/sec):1618. #Correct:315 #Tested:353 Testing Accuracy:89.2%\r",
      "Progress:35.3% Speed(reviews/sec):1620. #Correct:315 #Tested:354 Testing Accuracy:88.9%\r",
      "Progress:35.4% Speed(reviews/sec):1622. #Correct:316 #Tested:355 Testing Accuracy:89.0%\r",
      "Progress:35.5% Speed(reviews/sec):1620. #Correct:317 #Tested:356 Testing Accuracy:89.0%\r",
      "Progress:35.6% Speed(reviews/sec):1621. #Correct:318 #Tested:357 Testing Accuracy:89.0%\r",
      "Progress:35.7% Speed(reviews/sec):1621. #Correct:318 #Tested:358 Testing Accuracy:88.8%\r",
      "Progress:35.8% Speed(reviews/sec):1622. #Correct:319 #Tested:359 Testing Accuracy:88.8%\r",
      "Progress:35.9% Speed(reviews/sec):1624. #Correct:320 #Tested:360 Testing Accuracy:88.8%\r",
      "Progress:36.0% Speed(reviews/sec):1618. #Correct:321 #Tested:361 Testing Accuracy:88.9%\r",
      "Progress:36.1% Speed(reviews/sec):1618. #Correct:322 #Tested:362 Testing Accuracy:88.9%\r",
      "Progress:36.2% Speed(reviews/sec):1613. #Correct:323 #Tested:363 Testing Accuracy:88.9%\r",
      "Progress:36.3% Speed(reviews/sec):1614. #Correct:324 #Tested:364 Testing Accuracy:89.0%\r",
      "Progress:36.4% Speed(reviews/sec):1616. #Correct:325 #Tested:365 Testing Accuracy:89.0%\r",
      "Progress:36.5% Speed(reviews/sec):1617. #Correct:326 #Tested:366 Testing Accuracy:89.0%\r",
      "Progress:36.6% Speed(reviews/sec):1618. #Correct:327 #Tested:367 Testing Accuracy:89.1%\r",
      "Progress:36.7% Speed(reviews/sec):1621. #Correct:328 #Tested:368 Testing Accuracy:89.1%\r",
      "Progress:36.8% Speed(reviews/sec):1621. #Correct:329 #Tested:369 Testing Accuracy:89.1%\r",
      "Progress:36.9% Speed(reviews/sec):1623. #Correct:330 #Tested:370 Testing Accuracy:89.1%\r",
      "Progress:37.0% Speed(reviews/sec):1624. #Correct:331 #Tested:371 Testing Accuracy:89.2%\r",
      "Progress:37.1% Speed(reviews/sec):1620. #Correct:331 #Tested:372 Testing Accuracy:88.9%\r",
      "Progress:37.2% Speed(reviews/sec):1622. #Correct:331 #Tested:373 Testing Accuracy:88.7%\r",
      "Progress:37.3% Speed(reviews/sec):1624. #Correct:332 #Tested:374 Testing Accuracy:88.7%\r",
      "Progress:37.4% Speed(reviews/sec):1623. #Correct:333 #Tested:375 Testing Accuracy:88.8%\r",
      "Progress:37.5% Speed(reviews/sec):1615. #Correct:333 #Tested:376 Testing Accuracy:88.5%\r",
      "Progress:37.6% Speed(reviews/sec):1617. #Correct:334 #Tested:377 Testing Accuracy:88.5%\r",
      "Progress:37.7% Speed(reviews/sec):1618. #Correct:335 #Tested:378 Testing Accuracy:88.6%\r",
      "Progress:37.8% Speed(reviews/sec):1620. #Correct:336 #Tested:379 Testing Accuracy:88.6%\r",
      "Progress:37.9% Speed(reviews/sec):1621. #Correct:337 #Tested:380 Testing Accuracy:88.6%\r",
      "Progress:38.0% Speed(reviews/sec):1617. #Correct:338 #Tested:381 Testing Accuracy:88.7%\r",
      "Progress:38.1% Speed(reviews/sec):1619. #Correct:339 #Tested:382 Testing Accuracy:88.7%\r",
      "Progress:38.2% Speed(reviews/sec):1618. #Correct:340 #Tested:383 Testing Accuracy:88.7%\r",
      "Progress:38.3% Speed(reviews/sec):1619. #Correct:341 #Tested:384 Testing Accuracy:88.8%\r",
      "Progress:38.4% Speed(reviews/sec):1621. #Correct:341 #Tested:385 Testing Accuracy:88.5%\r",
      "Progress:38.5% Speed(reviews/sec):1616. #Correct:342 #Tested:386 Testing Accuracy:88.6%\r",
      "Progress:38.6% Speed(reviews/sec):1616. #Correct:343 #Tested:387 Testing Accuracy:88.6%\r",
      "Progress:38.7% Speed(reviews/sec):1619. #Correct:343 #Tested:388 Testing Accuracy:88.4%\r",
      "Progress:38.8% Speed(reviews/sec):1619. #Correct:343 #Tested:389 Testing Accuracy:88.1%\r",
      "Progress:38.9% Speed(reviews/sec):1620. #Correct:344 #Tested:390 Testing Accuracy:88.2%\r",
      "Progress:39.0% Speed(reviews/sec):1622. #Correct:345 #Tested:391 Testing Accuracy:88.2%\r",
      "Progress:39.1% Speed(reviews/sec):1624. #Correct:346 #Tested:392 Testing Accuracy:88.2%\r",
      "Progress:39.2% Speed(reviews/sec):1625. #Correct:347 #Tested:393 Testing Accuracy:88.2%\r",
      "Progress:39.3% Speed(reviews/sec):1622. #Correct:348 #Tested:394 Testing Accuracy:88.3%\r",
      "Progress:39.4% Speed(reviews/sec):1622. #Correct:349 #Tested:395 Testing Accuracy:88.3%\r",
      "Progress:39.5% Speed(reviews/sec):1624. #Correct:350 #Tested:396 Testing Accuracy:88.3%\r",
      "Progress:39.6% Speed(reviews/sec):1625. #Correct:351 #Tested:397 Testing Accuracy:88.4%\r",
      "Progress:39.7% Speed(reviews/sec):1626. #Correct:352 #Tested:398 Testing Accuracy:88.4%\r",
      "Progress:39.8% Speed(reviews/sec):1625. #Correct:353 #Tested:399 Testing Accuracy:88.4%\r",
      "Progress:39.9% Speed(reviews/sec):1627. #Correct:354 #Tested:400 Testing Accuracy:88.5%\r",
      "Progress:40.0% Speed(reviews/sec):1620. #Correct:355 #Tested:401 Testing Accuracy:88.5%\r",
      "Progress:40.1% Speed(reviews/sec):1622. #Correct:356 #Tested:402 Testing Accuracy:88.5%\r",
      "Progress:40.2% Speed(reviews/sec):1621. #Correct:357 #Tested:403 Testing Accuracy:88.5%\r",
      "Progress:40.3% Speed(reviews/sec):1622. #Correct:357 #Tested:404 Testing Accuracy:88.3%\r",
      "Progress:40.4% Speed(reviews/sec):1624. #Correct:358 #Tested:405 Testing Accuracy:88.3%\r",
      "Progress:40.5% Speed(reviews/sec):1625. #Correct:359 #Tested:406 Testing Accuracy:88.4%\r",
      "Progress:40.6% Speed(reviews/sec):1623. #Correct:360 #Tested:407 Testing Accuracy:88.4%\r",
      "Progress:40.7% Speed(reviews/sec):1624. #Correct:361 #Tested:408 Testing Accuracy:88.4%\r",
      "Progress:40.8% Speed(reviews/sec):1624. #Correct:362 #Tested:409 Testing Accuracy:88.5%\r",
      "Progress:40.9% Speed(reviews/sec):1622. #Correct:363 #Tested:410 Testing Accuracy:88.5%\r",
      "Progress:41.0% Speed(reviews/sec):1623. #Correct:364 #Tested:411 Testing Accuracy:88.5%\r",
      "Progress:41.1% Speed(reviews/sec):1622. #Correct:365 #Tested:412 Testing Accuracy:88.5%\r",
      "Progress:41.2% Speed(reviews/sec):1624. #Correct:366 #Tested:413 Testing Accuracy:88.6%\r",
      "Progress:41.3% Speed(reviews/sec):1624. #Correct:367 #Tested:414 Testing Accuracy:88.6%\r",
      "Progress:41.4% Speed(reviews/sec):1625. #Correct:368 #Tested:415 Testing Accuracy:88.6%\r",
      "Progress:41.5% Speed(reviews/sec):1625. #Correct:369 #Tested:416 Testing Accuracy:88.7%"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\r",
      "Progress:41.6% Speed(reviews/sec):1624. #Correct:370 #Tested:417 Testing Accuracy:88.7%\r",
      "Progress:41.7% Speed(reviews/sec):1624. #Correct:371 #Tested:418 Testing Accuracy:88.7%\r",
      "Progress:41.8% Speed(reviews/sec):1624. #Correct:372 #Tested:419 Testing Accuracy:88.7%\r",
      "Progress:41.9% Speed(reviews/sec):1618. #Correct:373 #Tested:420 Testing Accuracy:88.8%\r",
      "Progress:42.0% Speed(reviews/sec):1616. #Correct:374 #Tested:421 Testing Accuracy:88.8%\r",
      "Progress:42.1% Speed(reviews/sec):1614. #Correct:375 #Tested:422 Testing Accuracy:88.8%\r",
      "Progress:42.2% Speed(reviews/sec):1615. #Correct:376 #Tested:423 Testing Accuracy:88.8%\r",
      "Progress:42.3% Speed(reviews/sec):1615. #Correct:377 #Tested:424 Testing Accuracy:88.9%\r",
      "Progress:42.4% Speed(reviews/sec):1613. #Correct:378 #Tested:425 Testing Accuracy:88.9%\r",
      "Progress:42.5% Speed(reviews/sec):1611. #Correct:379 #Tested:426 Testing Accuracy:88.9%\r",
      "Progress:42.6% Speed(reviews/sec):1592. #Correct:380 #Tested:427 Testing Accuracy:88.9%\r",
      "Progress:42.7% Speed(reviews/sec):1588. #Correct:381 #Tested:428 Testing Accuracy:89.0%\r",
      "Progress:42.8% Speed(reviews/sec):1585. #Correct:382 #Tested:429 Testing Accuracy:89.0%\r",
      "Progress:42.9% Speed(reviews/sec):1581. #Correct:383 #Tested:430 Testing Accuracy:89.0%\r",
      "Progress:43.0% Speed(reviews/sec):1574. #Correct:384 #Tested:431 Testing Accuracy:89.0%\r",
      "Progress:43.1% Speed(reviews/sec):1576. #Correct:385 #Tested:432 Testing Accuracy:89.1%\r",
      "Progress:43.2% Speed(reviews/sec):1573. #Correct:386 #Tested:433 Testing Accuracy:89.1%\r",
      "Progress:43.3% Speed(reviews/sec):1575. #Correct:387 #Tested:434 Testing Accuracy:89.1%\r",
      "Progress:43.4% Speed(reviews/sec):1576. #Correct:388 #Tested:435 Testing Accuracy:89.1%\r",
      "Progress:43.5% Speed(reviews/sec):1574. #Correct:389 #Tested:436 Testing Accuracy:89.2%\r",
      "Progress:43.6% Speed(reviews/sec):1575. #Correct:390 #Tested:437 Testing Accuracy:89.2%\r",
      "Progress:43.7% Speed(reviews/sec):1573. #Correct:391 #Tested:438 Testing Accuracy:89.2%\r",
      "Progress:43.8% Speed(reviews/sec):1573. #Correct:392 #Tested:439 Testing Accuracy:89.2%\r",
      "Progress:43.9% Speed(reviews/sec):1572. #Correct:393 #Tested:440 Testing Accuracy:89.3%\r",
      "Progress:44.0% Speed(reviews/sec):1571. #Correct:394 #Tested:441 Testing Accuracy:89.3%\r",
      "Progress:44.1% Speed(reviews/sec):1569. #Correct:395 #Tested:442 Testing Accuracy:89.3%\r",
      "Progress:44.2% Speed(reviews/sec):1572. #Correct:396 #Tested:443 Testing Accuracy:89.3%\r",
      "Progress:44.3% Speed(reviews/sec):1571. #Correct:397 #Tested:444 Testing Accuracy:89.4%\r",
      "Progress:44.4% Speed(reviews/sec):1567. #Correct:398 #Tested:445 Testing Accuracy:89.4%\r",
      "Progress:44.5% Speed(reviews/sec):1564. #Correct:399 #Tested:446 Testing Accuracy:89.4%\r",
      "Progress:44.6% Speed(reviews/sec):1565. #Correct:400 #Tested:447 Testing Accuracy:89.4%\r",
      "Progress:44.7% Speed(reviews/sec):1564. #Correct:401 #Tested:448 Testing Accuracy:89.5%\r",
      "Progress:44.8% Speed(reviews/sec):1561. #Correct:402 #Tested:449 Testing Accuracy:89.5%\r",
      "Progress:44.9% Speed(reviews/sec):1560. #Correct:402 #Tested:450 Testing Accuracy:89.3%\r",
      "Progress:45.0% Speed(reviews/sec):1562. #Correct:403 #Tested:451 Testing Accuracy:89.3%\r",
      "Progress:45.1% Speed(reviews/sec):1559. #Correct:404 #Tested:452 Testing Accuracy:89.3%\r",
      "Progress:45.2% Speed(reviews/sec):1560. #Correct:405 #Tested:453 Testing Accuracy:89.4%\r",
      "Progress:45.3% Speed(reviews/sec):1562. #Correct:406 #Tested:454 Testing Accuracy:89.4%\r",
      "Progress:45.4% Speed(reviews/sec):1564. #Correct:407 #Tested:455 Testing Accuracy:89.4%\r",
      "Progress:45.5% Speed(reviews/sec):1565. #Correct:408 #Tested:456 Testing Accuracy:89.4%\r",
      "Progress:45.6% Speed(reviews/sec):1567. #Correct:409 #Tested:457 Testing Accuracy:89.4%\r",
      "Progress:45.7% Speed(reviews/sec):1569. #Correct:410 #Tested:458 Testing Accuracy:89.5%\r",
      "Progress:45.8% Speed(reviews/sec):1570. #Correct:411 #Tested:459 Testing Accuracy:89.5%\r",
      "Progress:45.9% Speed(reviews/sec):1568. #Correct:411 #Tested:460 Testing Accuracy:89.3%\r",
      "Progress:46.0% Speed(reviews/sec):1563. #Correct:412 #Tested:461 Testing Accuracy:89.3%\r",
      "Progress:46.1% Speed(reviews/sec):1561. #Correct:413 #Tested:462 Testing Accuracy:89.3%\r",
      "Progress:46.2% Speed(reviews/sec):1554. #Correct:414 #Tested:463 Testing Accuracy:89.4%\r",
      "Progress:46.3% Speed(reviews/sec):1555. #Correct:414 #Tested:464 Testing Accuracy:89.2%\r",
      "Progress:46.4% Speed(reviews/sec):1557. #Correct:415 #Tested:465 Testing Accuracy:89.2%\r",
      "Progress:46.5% Speed(reviews/sec):1558. #Correct:416 #Tested:466 Testing Accuracy:89.2%\r",
      "Progress:46.6% Speed(reviews/sec):1559. #Correct:417 #Tested:467 Testing Accuracy:89.2%\r",
      "Progress:46.7% Speed(reviews/sec):1558. #Correct:418 #Tested:468 Testing Accuracy:89.3%\r",
      "Progress:46.8% Speed(reviews/sec):1555. #Correct:419 #Tested:469 Testing Accuracy:89.3%\r",
      "Progress:46.9% Speed(reviews/sec):1557. #Correct:419 #Tested:470 Testing Accuracy:89.1%\r",
      "Progress:47.0% Speed(reviews/sec):1558. #Correct:420 #Tested:471 Testing Accuracy:89.1%\r",
      "Progress:47.1% Speed(reviews/sec):1557. #Correct:420 #Tested:472 Testing Accuracy:88.9%\r",
      "Progress:47.2% Speed(reviews/sec):1559. #Correct:421 #Tested:473 Testing Accuracy:89.0%\r",
      "Progress:47.3% Speed(reviews/sec):1558. #Correct:422 #Tested:474 Testing Accuracy:89.0%\r",
      "Progress:47.4% Speed(reviews/sec):1558. #Correct:423 #Tested:475 Testing Accuracy:89.0%\r",
      "Progress:47.5% Speed(reviews/sec):1557. #Correct:424 #Tested:476 Testing Accuracy:89.0%\r",
      "Progress:47.6% Speed(reviews/sec):1555. #Correct:425 #Tested:477 Testing Accuracy:89.0%\r",
      "Progress:47.7% Speed(reviews/sec):1555. #Correct:426 #Tested:478 Testing Accuracy:89.1%\r",
      "Progress:47.8% Speed(reviews/sec):1553. #Correct:427 #Tested:479 Testing Accuracy:89.1%\r",
      "Progress:47.9% Speed(reviews/sec):1554. #Correct:428 #Tested:480 Testing Accuracy:89.1%\r",
      "Progress:48.0% Speed(reviews/sec):1555. #Correct:429 #Tested:481 Testing Accuracy:89.1%\r",
      "Progress:48.1% Speed(reviews/sec):1551. #Correct:430 #Tested:482 Testing Accuracy:89.2%\r",
      "Progress:48.2% Speed(reviews/sec):1552. #Correct:430 #Tested:483 Testing Accuracy:89.0%\r",
      "Progress:48.3% Speed(reviews/sec):1554. #Correct:431 #Tested:484 Testing Accuracy:89.0%\r",
      "Progress:48.4% Speed(reviews/sec):1554. #Correct:432 #Tested:485 Testing Accuracy:89.0%\r",
      "Progress:48.5% Speed(reviews/sec):1554. #Correct:433 #Tested:486 Testing Accuracy:89.0%\r",
      "Progress:48.6% Speed(reviews/sec):1556. #Correct:434 #Tested:487 Testing Accuracy:89.1%\r",
      "Progress:48.7% Speed(reviews/sec):1558. #Correct:434 #Tested:488 Testing Accuracy:88.9%\r",
      "Progress:48.8% Speed(reviews/sec):1559. #Correct:435 #Tested:489 Testing Accuracy:88.9%\r",
      "Progress:48.9% Speed(reviews/sec):1557. #Correct:435 #Tested:490 Testing Accuracy:88.7%\r",
      "Progress:49.0% Speed(reviews/sec):1558. #Correct:436 #Tested:491 Testing Accuracy:88.7%\r",
      "Progress:49.1% Speed(reviews/sec):1560. #Correct:437 #Tested:492 Testing Accuracy:88.8%\r",
      "Progress:49.2% Speed(reviews/sec):1560. #Correct:438 #Tested:493 Testing Accuracy:88.8%\r",
      "Progress:49.3% Speed(reviews/sec):1559. #Correct:439 #Tested:494 Testing Accuracy:88.8%\r",
      "Progress:49.4% Speed(reviews/sec):1560. #Correct:440 #Tested:495 Testing Accuracy:88.8%\r",
      "Progress:49.5% Speed(reviews/sec):1553. #Correct:441 #Tested:496 Testing Accuracy:88.9%\r",
      "Progress:49.6% Speed(reviews/sec):1554. #Correct:442 #Tested:497 Testing Accuracy:88.9%\r",
      "Progress:49.7% Speed(reviews/sec):1556. #Correct:442 #Tested:498 Testing Accuracy:88.7%\r",
      "Progress:49.8% Speed(reviews/sec):1555. #Correct:443 #Tested:499 Testing Accuracy:88.7%\r",
      "Progress:49.9% Speed(reviews/sec):1556. #Correct:444 #Tested:500 Testing Accuracy:88.8%\r",
      "Progress:50.0% Speed(reviews/sec):1557. #Correct:445 #Tested:501 Testing Accuracy:88.8%\r",
      "Progress:50.1% Speed(reviews/sec):1556. #Correct:446 #Tested:502 Testing Accuracy:88.8%\r",
      "Progress:50.2% Speed(reviews/sec):1556. #Correct:447 #Tested:503 Testing Accuracy:88.8%\r",
      "Progress:50.3% Speed(reviews/sec):1557. #Correct:448 #Tested:504 Testing Accuracy:88.8%\r",
      "Progress:50.4% Speed(reviews/sec):1558. #Correct:449 #Tested:505 Testing Accuracy:88.9%\r",
      "Progress:50.5% Speed(reviews/sec):1559. #Correct:450 #Tested:506 Testing Accuracy:88.9%\r",
      "Progress:50.6% Speed(reviews/sec):1556. #Correct:451 #Tested:507 Testing Accuracy:88.9%\r",
      "Progress:50.7% Speed(reviews/sec):1553. #Correct:452 #Tested:508 Testing Accuracy:88.9%\r",
      "Progress:50.8% Speed(reviews/sec):1554. #Correct:453 #Tested:509 Testing Accuracy:88.9%\r",
      "Progress:50.9% Speed(reviews/sec):1555. #Correct:454 #Tested:510 Testing Accuracy:89.0%\r",
      "Progress:51.0% Speed(reviews/sec):1549. #Correct:455 #Tested:511 Testing Accuracy:89.0%\r",
      "Progress:51.1% Speed(reviews/sec):1548. #Correct:456 #Tested:512 Testing Accuracy:89.0%\r",
      "Progress:51.2% Speed(reviews/sec):1548. #Correct:456 #Tested:513 Testing Accuracy:88.8%\r",
      "Progress:51.3% Speed(reviews/sec):1550. #Correct:457 #Tested:514 Testing Accuracy:88.9%\r",
      "Progress:51.4% Speed(reviews/sec):1550. #Correct:458 #Tested:515 Testing Accuracy:88.9%\r",
      "Progress:51.5% Speed(reviews/sec):1548. #Correct:459 #Tested:516 Testing Accuracy:88.9%\r",
      "Progress:51.6% Speed(reviews/sec):1546. #Correct:460 #Tested:517 Testing Accuracy:88.9%\r",
      "Progress:51.7% Speed(reviews/sec):1547. #Correct:461 #Tested:518 Testing Accuracy:88.9%\r",
      "Progress:51.8% Speed(reviews/sec):1548. #Correct:461 #Tested:519 Testing Accuracy:88.8%\r",
      "Progress:51.9% Speed(reviews/sec):1548. #Correct:462 #Tested:520 Testing Accuracy:88.8%\r",
      "Progress:52.0% Speed(reviews/sec):1550. #Correct:463 #Tested:521 Testing Accuracy:88.8%\r",
      "Progress:52.1% Speed(reviews/sec):1550. #Correct:464 #Tested:522 Testing Accuracy:88.8%\r",
      "Progress:52.2% Speed(reviews/sec):1547. #Correct:465 #Tested:523 Testing Accuracy:88.9%\r",
      "Progress:52.3% Speed(reviews/sec):1548. #Correct:466 #Tested:524 Testing Accuracy:88.9%\r",
      "Progress:52.4% Speed(reviews/sec):1549. #Correct:467 #Tested:525 Testing Accuracy:88.9%\r",
      "Progress:52.5% Speed(reviews/sec):1549. #Correct:468 #Tested:526 Testing Accuracy:88.9%\r",
      "Progress:52.6% Speed(reviews/sec):1551. #Correct:469 #Tested:527 Testing Accuracy:88.9%\r",
      "Progress:52.7% Speed(reviews/sec):1552. #Correct:470 #Tested:528 Testing Accuracy:89.0%\r",
      "Progress:52.8% Speed(reviews/sec):1552. #Correct:471 #Tested:529 Testing Accuracy:89.0%\r",
      "Progress:52.9% Speed(reviews/sec):1552. #Correct:472 #Tested:530 Testing Accuracy:89.0%\r",
      "Progress:53.0% Speed(reviews/sec):1552. #Correct:473 #Tested:531 Testing Accuracy:89.0%\r",
      "Progress:53.1% Speed(reviews/sec):1553. #Correct:473 #Tested:532 Testing Accuracy:88.9%\r",
      "Progress:53.2% Speed(reviews/sec):1554. #Correct:474 #Tested:533 Testing Accuracy:88.9%\r",
      "Progress:53.3% Speed(reviews/sec):1555. #Correct:475 #Tested:534 Testing Accuracy:88.9%\r",
      "Progress:53.4% Speed(reviews/sec):1557. #Correct:476 #Tested:535 Testing Accuracy:88.9%\r",
      "Progress:53.5% Speed(reviews/sec):1558. #Correct:477 #Tested:536 Testing Accuracy:88.9%\r",
      "Progress:53.6% Speed(reviews/sec):1560. #Correct:478 #Tested:537 Testing Accuracy:89.0%\r",
      "Progress:53.7% Speed(reviews/sec):1561. #Correct:479 #Tested:538 Testing Accuracy:89.0%\r",
      "Progress:53.8% Speed(reviews/sec):1563. #Correct:480 #Tested:539 Testing Accuracy:89.0%\r",
      "Progress:53.9% Speed(reviews/sec):1564. #Correct:481 #Tested:540 Testing Accuracy:89.0%\r",
      "Progress:54.0% Speed(reviews/sec):1565. #Correct:482 #Tested:541 Testing Accuracy:89.0%\r",
      "Progress:54.1% Speed(reviews/sec):1566. #Correct:483 #Tested:542 Testing Accuracy:89.1%\r",
      "Progress:54.2% Speed(reviews/sec):1566. #Correct:484 #Tested:543 Testing Accuracy:89.1%\r",
      "Progress:54.3% Speed(reviews/sec):1567. #Correct:485 #Tested:544 Testing Accuracy:89.1%\r",
      "Progress:54.4% Speed(reviews/sec):1567. #Correct:485 #Tested:545 Testing Accuracy:88.9%\r",
      "Progress:54.5% Speed(reviews/sec):1568. #Correct:486 #Tested:546 Testing Accuracy:89.0%\r",
      "Progress:54.6% Speed(reviews/sec):1566. #Correct:487 #Tested:547 Testing Accuracy:89.0%\r",
      "Progress:54.7% Speed(reviews/sec):1566. #Correct:487 #Tested:548 Testing Accuracy:88.8%\r",
      "Progress:54.8% Speed(reviews/sec):1568. #Correct:488 #Tested:549 Testing Accuracy:88.8%\r",
      "Progress:54.9% Speed(reviews/sec):1569. #Correct:489 #Tested:550 Testing Accuracy:88.9%\r",
      "Progress:55.0% Speed(reviews/sec):1571. #Correct:490 #Tested:551 Testing Accuracy:88.9%\r",
      "Progress:55.1% Speed(reviews/sec):1562. #Correct:490 #Tested:552 Testing Accuracy:88.7%\r",
      "Progress:55.2% Speed(reviews/sec):1563. #Correct:491 #Tested:553 Testing Accuracy:88.7%\r",
      "Progress:55.3% Speed(reviews/sec):1565. #Correct:492 #Tested:554 Testing Accuracy:88.8%\r",
      "Progress:55.4% Speed(reviews/sec):1563. #Correct:493 #Tested:555 Testing Accuracy:88.8%\r",
      "Progress:55.5% Speed(reviews/sec):1564. #Correct:494 #Tested:556 Testing Accuracy:88.8%\r",
      "Progress:55.6% Speed(reviews/sec):1562. #Correct:495 #Tested:557 Testing Accuracy:88.8%\r",
      "Progress:55.7% Speed(reviews/sec):1562. #Correct:496 #Tested:558 Testing Accuracy:88.8%\r",
      "Progress:55.8% Speed(reviews/sec):1561. #Correct:496 #Tested:559 Testing Accuracy:88.7%\r",
      "Progress:55.9% Speed(reviews/sec):1562. #Correct:497 #Tested:560 Testing Accuracy:88.7%\r",
      "Progress:56.0% Speed(reviews/sec):1564. #Correct:498 #Tested:561 Testing Accuracy:88.7%\r",
      "Progress:56.1% Speed(reviews/sec):1565. #Correct:499 #Tested:562 Testing Accuracy:88.7%\r",
      "Progress:56.2% Speed(reviews/sec):1565. #Correct:500 #Tested:563 Testing Accuracy:88.8%\r",
      "Progress:56.3% Speed(reviews/sec):1566. #Correct:501 #Tested:564 Testing Accuracy:88.8%\r",
      "Progress:56.4% Speed(reviews/sec):1565. #Correct:502 #Tested:565 Testing Accuracy:88.8%\r",
      "Progress:56.5% Speed(reviews/sec):1566. #Correct:503 #Tested:566 Testing Accuracy:88.8%\r",
      "Progress:56.6% Speed(reviews/sec):1566. #Correct:504 #Tested:567 Testing Accuracy:88.8%\r",
      "Progress:56.7% Speed(reviews/sec):1563. #Correct:505 #Tested:568 Testing Accuracy:88.9%\r",
      "Progress:56.8% Speed(reviews/sec):1564. #Correct:506 #Tested:569 Testing Accuracy:88.9%\r",
      "Progress:56.9% Speed(reviews/sec):1563. #Correct:507 #Tested:570 Testing Accuracy:88.9%\r",
      "Progress:57.0% Speed(reviews/sec):1558. #Correct:507 #Tested:571 Testing Accuracy:88.7%\r",
      "Progress:57.1% Speed(reviews/sec):1555. #Correct:508 #Tested:572 Testing Accuracy:88.8%\r",
      "Progress:57.2% Speed(reviews/sec):1554. #Correct:509 #Tested:573 Testing Accuracy:88.8%\r",
      "Progress:57.3% Speed(reviews/sec):1555. #Correct:510 #Tested:574 Testing Accuracy:88.8%\r",
      "Progress:57.4% Speed(reviews/sec):1554. #Correct:511 #Tested:575 Testing Accuracy:88.8%\r",
      "Progress:57.5% Speed(reviews/sec):1554. #Correct:511 #Tested:576 Testing Accuracy:88.7%\r",
      "Progress:57.6% Speed(reviews/sec):1554. #Correct:512 #Tested:577 Testing Accuracy:88.7%\r",
      "Progress:57.7% Speed(reviews/sec):1554. #Correct:513 #Tested:578 Testing Accuracy:88.7%\r",
      "Progress:57.8% Speed(reviews/sec):1551. #Correct:514 #Tested:579 Testing Accuracy:88.7%\r",
      "Progress:57.9% Speed(reviews/sec):1550. #Correct:515 #Tested:580 Testing Accuracy:88.7%\r",
      "Progress:58.0% Speed(reviews/sec):1550. #Correct:516 #Tested:581 Testing Accuracy:88.8%\r",
      "Progress:58.1% Speed(reviews/sec):1551. #Correct:517 #Tested:582 Testing Accuracy:88.8%\r",
      "Progress:58.2% Speed(reviews/sec):1551. #Correct:518 #Tested:583 Testing Accuracy:88.8%\r",
      "Progress:58.3% Speed(reviews/sec):1552. #Correct:519 #Tested:584 Testing Accuracy:88.8%\r",
      "Progress:58.4% Speed(reviews/sec):1553. #Correct:520 #Tested:585 Testing Accuracy:88.8%\r",
      "Progress:58.5% Speed(reviews/sec):1554. #Correct:521 #Tested:586 Testing Accuracy:88.9%\r",
      "Progress:58.6% Speed(reviews/sec):1556. #Correct:522 #Tested:587 Testing Accuracy:88.9%\r",
      "Progress:58.7% Speed(reviews/sec):1557. #Correct:523 #Tested:588 Testing Accuracy:88.9%\r",
      "Progress:58.8% Speed(reviews/sec):1558. #Correct:524 #Tested:589 Testing Accuracy:88.9%\r",
      "Progress:58.9% Speed(reviews/sec):1560. #Correct:525 #Tested:590 Testing Accuracy:88.9%\r",
      "Progress:59.0% Speed(reviews/sec):1561. #Correct:525 #Tested:591 Testing Accuracy:88.8%\r",
      "Progress:59.1% Speed(reviews/sec):1562. #Correct:526 #Tested:592 Testing Accuracy:88.8%\r",
      "Progress:59.2% Speed(reviews/sec):1562. #Correct:526 #Tested:593 Testing Accuracy:88.7%\r",
      "Progress:59.3% Speed(reviews/sec):1563. #Correct:527 #Tested:594 Testing Accuracy:88.7%\r",
      "Progress:59.4% Speed(reviews/sec):1564. #Correct:528 #Tested:595 Testing Accuracy:88.7%\r",
      "Progress:59.5% Speed(reviews/sec):1566. #Correct:529 #Tested:596 Testing Accuracy:88.7%\r",
      "Progress:59.6% Speed(reviews/sec):1567. #Correct:530 #Tested:597 Testing Accuracy:88.7%\r",
      "Progress:59.7% Speed(reviews/sec):1568. #Correct:531 #Tested:598 Testing Accuracy:88.7%\r",
      "Progress:59.8% Speed(reviews/sec):1569. #Correct:531 #Tested:599 Testing Accuracy:88.6%\r",
      "Progress:59.9% Speed(reviews/sec):1568. #Correct:532 #Tested:600 Testing Accuracy:88.6%\r",
      "Progress:60.0% Speed(reviews/sec):1568. #Correct:532 #Tested:601 Testing Accuracy:88.5%\r",
      "Progress:60.1% Speed(reviews/sec):1569. #Correct:533 #Tested:602 Testing Accuracy:88.5%\r",
      "Progress:60.2% Speed(reviews/sec):1570. #Correct:533 #Tested:603 Testing Accuracy:88.3%\r",
      "Progress:60.3% Speed(reviews/sec):1565. #Correct:534 #Tested:604 Testing Accuracy:88.4%\r",
      "Progress:60.4% Speed(reviews/sec):1565. #Correct:535 #Tested:605 Testing Accuracy:88.4%\r",
      "Progress:60.5% Speed(reviews/sec):1564. #Correct:535 #Tested:606 Testing Accuracy:88.2%\r",
      "Progress:60.6% Speed(reviews/sec):1565. #Correct:535 #Tested:607 Testing Accuracy:88.1%\r",
      "Progress:60.7% Speed(reviews/sec):1566. #Correct:536 #Tested:608 Testing Accuracy:88.1%\r",
      "Progress:60.8% Speed(reviews/sec):1568. #Correct:537 #Tested:609 Testing Accuracy:88.1%\r",
      "Progress:60.9% Speed(reviews/sec):1569. #Correct:538 #Tested:610 Testing Accuracy:88.1%\r",
      "Progress:61.0% Speed(reviews/sec):1569. #Correct:538 #Tested:611 Testing Accuracy:88.0%\r",
      "Progress:61.1% Speed(reviews/sec):1570. #Correct:539 #Tested:612 Testing Accuracy:88.0%\r",
      "Progress:61.2% Speed(reviews/sec):1572. #Correct:540 #Tested:613 Testing Accuracy:88.0%\r",
      "Progress:61.3% Speed(reviews/sec):1573. #Correct:541 #Tested:614 Testing Accuracy:88.1%\r",
      "Progress:61.4% Speed(reviews/sec):1575. #Correct:542 #Tested:615 Testing Accuracy:88.1%\r",
      "Progress:61.5% Speed(reviews/sec):1576. #Correct:542 #Tested:616 Testing Accuracy:87.9%\r",
      "Progress:61.6% Speed(reviews/sec):1577. #Correct:543 #Tested:617 Testing Accuracy:88.0%\r",
      "Progress:61.7% Speed(reviews/sec):1578. #Correct:544 #Tested:618 Testing Accuracy:88.0%\r",
      "Progress:61.8% Speed(reviews/sec):1579. #Correct:545 #Tested:619 Testing Accuracy:88.0%\r",
      "Progress:61.9% Speed(reviews/sec):1581. #Correct:546 #Tested:620 Testing Accuracy:88.0%\r",
      "Progress:62.0% Speed(reviews/sec):1580. #Correct:546 #Tested:621 Testing Accuracy:87.9%\r",
      "Progress:62.1% Speed(reviews/sec):1581. #Correct:547 #Tested:622 Testing Accuracy:87.9%\r",
      "Progress:62.2% Speed(reviews/sec):1582. #Correct:547 #Tested:623 Testing Accuracy:87.8%\r",
      "Progress:62.3% Speed(reviews/sec):1583. #Correct:547 #Tested:624 Testing Accuracy:87.6%\r",
      "Progress:62.4% Speed(reviews/sec):1584. #Correct:547 #Tested:625 Testing Accuracy:87.5%\r",
      "Progress:62.5% Speed(reviews/sec):1581. #Correct:547 #Tested:626 Testing Accuracy:87.3%\r",
      "Progress:62.6% Speed(reviews/sec):1581. #Correct:547 #Tested:627 Testing Accuracy:87.2%\r",
      "Progress:62.7% Speed(reviews/sec):1580. #Correct:548 #Tested:628 Testing Accuracy:87.2%\r",
      "Progress:62.8% Speed(reviews/sec):1582. #Correct:549 #Tested:629 Testing Accuracy:87.2%\r",
      "Progress:62.9% Speed(reviews/sec):1581. #Correct:549 #Tested:630 Testing Accuracy:87.1%\r",
      "Progress:63.0% Speed(reviews/sec):1582. #Correct:549 #Tested:631 Testing Accuracy:87.0%\r",
      "Progress:63.1% Speed(reviews/sec):1573. #Correct:550 #Tested:632 Testing Accuracy:87.0%\r",
      "Progress:63.2% Speed(reviews/sec):1574. #Correct:551 #Tested:633 Testing Accuracy:87.0%\r",
      "Progress:63.3% Speed(reviews/sec):1574. #Correct:551 #Tested:634 Testing Accuracy:86.9%\r",
      "Progress:63.4% Speed(reviews/sec):1572. #Correct:551 #Tested:635 Testing Accuracy:86.7%\r",
      "Progress:63.5% Speed(reviews/sec):1572. #Correct:551 #Tested:636 Testing Accuracy:86.6%\r",
      "Progress:63.6% Speed(reviews/sec):1573. #Correct:552 #Tested:637 Testing Accuracy:86.6%\r",
      "Progress:63.7% Speed(reviews/sec):1573. #Correct:553 #Tested:638 Testing Accuracy:86.6%\r",
      "Progress:63.8% Speed(reviews/sec):1575. #Correct:554 #Tested:639 Testing Accuracy:86.6%\r",
      "Progress:63.9% Speed(reviews/sec):1576. #Correct:554 #Tested:640 Testing Accuracy:86.5%\r",
      "Progress:64.0% Speed(reviews/sec):1578. #Correct:555 #Tested:641 Testing Accuracy:86.5%\r",
      "Progress:64.1% Speed(reviews/sec):1577. #Correct:555 #Tested:642 Testing Accuracy:86.4%\r",
      "Progress:64.2% Speed(reviews/sec):1578. #Correct:556 #Tested:643 Testing Accuracy:86.4%\r",
      "Progress:64.3% Speed(reviews/sec):1578. #Correct:557 #Tested:644 Testing Accuracy:86.4%\r",
      "Progress:64.4% Speed(reviews/sec):1578. #Correct:558 #Tested:645 Testing Accuracy:86.5%\r",
      "Progress:64.5% Speed(reviews/sec):1577. #Correct:559 #Tested:646 Testing Accuracy:86.5%\r",
      "Progress:64.6% Speed(reviews/sec):1576. #Correct:560 #Tested:647 Testing Accuracy:86.5%\r",
      "Progress:64.7% Speed(reviews/sec):1571. #Correct:561 #Tested:648 Testing Accuracy:86.5%\r",
      "Progress:64.8% Speed(reviews/sec):1570. #Correct:562 #Tested:649 Testing Accuracy:86.5%\r",
      "Progress:64.9% Speed(reviews/sec):1571. #Correct:563 #Tested:650 Testing Accuracy:86.6%\r",
      "Progress:65.0% Speed(reviews/sec):1572. #Correct:564 #Tested:651 Testing Accuracy:86.6%\r",
      "Progress:65.1% Speed(reviews/sec):1572. #Correct:565 #Tested:652 Testing Accuracy:86.6%\r",
      "Progress:65.2% Speed(reviews/sec):1573. #Correct:566 #Tested:653 Testing Accuracy:86.6%\r",
      "Progress:65.3% Speed(reviews/sec):1573. #Correct:566 #Tested:654 Testing Accuracy:86.5%\r",
      "Progress:65.4% Speed(reviews/sec):1574. #Correct:566 #Tested:655 Testing Accuracy:86.4%\r",
      "Progress:65.5% Speed(reviews/sec):1573. #Correct:567 #Tested:656 Testing Accuracy:86.4%\r",
      "Progress:65.6% Speed(reviews/sec):1574. #Correct:568 #Tested:657 Testing Accuracy:86.4%\r",
      "Progress:65.7% Speed(reviews/sec):1573. #Correct:569 #Tested:658 Testing Accuracy:86.4%\r",
      "Progress:65.8% Speed(reviews/sec):1574. #Correct:570 #Tested:659 Testing Accuracy:86.4%\r",
      "Progress:65.9% Speed(reviews/sec):1574. #Correct:571 #Tested:660 Testing Accuracy:86.5%\r",
      "Progress:66.0% Speed(reviews/sec):1575. #Correct:572 #Tested:661 Testing Accuracy:86.5%\r",
      "Progress:66.1% Speed(reviews/sec):1576. #Correct:573 #Tested:662 Testing Accuracy:86.5%\r",
      "Progress:66.2% Speed(reviews/sec):1577. #Correct:574 #Tested:663 Testing Accuracy:86.5%\r",
      "Progress:66.3% Speed(reviews/sec):1577. #Correct:575 #Tested:664 Testing Accuracy:86.5%\r",
      "Progress:66.4% Speed(reviews/sec):1578. #Correct:576 #Tested:665 Testing Accuracy:86.6%\r",
      "Progress:66.5% Speed(reviews/sec):1574. #Correct:577 #Tested:666 Testing Accuracy:86.6%\r",
      "Progress:66.6% Speed(reviews/sec):1575. #Correct:578 #Tested:667 Testing Accuracy:86.6%\r",
      "Progress:66.7% Speed(reviews/sec):1576. #Correct:579 #Tested:668 Testing Accuracy:86.6%\r",
      "Progress:66.8% Speed(reviews/sec):1576. #Correct:580 #Tested:669 Testing Accuracy:86.6%\r",
      "Progress:66.9% Speed(reviews/sec):1577. #Correct:580 #Tested:670 Testing Accuracy:86.5%\r",
      "Progress:67.0% Speed(reviews/sec):1578. #Correct:581 #Tested:671 Testing Accuracy:86.5%\r",
      "Progress:67.1% Speed(reviews/sec):1579. #Correct:582 #Tested:672 Testing Accuracy:86.6%\r",
      "Progress:67.2% Speed(reviews/sec):1581. #Correct:582 #Tested:673 Testing Accuracy:86.4%\r",
      "Progress:67.3% Speed(reviews/sec):1582. #Correct:583 #Tested:674 Testing Accuracy:86.4%\r",
      "Progress:67.4% Speed(reviews/sec):1583. #Correct:584 #Tested:675 Testing Accuracy:86.5%\r",
      "Progress:67.5% Speed(reviews/sec):1584. #Correct:584 #Tested:676 Testing Accuracy:86.3%\r",
      "Progress:67.6% Speed(reviews/sec):1585. #Correct:584 #Tested:677 Testing Accuracy:86.2%\r",
      "Progress:67.7% Speed(reviews/sec):1586. #Correct:584 #Tested:678 Testing Accuracy:86.1%\r",
      "Progress:67.8% Speed(reviews/sec):1587. #Correct:585 #Tested:679 Testing Accuracy:86.1%\r",
      "Progress:67.9% Speed(reviews/sec):1588. #Correct:586 #Tested:680 Testing Accuracy:86.1%\r",
      "Progress:68.0% Speed(reviews/sec):1589. #Correct:586 #Tested:681 Testing Accuracy:86.0%\r",
      "Progress:68.1% Speed(reviews/sec):1590. #Correct:587 #Tested:682 Testing Accuracy:86.0%\r",
      "Progress:68.2% Speed(reviews/sec):1591. #Correct:588 #Tested:683 Testing Accuracy:86.0%\r",
      "Progress:68.3% Speed(reviews/sec):1593. #Correct:589 #Tested:684 Testing Accuracy:86.1%\r",
      "Progress:68.4% Speed(reviews/sec):1594. #Correct:590 #Tested:685 Testing Accuracy:86.1%\r",
      "Progress:68.5% Speed(reviews/sec):1590. #Correct:591 #Tested:686 Testing Accuracy:86.1%\r",
      "Progress:68.6% Speed(reviews/sec):1591. #Correct:591 #Tested:687 Testing Accuracy:86.0%\r",
      "Progress:68.7% Speed(reviews/sec):1592. #Correct:592 #Tested:688 Testing Accuracy:86.0%\r",
      "Progress:68.8% Speed(reviews/sec):1590. #Correct:593 #Tested:689 Testing Accuracy:86.0%\r",
      "Progress:68.9% Speed(reviews/sec):1591. #Correct:594 #Tested:690 Testing Accuracy:86.0%\r",
      "Progress:69.0% Speed(reviews/sec):1592. #Correct:595 #Tested:691 Testing Accuracy:86.1%\r",
      "Progress:69.1% Speed(reviews/sec):1589. #Correct:596 #Tested:692 Testing Accuracy:86.1%\r",
      "Progress:69.2% Speed(reviews/sec):1590. #Correct:597 #Tested:693 Testing Accuracy:86.1%\r",
      "Progress:69.3% Speed(reviews/sec):1590. #Correct:598 #Tested:694 Testing Accuracy:86.1%\r",
      "Progress:69.4% Speed(reviews/sec):1589. #Correct:599 #Tested:695 Testing Accuracy:86.1%\r",
      "Progress:69.5% Speed(reviews/sec):1589. #Correct:600 #Tested:696 Testing Accuracy:86.2%\r",
      "Progress:69.6% Speed(reviews/sec):1590. #Correct:601 #Tested:697 Testing Accuracy:86.2%\r",
      "Progress:69.7% Speed(reviews/sec):1591. #Correct:602 #Tested:698 Testing Accuracy:86.2%\r",
      "Progress:69.8% Speed(reviews/sec):1592. #Correct:603 #Tested:699 Testing Accuracy:86.2%\r",
      "Progress:69.9% Speed(reviews/sec):1593. #Correct:604 #Tested:700 Testing Accuracy:86.2%\r",
      "Progress:70.0% Speed(reviews/sec):1593. #Correct:605 #Tested:701 Testing Accuracy:86.3%\r",
      "Progress:70.1% Speed(reviews/sec):1594. #Correct:606 #Tested:702 Testing Accuracy:86.3%\r",
      "Progress:70.2% Speed(reviews/sec):1591. #Correct:607 #Tested:703 Testing Accuracy:86.3%\r",
      "Progress:70.3% Speed(reviews/sec):1591. #Correct:608 #Tested:704 Testing Accuracy:86.3%\r",
      "Progress:70.4% Speed(reviews/sec):1591. #Correct:609 #Tested:705 Testing Accuracy:86.3%\r",
      "Progress:70.5% Speed(reviews/sec):1592. #Correct:610 #Tested:706 Testing Accuracy:86.4%\r",
      "Progress:70.6% Speed(reviews/sec):1593. #Correct:611 #Tested:707 Testing Accuracy:86.4%\r",
      "Progress:70.7% Speed(reviews/sec):1594. #Correct:612 #Tested:708 Testing Accuracy:86.4%\r",
      "Progress:70.8% Speed(reviews/sec):1594. #Correct:613 #Tested:709 Testing Accuracy:86.4%\r",
      "Progress:70.9% Speed(reviews/sec):1592. #Correct:614 #Tested:710 Testing Accuracy:86.4%\r",
      "Progress:71.0% Speed(reviews/sec):1593. #Correct:614 #Tested:711 Testing Accuracy:86.3%\r",
      "Progress:71.1% Speed(reviews/sec):1594. #Correct:615 #Tested:712 Testing Accuracy:86.3%\r",
      "Progress:71.2% Speed(reviews/sec):1595. #Correct:616 #Tested:713 Testing Accuracy:86.3%\r",
      "Progress:71.3% Speed(reviews/sec):1596. #Correct:617 #Tested:714 Testing Accuracy:86.4%\r",
      "Progress:71.4% Speed(reviews/sec):1596. #Correct:618 #Tested:715 Testing Accuracy:86.4%\r",
      "Progress:71.5% Speed(reviews/sec):1598. #Correct:619 #Tested:716 Testing Accuracy:86.4%\r",
      "Progress:71.6% Speed(reviews/sec):1599. #Correct:620 #Tested:717 Testing Accuracy:86.4%\r",
      "Progress:71.7% Speed(reviews/sec):1600. #Correct:621 #Tested:718 Testing Accuracy:86.4%\r",
      "Progress:71.8% Speed(reviews/sec):1598. #Correct:621 #Tested:719 Testing Accuracy:86.3%\r",
      "Progress:71.9% Speed(reviews/sec):1599. #Correct:622 #Tested:720 Testing Accuracy:86.3%\r",
      "Progress:72.0% Speed(reviews/sec):1600. #Correct:623 #Tested:721 Testing Accuracy:86.4%\r",
      "Progress:72.1% Speed(reviews/sec):1601. #Correct:624 #Tested:722 Testing Accuracy:86.4%\r",
      "Progress:72.2% Speed(reviews/sec):1600. #Correct:625 #Tested:723 Testing Accuracy:86.4%\r",
      "Progress:72.3% Speed(reviews/sec):1601. #Correct:625 #Tested:724 Testing Accuracy:86.3%\r",
      "Progress:72.4% Speed(reviews/sec):1602. #Correct:625 #Tested:725 Testing Accuracy:86.2%\r",
      "Progress:72.5% Speed(reviews/sec):1603. #Correct:626 #Tested:726 Testing Accuracy:86.2%\r",
      "Progress:72.6% Speed(reviews/sec):1601. #Correct:627 #Tested:727 Testing Accuracy:86.2%\r",
      "Progress:72.7% Speed(reviews/sec):1600. #Correct:628 #Tested:728 Testing Accuracy:86.2%\r",
      "Progress:72.8% Speed(reviews/sec):1602. #Correct:629 #Tested:729 Testing Accuracy:86.2%\r",
      "Progress:72.9% Speed(reviews/sec):1601. #Correct:630 #Tested:730 Testing Accuracy:86.3%\r",
      "Progress:73.0% Speed(reviews/sec):1601. #Correct:631 #Tested:731 Testing Accuracy:86.3%\r",
      "Progress:73.1% Speed(reviews/sec):1603. #Correct:632 #Tested:732 Testing Accuracy:86.3%\r",
      "Progress:73.2% Speed(reviews/sec):1603. #Correct:633 #Tested:733 Testing Accuracy:86.3%\r",
      "Progress:73.3% Speed(reviews/sec):1602. #Correct:634 #Tested:734 Testing Accuracy:86.3%\r",
      "Progress:73.4% Speed(reviews/sec):1601. #Correct:635 #Tested:735 Testing Accuracy:86.3%\r",
      "Progress:73.5% Speed(reviews/sec):1602. #Correct:636 #Tested:736 Testing Accuracy:86.4%\r",
      "Progress:73.6% Speed(reviews/sec):1603. #Correct:637 #Tested:737 Testing Accuracy:86.4%\r",
      "Progress:73.7% Speed(reviews/sec):1604. #Correct:638 #Tested:738 Testing Accuracy:86.4%\r",
      "Progress:73.8% Speed(reviews/sec):1604. #Correct:639 #Tested:739 Testing Accuracy:86.4%\r",
      "Progress:73.9% Speed(reviews/sec):1605. #Correct:640 #Tested:740 Testing Accuracy:86.4%\r",
      "Progress:74.0% Speed(reviews/sec):1606. #Correct:641 #Tested:741 Testing Accuracy:86.5%\r",
      "Progress:74.1% Speed(reviews/sec):1607. #Correct:642 #Tested:742 Testing Accuracy:86.5%\r",
      "Progress:74.2% Speed(reviews/sec):1608. #Correct:643 #Tested:743 Testing Accuracy:86.5%\r",
      "Progress:74.3% Speed(reviews/sec):1609. #Correct:643 #Tested:744 Testing Accuracy:86.4%\r",
      "Progress:74.4% Speed(reviews/sec):1609. #Correct:644 #Tested:745 Testing Accuracy:86.4%\r",
      "Progress:74.5% Speed(reviews/sec):1608. #Correct:645 #Tested:746 Testing Accuracy:86.4%"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\r",
      "Progress:74.6% Speed(reviews/sec):1607. #Correct:646 #Tested:747 Testing Accuracy:86.4%\r",
      "Progress:74.7% Speed(reviews/sec):1606. #Correct:646 #Tested:748 Testing Accuracy:86.3%\r",
      "Progress:74.8% Speed(reviews/sec):1606. #Correct:647 #Tested:749 Testing Accuracy:86.3%\r",
      "Progress:74.9% Speed(reviews/sec):1603. #Correct:648 #Tested:750 Testing Accuracy:86.4%\r",
      "Progress:75.0% Speed(reviews/sec):1604. #Correct:649 #Tested:751 Testing Accuracy:86.4%\r",
      "Progress:75.1% Speed(reviews/sec):1605. #Correct:650 #Tested:752 Testing Accuracy:86.4%\r",
      "Progress:75.2% Speed(reviews/sec):1606. #Correct:651 #Tested:753 Testing Accuracy:86.4%\r",
      "Progress:75.3% Speed(reviews/sec):1606. #Correct:652 #Tested:754 Testing Accuracy:86.4%\r",
      "Progress:75.4% Speed(reviews/sec):1605. #Correct:653 #Tested:755 Testing Accuracy:86.4%\r",
      "Progress:75.5% Speed(reviews/sec):1605. #Correct:653 #Tested:756 Testing Accuracy:86.3%\r",
      "Progress:75.6% Speed(reviews/sec):1606. #Correct:653 #Tested:757 Testing Accuracy:86.2%\r",
      "Progress:75.7% Speed(reviews/sec):1607. #Correct:654 #Tested:758 Testing Accuracy:86.2%\r",
      "Progress:75.8% Speed(reviews/sec):1607. #Correct:654 #Tested:759 Testing Accuracy:86.1%\r",
      "Progress:75.9% Speed(reviews/sec):1608. #Correct:655 #Tested:760 Testing Accuracy:86.1%\r",
      "Progress:76.0% Speed(reviews/sec):1609. #Correct:655 #Tested:761 Testing Accuracy:86.0%\r",
      "Progress:76.1% Speed(reviews/sec):1609. #Correct:656 #Tested:762 Testing Accuracy:86.0%\r",
      "Progress:76.2% Speed(reviews/sec):1609. #Correct:656 #Tested:763 Testing Accuracy:85.9%\r",
      "Progress:76.3% Speed(reviews/sec):1610. #Correct:657 #Tested:764 Testing Accuracy:85.9%\r",
      "Progress:76.4% Speed(reviews/sec):1610. #Correct:658 #Tested:765 Testing Accuracy:86.0%\r",
      "Progress:76.5% Speed(reviews/sec):1610. #Correct:658 #Tested:766 Testing Accuracy:85.9%\r",
      "Progress:76.6% Speed(reviews/sec):1611. #Correct:659 #Tested:767 Testing Accuracy:85.9%\r",
      "Progress:76.7% Speed(reviews/sec):1612. #Correct:660 #Tested:768 Testing Accuracy:85.9%\r",
      "Progress:76.8% Speed(reviews/sec):1613. #Correct:660 #Tested:769 Testing Accuracy:85.8%\r",
      "Progress:76.9% Speed(reviews/sec):1613. #Correct:661 #Tested:770 Testing Accuracy:85.8%\r",
      "Progress:77.0% Speed(reviews/sec):1613. #Correct:662 #Tested:771 Testing Accuracy:85.8%\r",
      "Progress:77.1% Speed(reviews/sec):1614. #Correct:662 #Tested:772 Testing Accuracy:85.7%\r",
      "Progress:77.2% Speed(reviews/sec):1615. #Correct:663 #Tested:773 Testing Accuracy:85.7%\r",
      "Progress:77.3% Speed(reviews/sec):1615. #Correct:664 #Tested:774 Testing Accuracy:85.7%\r",
      "Progress:77.4% Speed(reviews/sec):1615. #Correct:665 #Tested:775 Testing Accuracy:85.8%\r",
      "Progress:77.5% Speed(reviews/sec):1615. #Correct:666 #Tested:776 Testing Accuracy:85.8%\r",
      "Progress:77.6% Speed(reviews/sec):1615. #Correct:667 #Tested:777 Testing Accuracy:85.8%\r",
      "Progress:77.7% Speed(reviews/sec):1615. #Correct:667 #Tested:778 Testing Accuracy:85.7%\r",
      "Progress:77.8% Speed(reviews/sec):1613. #Correct:668 #Tested:779 Testing Accuracy:85.7%\r",
      "Progress:77.9% Speed(reviews/sec):1614. #Correct:669 #Tested:780 Testing Accuracy:85.7%\r",
      "Progress:78.0% Speed(reviews/sec):1612. #Correct:669 #Tested:781 Testing Accuracy:85.6%\r",
      "Progress:78.1% Speed(reviews/sec):1613. #Correct:670 #Tested:782 Testing Accuracy:85.6%\r",
      "Progress:78.2% Speed(reviews/sec):1613. #Correct:670 #Tested:783 Testing Accuracy:85.5%\r",
      "Progress:78.3% Speed(reviews/sec):1614. #Correct:671 #Tested:784 Testing Accuracy:85.5%\r",
      "Progress:78.4% Speed(reviews/sec):1613. #Correct:672 #Tested:785 Testing Accuracy:85.6%\r",
      "Progress:78.5% Speed(reviews/sec):1614. #Correct:672 #Tested:786 Testing Accuracy:85.4%\r",
      "Progress:78.6% Speed(reviews/sec):1614. #Correct:673 #Tested:787 Testing Accuracy:85.5%\r",
      "Progress:78.7% Speed(reviews/sec):1614. #Correct:674 #Tested:788 Testing Accuracy:85.5%\r",
      "Progress:78.8% Speed(reviews/sec):1615. #Correct:674 #Tested:789 Testing Accuracy:85.4%\r",
      "Progress:78.9% Speed(reviews/sec):1614. #Correct:675 #Tested:790 Testing Accuracy:85.4%\r",
      "Progress:79.0% Speed(reviews/sec):1614. #Correct:676 #Tested:791 Testing Accuracy:85.4%\r",
      "Progress:79.1% Speed(reviews/sec):1614. #Correct:677 #Tested:792 Testing Accuracy:85.4%\r",
      "Progress:79.2% Speed(reviews/sec):1613. #Correct:677 #Tested:793 Testing Accuracy:85.3%\r",
      "Progress:79.3% Speed(reviews/sec):1614. #Correct:678 #Tested:794 Testing Accuracy:85.3%\r",
      "Progress:79.4% Speed(reviews/sec):1614. #Correct:679 #Tested:795 Testing Accuracy:85.4%\r",
      "Progress:79.5% Speed(reviews/sec):1615. #Correct:680 #Tested:796 Testing Accuracy:85.4%\r",
      "Progress:79.6% Speed(reviews/sec):1616. #Correct:681 #Tested:797 Testing Accuracy:85.4%\r",
      "Progress:79.7% Speed(reviews/sec):1616. #Correct:682 #Tested:798 Testing Accuracy:85.4%\r",
      "Progress:79.8% Speed(reviews/sec):1618. #Correct:683 #Tested:799 Testing Accuracy:85.4%\r",
      "Progress:79.9% Speed(reviews/sec):1619. #Correct:684 #Tested:800 Testing Accuracy:85.5%\r",
      "Progress:80.0% Speed(reviews/sec):1620. #Correct:685 #Tested:801 Testing Accuracy:85.5%\r",
      "Progress:80.1% Speed(reviews/sec):1621. #Correct:686 #Tested:802 Testing Accuracy:85.5%\r",
      "Progress:80.2% Speed(reviews/sec):1622. #Correct:687 #Tested:803 Testing Accuracy:85.5%\r",
      "Progress:80.3% Speed(reviews/sec):1623. #Correct:688 #Tested:804 Testing Accuracy:85.5%\r",
      "Progress:80.4% Speed(reviews/sec):1624. #Correct:689 #Tested:805 Testing Accuracy:85.5%\r",
      "Progress:80.5% Speed(reviews/sec):1625. #Correct:690 #Tested:806 Testing Accuracy:85.6%\r",
      "Progress:80.6% Speed(reviews/sec):1626. #Correct:691 #Tested:807 Testing Accuracy:85.6%\r",
      "Progress:80.7% Speed(reviews/sec):1626. #Correct:691 #Tested:808 Testing Accuracy:85.5%\r",
      "Progress:80.8% Speed(reviews/sec):1627. #Correct:692 #Tested:809 Testing Accuracy:85.5%\r",
      "Progress:80.9% Speed(reviews/sec):1627. #Correct:693 #Tested:810 Testing Accuracy:85.5%\r",
      "Progress:81.0% Speed(reviews/sec):1625. #Correct:694 #Tested:811 Testing Accuracy:85.5%\r",
      "Progress:81.1% Speed(reviews/sec):1625. #Correct:695 #Tested:812 Testing Accuracy:85.5%\r",
      "Progress:81.2% Speed(reviews/sec):1618. #Correct:696 #Tested:813 Testing Accuracy:85.6%\r",
      "Progress:81.3% Speed(reviews/sec):1618. #Correct:697 #Tested:814 Testing Accuracy:85.6%\r",
      "Progress:81.4% Speed(reviews/sec):1610. #Correct:698 #Tested:815 Testing Accuracy:85.6%\r",
      "Progress:81.5% Speed(reviews/sec):1596. #Correct:699 #Tested:816 Testing Accuracy:85.6%\r",
      "Progress:81.6% Speed(reviews/sec):1595. #Correct:700 #Tested:817 Testing Accuracy:85.6%\r",
      "Progress:81.7% Speed(reviews/sec):1595. #Correct:701 #Tested:818 Testing Accuracy:85.6%\r",
      "Progress:81.8% Speed(reviews/sec):1595. #Correct:702 #Tested:819 Testing Accuracy:85.7%\r",
      "Progress:81.9% Speed(reviews/sec):1593. #Correct:703 #Tested:820 Testing Accuracy:85.7%\r",
      "Progress:82.0% Speed(reviews/sec):1589. #Correct:704 #Tested:821 Testing Accuracy:85.7%\r",
      "Progress:82.1% Speed(reviews/sec):1585. #Correct:704 #Tested:822 Testing Accuracy:85.6%\r",
      "Progress:82.2% Speed(reviews/sec):1586. #Correct:705 #Tested:823 Testing Accuracy:85.6%\r",
      "Progress:82.3% Speed(reviews/sec):1586. #Correct:706 #Tested:824 Testing Accuracy:85.6%\r",
      "Progress:82.4% Speed(reviews/sec):1585. #Correct:707 #Tested:825 Testing Accuracy:85.6%\r",
      "Progress:82.5% Speed(reviews/sec):1586. #Correct:707 #Tested:826 Testing Accuracy:85.5%\r",
      "Progress:82.6% Speed(reviews/sec):1578. #Correct:708 #Tested:827 Testing Accuracy:85.6%\r",
      "Progress:82.7% Speed(reviews/sec):1578. #Correct:709 #Tested:828 Testing Accuracy:85.6%\r",
      "Progress:82.8% Speed(reviews/sec):1577. #Correct:710 #Tested:829 Testing Accuracy:85.6%\r",
      "Progress:82.9% Speed(reviews/sec):1578. #Correct:711 #Tested:830 Testing Accuracy:85.6%\r",
      "Progress:83.0% Speed(reviews/sec):1579. #Correct:712 #Tested:831 Testing Accuracy:85.6%\r",
      "Progress:83.1% Speed(reviews/sec):1578. #Correct:713 #Tested:832 Testing Accuracy:85.6%\r",
      "Progress:83.2% Speed(reviews/sec):1579. #Correct:714 #Tested:833 Testing Accuracy:85.7%\r",
      "Progress:83.3% Speed(reviews/sec):1579. #Correct:715 #Tested:834 Testing Accuracy:85.7%\r",
      "Progress:83.4% Speed(reviews/sec):1580. #Correct:716 #Tested:835 Testing Accuracy:85.7%\r",
      "Progress:83.5% Speed(reviews/sec):1581. #Correct:717 #Tested:836 Testing Accuracy:85.7%\r",
      "Progress:83.6% Speed(reviews/sec):1580. #Correct:718 #Tested:837 Testing Accuracy:85.7%\r",
      "Progress:83.7% Speed(reviews/sec):1578. #Correct:719 #Tested:838 Testing Accuracy:85.7%\r",
      "Progress:83.8% Speed(reviews/sec):1578. #Correct:719 #Tested:839 Testing Accuracy:85.6%\r",
      "Progress:83.9% Speed(reviews/sec):1579. #Correct:720 #Tested:840 Testing Accuracy:85.7%\r",
      "Progress:84.0% Speed(reviews/sec):1580. #Correct:721 #Tested:841 Testing Accuracy:85.7%\r",
      "Progress:84.1% Speed(reviews/sec):1580. #Correct:722 #Tested:842 Testing Accuracy:85.7%\r",
      "Progress:84.2% Speed(reviews/sec):1581. #Correct:723 #Tested:843 Testing Accuracy:85.7%\r",
      "Progress:84.3% Speed(reviews/sec):1581. #Correct:724 #Tested:844 Testing Accuracy:85.7%\r",
      "Progress:84.4% Speed(reviews/sec):1581. #Correct:725 #Tested:845 Testing Accuracy:85.7%\r",
      "Progress:84.5% Speed(reviews/sec):1582. #Correct:726 #Tested:846 Testing Accuracy:85.8%\r",
      "Progress:84.6% Speed(reviews/sec):1582. #Correct:727 #Tested:847 Testing Accuracy:85.8%\r",
      "Progress:84.7% Speed(reviews/sec):1582. #Correct:727 #Tested:848 Testing Accuracy:85.7%\r",
      "Progress:84.8% Speed(reviews/sec):1583. #Correct:728 #Tested:849 Testing Accuracy:85.7%\r",
      "Progress:84.9% Speed(reviews/sec):1582. #Correct:729 #Tested:850 Testing Accuracy:85.7%\r",
      "Progress:85.0% Speed(reviews/sec):1581. #Correct:730 #Tested:851 Testing Accuracy:85.7%\r",
      "Progress:85.1% Speed(reviews/sec):1582. #Correct:731 #Tested:852 Testing Accuracy:85.7%\r",
      "Progress:85.2% Speed(reviews/sec):1581. #Correct:732 #Tested:853 Testing Accuracy:85.8%\r",
      "Progress:85.3% Speed(reviews/sec):1581. #Correct:733 #Tested:854 Testing Accuracy:85.8%\r",
      "Progress:85.4% Speed(reviews/sec):1582. #Correct:734 #Tested:855 Testing Accuracy:85.8%\r",
      "Progress:85.5% Speed(reviews/sec):1583. #Correct:735 #Tested:856 Testing Accuracy:85.8%\r",
      "Progress:85.6% Speed(reviews/sec):1584. #Correct:736 #Tested:857 Testing Accuracy:85.8%\r",
      "Progress:85.7% Speed(reviews/sec):1584. #Correct:736 #Tested:858 Testing Accuracy:85.7%\r",
      "Progress:85.8% Speed(reviews/sec):1575. #Correct:737 #Tested:859 Testing Accuracy:85.7%\r",
      "Progress:85.9% Speed(reviews/sec):1576. #Correct:738 #Tested:860 Testing Accuracy:85.8%\r",
      "Progress:86.0% Speed(reviews/sec):1577. #Correct:739 #Tested:861 Testing Accuracy:85.8%\r",
      "Progress:86.1% Speed(reviews/sec):1577. #Correct:740 #Tested:862 Testing Accuracy:85.8%\r",
      "Progress:86.2% Speed(reviews/sec):1577. #Correct:741 #Tested:863 Testing Accuracy:85.8%\r",
      "Progress:86.3% Speed(reviews/sec):1576. #Correct:742 #Tested:864 Testing Accuracy:85.8%\r",
      "Progress:86.4% Speed(reviews/sec):1576. #Correct:743 #Tested:865 Testing Accuracy:85.8%\r",
      "Progress:86.5% Speed(reviews/sec):1577. #Correct:744 #Tested:866 Testing Accuracy:85.9%\r",
      "Progress:86.6% Speed(reviews/sec):1578. #Correct:745 #Tested:867 Testing Accuracy:85.9%\r",
      "Progress:86.7% Speed(reviews/sec):1579. #Correct:746 #Tested:868 Testing Accuracy:85.9%\r",
      "Progress:86.8% Speed(reviews/sec):1580. #Correct:747 #Tested:869 Testing Accuracy:85.9%\r",
      "Progress:86.9% Speed(reviews/sec):1581. #Correct:748 #Tested:870 Testing Accuracy:85.9%\r",
      "Progress:87.0% Speed(reviews/sec):1581. #Correct:749 #Tested:871 Testing Accuracy:85.9%\r",
      "Progress:87.1% Speed(reviews/sec):1581. #Correct:750 #Tested:872 Testing Accuracy:86.0%\r",
      "Progress:87.2% Speed(reviews/sec):1581. #Correct:751 #Tested:873 Testing Accuracy:86.0%\r",
      "Progress:87.3% Speed(reviews/sec):1579. #Correct:752 #Tested:874 Testing Accuracy:86.0%\r",
      "Progress:87.4% Speed(reviews/sec):1580. #Correct:753 #Tested:875 Testing Accuracy:86.0%\r",
      "Progress:87.5% Speed(reviews/sec):1581. #Correct:754 #Tested:876 Testing Accuracy:86.0%\r",
      "Progress:87.6% Speed(reviews/sec):1582. #Correct:754 #Tested:877 Testing Accuracy:85.9%\r",
      "Progress:87.7% Speed(reviews/sec):1580. #Correct:755 #Tested:878 Testing Accuracy:85.9%\r",
      "Progress:87.8% Speed(reviews/sec):1581. #Correct:756 #Tested:879 Testing Accuracy:86.0%\r",
      "Progress:87.9% Speed(reviews/sec):1582. #Correct:757 #Tested:880 Testing Accuracy:86.0%\r",
      "Progress:88.0% Speed(reviews/sec):1582. #Correct:758 #Tested:881 Testing Accuracy:86.0%\r",
      "Progress:88.1% Speed(reviews/sec):1583. #Correct:759 #Tested:882 Testing Accuracy:86.0%\r",
      "Progress:88.2% Speed(reviews/sec):1584. #Correct:760 #Tested:883 Testing Accuracy:86.0%\r",
      "Progress:88.3% Speed(reviews/sec):1585. #Correct:761 #Tested:884 Testing Accuracy:86.0%\r",
      "Progress:88.4% Speed(reviews/sec):1586. #Correct:762 #Tested:885 Testing Accuracy:86.1%\r",
      "Progress:88.5% Speed(reviews/sec):1586. #Correct:763 #Tested:886 Testing Accuracy:86.1%\r",
      "Progress:88.6% Speed(reviews/sec):1584. #Correct:764 #Tested:887 Testing Accuracy:86.1%\r",
      "Progress:88.7% Speed(reviews/sec):1585. #Correct:765 #Tested:888 Testing Accuracy:86.1%\r",
      "Progress:88.8% Speed(reviews/sec):1586. #Correct:765 #Tested:889 Testing Accuracy:86.0%\r",
      "Progress:88.9% Speed(reviews/sec):1585. #Correct:765 #Tested:890 Testing Accuracy:85.9%\r",
      "Progress:89.0% Speed(reviews/sec):1586. #Correct:766 #Tested:891 Testing Accuracy:85.9%\r",
      "Progress:89.1% Speed(reviews/sec):1586. #Correct:767 #Tested:892 Testing Accuracy:85.9%\r",
      "Progress:89.2% Speed(reviews/sec):1588. #Correct:768 #Tested:893 Testing Accuracy:86.0%\r",
      "Progress:89.3% Speed(reviews/sec):1588. #Correct:769 #Tested:894 Testing Accuracy:86.0%\r",
      "Progress:89.4% Speed(reviews/sec):1589. #Correct:770 #Tested:895 Testing Accuracy:86.0%\r",
      "Progress:89.5% Speed(reviews/sec):1590. #Correct:771 #Tested:896 Testing Accuracy:86.0%\r",
      "Progress:89.6% Speed(reviews/sec):1591. #Correct:772 #Tested:897 Testing Accuracy:86.0%\r",
      "Progress:89.7% Speed(reviews/sec):1592. #Correct:773 #Tested:898 Testing Accuracy:86.0%\r",
      "Progress:89.8% Speed(reviews/sec):1590. #Correct:774 #Tested:899 Testing Accuracy:86.0%\r",
      "Progress:89.9% Speed(reviews/sec):1587. #Correct:775 #Tested:900 Testing Accuracy:86.1%\r",
      "Progress:90.0% Speed(reviews/sec):1587. #Correct:775 #Tested:901 Testing Accuracy:86.0%\r",
      "Progress:90.1% Speed(reviews/sec):1585. #Correct:776 #Tested:902 Testing Accuracy:86.0%\r",
      "Progress:90.2% Speed(reviews/sec):1585. #Correct:777 #Tested:903 Testing Accuracy:86.0%\r",
      "Progress:90.3% Speed(reviews/sec):1586. #Correct:778 #Tested:904 Testing Accuracy:86.0%\r",
      "Progress:90.4% Speed(reviews/sec):1586. #Correct:779 #Tested:905 Testing Accuracy:86.0%\r",
      "Progress:90.5% Speed(reviews/sec):1587. #Correct:780 #Tested:906 Testing Accuracy:86.0%\r",
      "Progress:90.6% Speed(reviews/sec):1588. #Correct:781 #Tested:907 Testing Accuracy:86.1%\r",
      "Progress:90.7% Speed(reviews/sec):1589. #Correct:782 #Tested:908 Testing Accuracy:86.1%\r",
      "Progress:90.8% Speed(reviews/sec):1587. #Correct:782 #Tested:909 Testing Accuracy:86.0%\r",
      "Progress:90.9% Speed(reviews/sec):1588. #Correct:782 #Tested:910 Testing Accuracy:85.9%\r",
      "Progress:91.0% Speed(reviews/sec):1588. #Correct:783 #Tested:911 Testing Accuracy:85.9%\r",
      "Progress:91.1% Speed(reviews/sec):1588. #Correct:783 #Tested:912 Testing Accuracy:85.8%\r",
      "Progress:91.2% Speed(reviews/sec):1588. #Correct:784 #Tested:913 Testing Accuracy:85.8%\r",
      "Progress:91.3% Speed(reviews/sec):1588. #Correct:784 #Tested:914 Testing Accuracy:85.7%\r",
      "Progress:91.4% Speed(reviews/sec):1589. #Correct:785 #Tested:915 Testing Accuracy:85.7%\r",
      "Progress:91.5% Speed(reviews/sec):1589. #Correct:786 #Tested:916 Testing Accuracy:85.8%\r",
      "Progress:91.6% Speed(reviews/sec):1589. #Correct:786 #Tested:917 Testing Accuracy:85.7%\r",
      "Progress:91.7% Speed(reviews/sec):1590. #Correct:787 #Tested:918 Testing Accuracy:85.7%\r",
      "Progress:91.8% Speed(reviews/sec):1590. #Correct:788 #Tested:919 Testing Accuracy:85.7%\r",
      "Progress:91.9% Speed(reviews/sec):1591. #Correct:789 #Tested:920 Testing Accuracy:85.7%\r",
      "Progress:92.0% Speed(reviews/sec):1591. #Correct:789 #Tested:921 Testing Accuracy:85.6%\r",
      "Progress:92.1% Speed(reviews/sec):1592. #Correct:790 #Tested:922 Testing Accuracy:85.6%\r",
      "Progress:92.2% Speed(reviews/sec):1593. #Correct:791 #Tested:923 Testing Accuracy:85.6%\r",
      "Progress:92.3% Speed(reviews/sec):1594. #Correct:792 #Tested:924 Testing Accuracy:85.7%\r",
      "Progress:92.4% Speed(reviews/sec):1595. #Correct:793 #Tested:925 Testing Accuracy:85.7%\r",
      "Progress:92.5% Speed(reviews/sec):1595. #Correct:794 #Tested:926 Testing Accuracy:85.7%\r",
      "Progress:92.6% Speed(reviews/sec):1596. #Correct:795 #Tested:927 Testing Accuracy:85.7%\r",
      "Progress:92.7% Speed(reviews/sec):1597. #Correct:796 #Tested:928 Testing Accuracy:85.7%\r",
      "Progress:92.8% Speed(reviews/sec):1598. #Correct:797 #Tested:929 Testing Accuracy:85.7%\r",
      "Progress:92.9% Speed(reviews/sec):1599. #Correct:798 #Tested:930 Testing Accuracy:85.8%\r",
      "Progress:93.0% Speed(reviews/sec):1596. #Correct:799 #Tested:931 Testing Accuracy:85.8%\r",
      "Progress:93.1% Speed(reviews/sec):1595. #Correct:800 #Tested:932 Testing Accuracy:85.8%\r",
      "Progress:93.2% Speed(reviews/sec):1595. #Correct:801 #Tested:933 Testing Accuracy:85.8%\r",
      "Progress:93.3% Speed(reviews/sec):1594. #Correct:802 #Tested:934 Testing Accuracy:85.8%\r",
      "Progress:93.4% Speed(reviews/sec):1595. #Correct:803 #Tested:935 Testing Accuracy:85.8%\r",
      "Progress:93.5% Speed(reviews/sec):1596. #Correct:804 #Tested:936 Testing Accuracy:85.8%\r",
      "Progress:93.6% Speed(reviews/sec):1595. #Correct:804 #Tested:937 Testing Accuracy:85.8%\r",
      "Progress:93.7% Speed(reviews/sec):1596. #Correct:805 #Tested:938 Testing Accuracy:85.8%\r",
      "Progress:93.8% Speed(reviews/sec):1597. #Correct:805 #Tested:939 Testing Accuracy:85.7%\r",
      "Progress:93.9% Speed(reviews/sec):1598. #Correct:806 #Tested:940 Testing Accuracy:85.7%\r",
      "Progress:94.0% Speed(reviews/sec):1599. #Correct:807 #Tested:941 Testing Accuracy:85.7%\r",
      "Progress:94.1% Speed(reviews/sec):1597. #Correct:807 #Tested:942 Testing Accuracy:85.6%\r",
      "Progress:94.2% Speed(reviews/sec):1598. #Correct:807 #Tested:943 Testing Accuracy:85.5%\r",
      "Progress:94.3% Speed(reviews/sec):1599. #Correct:808 #Tested:944 Testing Accuracy:85.5%\r",
      "Progress:94.4% Speed(reviews/sec):1597. #Correct:809 #Tested:945 Testing Accuracy:85.6%\r",
      "Progress:94.5% Speed(reviews/sec):1597. #Correct:809 #Tested:946 Testing Accuracy:85.5%\r",
      "Progress:94.6% Speed(reviews/sec):1598. #Correct:810 #Tested:947 Testing Accuracy:85.5%\r",
      "Progress:94.7% Speed(reviews/sec):1598. #Correct:811 #Tested:948 Testing Accuracy:85.5%\r",
      "Progress:94.8% Speed(reviews/sec):1599. #Correct:812 #Tested:949 Testing Accuracy:85.5%\r",
      "Progress:94.9% Speed(reviews/sec):1600. #Correct:813 #Tested:950 Testing Accuracy:85.5%\r",
      "Progress:95.0% Speed(reviews/sec):1601. #Correct:814 #Tested:951 Testing Accuracy:85.5%\r",
      "Progress:95.1% Speed(reviews/sec):1600. #Correct:815 #Tested:952 Testing Accuracy:85.6%\r",
      "Progress:95.2% Speed(reviews/sec):1599. #Correct:816 #Tested:953 Testing Accuracy:85.6%\r",
      "Progress:95.3% Speed(reviews/sec):1600. #Correct:817 #Tested:954 Testing Accuracy:85.6%\r",
      "Progress:95.4% Speed(reviews/sec):1599. #Correct:818 #Tested:955 Testing Accuracy:85.6%\r",
      "Progress:95.5% Speed(reviews/sec):1599. #Correct:819 #Tested:956 Testing Accuracy:85.6%\r",
      "Progress:95.6% Speed(reviews/sec):1599. #Correct:820 #Tested:957 Testing Accuracy:85.6%\r",
      "Progress:95.7% Speed(reviews/sec):1600. #Correct:821 #Tested:958 Testing Accuracy:85.6%\r",
      "Progress:95.8% Speed(reviews/sec):1601. #Correct:822 #Tested:959 Testing Accuracy:85.7%\r",
      "Progress:95.9% Speed(reviews/sec):1602. #Correct:823 #Tested:960 Testing Accuracy:85.7%\r",
      "Progress:96.0% Speed(reviews/sec):1602. #Correct:824 #Tested:961 Testing Accuracy:85.7%\r",
      "Progress:96.1% Speed(reviews/sec):1603. #Correct:825 #Tested:962 Testing Accuracy:85.7%\r",
      "Progress:96.2% Speed(reviews/sec):1603. #Correct:826 #Tested:963 Testing Accuracy:85.7%\r",
      "Progress:96.3% Speed(reviews/sec):1604. #Correct:827 #Tested:964 Testing Accuracy:85.7%\r",
      "Progress:96.4% Speed(reviews/sec):1605. #Correct:828 #Tested:965 Testing Accuracy:85.8%\r",
      "Progress:96.5% Speed(reviews/sec):1605. #Correct:829 #Tested:966 Testing Accuracy:85.8%\r",
      "Progress:96.6% Speed(reviews/sec):1606. #Correct:830 #Tested:967 Testing Accuracy:85.8%\r",
      "Progress:96.7% Speed(reviews/sec):1604. #Correct:831 #Tested:968 Testing Accuracy:85.8%\r",
      "Progress:96.8% Speed(reviews/sec):1605. #Correct:832 #Tested:969 Testing Accuracy:85.8%\r",
      "Progress:96.9% Speed(reviews/sec):1605. #Correct:833 #Tested:970 Testing Accuracy:85.8%\r",
      "Progress:97.0% Speed(reviews/sec):1605. #Correct:834 #Tested:971 Testing Accuracy:85.8%\r",
      "Progress:97.1% Speed(reviews/sec):1606. #Correct:835 #Tested:972 Testing Accuracy:85.9%\r",
      "Progress:97.2% Speed(reviews/sec):1607. #Correct:836 #Tested:973 Testing Accuracy:85.9%\r",
      "Progress:97.3% Speed(reviews/sec):1607. #Correct:837 #Tested:974 Testing Accuracy:85.9%\r",
      "Progress:97.4% Speed(reviews/sec):1605. #Correct:838 #Tested:975 Testing Accuracy:85.9%\r",
      "Progress:97.5% Speed(reviews/sec):1604. #Correct:839 #Tested:976 Testing Accuracy:85.9%\r",
      "Progress:97.6% Speed(reviews/sec):1605. #Correct:840 #Tested:977 Testing Accuracy:85.9%\r",
      "Progress:97.7% Speed(reviews/sec):1604. #Correct:840 #Tested:978 Testing Accuracy:85.8%\r",
      "Progress:97.8% Speed(reviews/sec):1605. #Correct:841 #Tested:979 Testing Accuracy:85.9%\r",
      "Progress:97.9% Speed(reviews/sec):1606. #Correct:842 #Tested:980 Testing Accuracy:85.9%\r",
      "Progress:98.0% Speed(reviews/sec):1605. #Correct:843 #Tested:981 Testing Accuracy:85.9%\r",
      "Progress:98.1% Speed(reviews/sec):1604. #Correct:844 #Tested:982 Testing Accuracy:85.9%\r",
      "Progress:98.2% Speed(reviews/sec):1604. #Correct:845 #Tested:983 Testing Accuracy:85.9%\r",
      "Progress:98.3% Speed(reviews/sec):1604. #Correct:846 #Tested:984 Testing Accuracy:85.9%\r",
      "Progress:98.4% Speed(reviews/sec):1605. #Correct:847 #Tested:985 Testing Accuracy:85.9%\r",
      "Progress:98.5% Speed(reviews/sec):1606. #Correct:848 #Tested:986 Testing Accuracy:86.0%\r",
      "Progress:98.6% Speed(reviews/sec):1606. #Correct:849 #Tested:987 Testing Accuracy:86.0%\r",
      "Progress:98.7% Speed(reviews/sec):1606. #Correct:850 #Tested:988 Testing Accuracy:86.0%\r",
      "Progress:98.8% Speed(reviews/sec):1607. #Correct:851 #Tested:989 Testing Accuracy:86.0%\r",
      "Progress:98.9% Speed(reviews/sec):1607. #Correct:852 #Tested:990 Testing Accuracy:86.0%\r",
      "Progress:99.0% Speed(reviews/sec):1608. #Correct:853 #Tested:991 Testing Accuracy:86.0%\r",
      "Progress:99.1% Speed(reviews/sec):1608. #Correct:853 #Tested:992 Testing Accuracy:85.9%\r",
      "Progress:99.2% Speed(reviews/sec):1606. #Correct:854 #Tested:993 Testing Accuracy:86.0%\r",
      "Progress:99.3% Speed(reviews/sec):1607. #Correct:855 #Tested:994 Testing Accuracy:86.0%\r",
      "Progress:99.4% Speed(reviews/sec):1606. #Correct:855 #Tested:995 Testing Accuracy:85.9%\r",
      "Progress:99.5% Speed(reviews/sec):1606. #Correct:856 #Tested:996 Testing Accuracy:85.9%\r",
      "Progress:99.6% Speed(reviews/sec):1606. #Correct:857 #Tested:997 Testing Accuracy:85.9%\r",
      "Progress:99.7% Speed(reviews/sec):1606. #Correct:858 #Tested:998 Testing Accuracy:85.9%\r",
      "Progress:99.8% Speed(reviews/sec):1607. #Correct:858 #Tested:999 Testing Accuracy:85.8%\r",
      "Progress:99.9% Speed(reviews/sec):1608. #Correct:859 #Tested:1000 Testing Accuracy:85.9%"
     ]
    }
   ],
   "source": [
    "mlp.test(reviews[-1000:],labels[-1000:])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "运行下方代码框，用大许多的 polarity cutoff 参数值来训练网络。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Progress:0.0% Speed(reviews/sec):0.0 #Correct:1 #Trained:1 Training Accuracy:100.%\n",
      "Progress:10.4% Speed(reviews/sec):8056. #Correct:2114 #Trained:2501 Training Accuracy:84.5%\n",
      "Progress:20.8% Speed(reviews/sec):7940. #Correct:4235 #Trained:5001 Training Accuracy:84.6%\n",
      "Progress:31.2% Speed(reviews/sec):7966. #Correct:6362 #Trained:7501 Training Accuracy:84.8%\n",
      "Progress:41.6% Speed(reviews/sec):7964. #Correct:8513 #Trained:10001 Training Accuracy:85.1%\n",
      "Progress:52.0% Speed(reviews/sec):7945. #Correct:10641 #Trained:12501 Training Accuracy:85.1%\n",
      "Progress:62.5% Speed(reviews/sec):7910. #Correct:12796 #Trained:15001 Training Accuracy:85.3%\n",
      "Progress:72.9% Speed(reviews/sec):7894. #Correct:14911 #Trained:17501 Training Accuracy:85.2%\n",
      "Progress:83.3% Speed(reviews/sec):7846. #Correct:17077 #Trained:20001 Training Accuracy:85.3%\n",
      "Progress:93.7% Speed(reviews/sec):7848. #Correct:19258 #Trained:22501 Training Accuracy:85.5%\n",
      "Progress:97.9% Speed(reviews/sec):7855. #Correct:20103 #Trained:23501 Training Accuracy:85.5%"
     ]
    }
   ],
   "source": [
    "mlp = SentimentNetwork(reviews[:-1000],labels[:-1000],min_count=20,polarity_cutoff=0.8,learning_rate=0.01)\n",
    "mlp.train(reviews[:-1000],labels[:-1000])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "运行下方代码框，测试网络性能。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\r",
      "Progress:0.0% Speed(reviews/sec):0.0 #Correct:0 #Tested:1 Testing Accuracy:0.0%\r",
      "Progress:0.1% Speed(reviews/sec):2206. #Correct:1 #Tested:2 Testing Accuracy:50.0%\r",
      "Progress:0.2% Speed(reviews/sec):3278. #Correct:2 #Tested:3 Testing Accuracy:66.6%\r",
      "Progress:0.3% Speed(reviews/sec):3971. #Correct:3 #Tested:4 Testing Accuracy:75.0%\r",
      "Progress:0.4% Speed(reviews/sec):4778. #Correct:3 #Tested:5 Testing Accuracy:60.0%\r",
      "Progress:0.5% Speed(reviews/sec):5235. #Correct:4 #Tested:6 Testing Accuracy:66.6%\r",
      "Progress:0.6% Speed(reviews/sec):5281. #Correct:5 #Tested:7 Testing Accuracy:71.4%\r",
      "Progress:0.7% Speed(reviews/sec):5413. #Correct:6 #Tested:8 Testing Accuracy:75.0%\r",
      "Progress:0.8% Speed(reviews/sec):5625. #Correct:7 #Tested:9 Testing Accuracy:77.7%\r",
      "Progress:0.9% Speed(reviews/sec):4706. #Correct:8 #Tested:10 Testing Accuracy:80.0%\r",
      "Progress:1.0% Speed(reviews/sec):4698. #Correct:9 #Tested:11 Testing Accuracy:81.8%\r",
      "Progress:1.1% Speed(reviews/sec):4767. #Correct:10 #Tested:12 Testing Accuracy:83.3%\r",
      "Progress:1.2% Speed(reviews/sec):4849. #Correct:10 #Tested:13 Testing Accuracy:76.9%\r",
      "Progress:1.3% Speed(reviews/sec):4881. #Correct:11 #Tested:14 Testing Accuracy:78.5%\r",
      "Progress:1.4% Speed(reviews/sec):4952. #Correct:11 #Tested:15 Testing Accuracy:73.3%\r",
      "Progress:1.5% Speed(reviews/sec):5013. #Correct:12 #Tested:16 Testing Accuracy:75.0%\r",
      "Progress:1.6% Speed(reviews/sec):4742. #Correct:13 #Tested:17 Testing Accuracy:76.4%\r",
      "Progress:1.7% Speed(reviews/sec):4772. #Correct:14 #Tested:18 Testing Accuracy:77.7%\r",
      "Progress:1.8% Speed(reviews/sec):4939. #Correct:15 #Tested:19 Testing Accuracy:78.9%\r",
      "Progress:1.9% Speed(reviews/sec):4969. #Correct:16 #Tested:20 Testing Accuracy:80.0%\r",
      "Progress:2.0% Speed(reviews/sec):4900. #Correct:17 #Tested:21 Testing Accuracy:80.9%\r",
      "Progress:2.1% Speed(reviews/sec):4709. #Correct:18 #Tested:22 Testing Accuracy:81.8%\r",
      "Progress:2.2% Speed(reviews/sec):4730. #Correct:19 #Tested:23 Testing Accuracy:82.6%\r",
      "Progress:2.3% Speed(reviews/sec):4733. #Correct:20 #Tested:24 Testing Accuracy:83.3%\r",
      "Progress:2.4% Speed(reviews/sec):4602. #Correct:21 #Tested:25 Testing Accuracy:84.0%\r",
      "Progress:2.5% Speed(reviews/sec):4683. #Correct:22 #Tested:26 Testing Accuracy:84.6%\r",
      "Progress:2.6% Speed(reviews/sec):4774. #Correct:22 #Tested:27 Testing Accuracy:81.4%\r",
      "Progress:2.7% Speed(reviews/sec):4523. #Correct:23 #Tested:28 Testing Accuracy:82.1%\r",
      "Progress:2.8% Speed(reviews/sec):4438. #Correct:24 #Tested:29 Testing Accuracy:82.7%\r",
      "Progress:2.9% Speed(reviews/sec):4427. #Correct:25 #Tested:30 Testing Accuracy:83.3%\r",
      "Progress:3.0% Speed(reviews/sec):4384. #Correct:25 #Tested:31 Testing Accuracy:80.6%\r",
      "Progress:3.1% Speed(reviews/sec):4431. #Correct:26 #Tested:32 Testing Accuracy:81.2%\r",
      "Progress:3.2% Speed(reviews/sec):4391. #Correct:27 #Tested:33 Testing Accuracy:81.8%\r",
      "Progress:3.3% Speed(reviews/sec):4425. #Correct:28 #Tested:34 Testing Accuracy:82.3%\r",
      "Progress:3.4% Speed(reviews/sec):4375. #Correct:29 #Tested:35 Testing Accuracy:82.8%\r",
      "Progress:3.5% Speed(reviews/sec):4403. #Correct:30 #Tested:36 Testing Accuracy:83.3%\r",
      "Progress:3.6% Speed(reviews/sec):4366. #Correct:31 #Tested:37 Testing Accuracy:83.7%\r",
      "Progress:3.7% Speed(reviews/sec):4424. #Correct:32 #Tested:38 Testing Accuracy:84.2%\r",
      "Progress:3.8% Speed(reviews/sec):4397. #Correct:33 #Tested:39 Testing Accuracy:84.6%\r",
      "Progress:3.9% Speed(reviews/sec):4356. #Correct:33 #Tested:40 Testing Accuracy:82.5%\r",
      "Progress:4.0% Speed(reviews/sec):4387. #Correct:34 #Tested:41 Testing Accuracy:82.9%\r",
      "Progress:4.1% Speed(reviews/sec):4390. #Correct:35 #Tested:42 Testing Accuracy:83.3%\r",
      "Progress:4.2% Speed(reviews/sec):4278. #Correct:36 #Tested:43 Testing Accuracy:83.7%\r",
      "Progress:4.3% Speed(reviews/sec):4249. #Correct:37 #Tested:44 Testing Accuracy:84.0%\r",
      "Progress:4.4% Speed(reviews/sec):4159. #Correct:38 #Tested:45 Testing Accuracy:84.4%\r",
      "Progress:4.5% Speed(reviews/sec):4038. #Correct:39 #Tested:46 Testing Accuracy:84.7%\r",
      "Progress:4.6% Speed(reviews/sec):4006. #Correct:40 #Tested:47 Testing Accuracy:85.1%\r",
      "Progress:4.7% Speed(reviews/sec):4042. #Correct:41 #Tested:48 Testing Accuracy:85.4%\r",
      "Progress:4.8% Speed(reviews/sec):3994. #Correct:42 #Tested:49 Testing Accuracy:85.7%\r",
      "Progress:4.9% Speed(reviews/sec):4012. #Correct:43 #Tested:50 Testing Accuracy:86.0%\r",
      "Progress:5.0% Speed(reviews/sec):4041. #Correct:44 #Tested:51 Testing Accuracy:86.2%\r",
      "Progress:5.1% Speed(reviews/sec):4084. #Correct:45 #Tested:52 Testing Accuracy:86.5%\r",
      "Progress:5.2% Speed(reviews/sec):4046. #Correct:46 #Tested:53 Testing Accuracy:86.7%\r",
      "Progress:5.3% Speed(reviews/sec):4062. #Correct:47 #Tested:54 Testing Accuracy:87.0%\r",
      "Progress:5.4% Speed(reviews/sec):4088. #Correct:48 #Tested:55 Testing Accuracy:87.2%\r",
      "Progress:5.5% Speed(reviews/sec):4124. #Correct:49 #Tested:56 Testing Accuracy:87.5%\r",
      "Progress:5.6% Speed(reviews/sec):4022. #Correct:50 #Tested:57 Testing Accuracy:87.7%\r",
      "Progress:5.7% Speed(reviews/sec):4056. #Correct:51 #Tested:58 Testing Accuracy:87.9%\r",
      "Progress:5.8% Speed(reviews/sec):4093. #Correct:52 #Tested:59 Testing Accuracy:88.1%\r",
      "Progress:5.9% Speed(reviews/sec):4135. #Correct:53 #Tested:60 Testing Accuracy:88.3%\r",
      "Progress:6.0% Speed(reviews/sec):4113. #Correct:53 #Tested:61 Testing Accuracy:86.8%\r",
      "Progress:6.1% Speed(reviews/sec):4110. #Correct:54 #Tested:62 Testing Accuracy:87.0%\r",
      "Progress:6.2% Speed(reviews/sec):4152. #Correct:55 #Tested:63 Testing Accuracy:87.3%\r",
      "Progress:6.3% Speed(reviews/sec):4184. #Correct:55 #Tested:64 Testing Accuracy:85.9%\r",
      "Progress:6.4% Speed(reviews/sec):4168. #Correct:55 #Tested:65 Testing Accuracy:84.6%\r",
      "Progress:6.5% Speed(reviews/sec):4197. #Correct:55 #Tested:66 Testing Accuracy:83.3%\r",
      "Progress:6.6% Speed(reviews/sec):4228. #Correct:56 #Tested:67 Testing Accuracy:83.5%\r",
      "Progress:6.7% Speed(reviews/sec):4251. #Correct:57 #Tested:68 Testing Accuracy:83.8%\r",
      "Progress:6.8% Speed(reviews/sec):4269. #Correct:58 #Tested:69 Testing Accuracy:84.0%\r",
      "Progress:6.9% Speed(reviews/sec):4268. #Correct:59 #Tested:70 Testing Accuracy:84.2%\r",
      "Progress:7.0% Speed(reviews/sec):4272. #Correct:59 #Tested:71 Testing Accuracy:83.0%\r",
      "Progress:7.1% Speed(reviews/sec):4308. #Correct:60 #Tested:72 Testing Accuracy:83.3%\r",
      "Progress:7.2% Speed(reviews/sec):4334. #Correct:61 #Tested:73 Testing Accuracy:83.5%\r",
      "Progress:7.3% Speed(reviews/sec):4371. #Correct:62 #Tested:74 Testing Accuracy:83.7%\r",
      "Progress:7.4% Speed(reviews/sec):4397. #Correct:63 #Tested:75 Testing Accuracy:84.0%\r",
      "Progress:7.5% Speed(reviews/sec):4400. #Correct:64 #Tested:76 Testing Accuracy:84.2%\r",
      "Progress:7.6% Speed(reviews/sec):4412. #Correct:65 #Tested:77 Testing Accuracy:84.4%\r",
      "Progress:7.7% Speed(reviews/sec):4412. #Correct:66 #Tested:78 Testing Accuracy:84.6%\r",
      "Progress:7.8% Speed(reviews/sec):4197. #Correct:67 #Tested:79 Testing Accuracy:84.8%\r",
      "Progress:7.9% Speed(reviews/sec):3259. #Correct:68 #Tested:80 Testing Accuracy:85.0%\r",
      "Progress:8.0% Speed(reviews/sec):3257. #Correct:69 #Tested:81 Testing Accuracy:85.1%\r",
      "Progress:8.1% Speed(reviews/sec):3172. #Correct:69 #Tested:82 Testing Accuracy:84.1%\r",
      "Progress:8.2% Speed(reviews/sec):3151. #Correct:70 #Tested:83 Testing Accuracy:84.3%\r",
      "Progress:8.3% Speed(reviews/sec):3153. #Correct:71 #Tested:84 Testing Accuracy:84.5%\r",
      "Progress:8.4% Speed(reviews/sec):3148. #Correct:72 #Tested:85 Testing Accuracy:84.7%\r",
      "Progress:8.5% Speed(reviews/sec):3163. #Correct:73 #Tested:86 Testing Accuracy:84.8%\r",
      "Progress:8.6% Speed(reviews/sec):3168. #Correct:74 #Tested:87 Testing Accuracy:85.0%\r",
      "Progress:8.7% Speed(reviews/sec):3187. #Correct:74 #Tested:88 Testing Accuracy:84.0%\r",
      "Progress:8.8% Speed(reviews/sec):3208. #Correct:75 #Tested:89 Testing Accuracy:84.2%\r",
      "Progress:8.9% Speed(reviews/sec):3159. #Correct:76 #Tested:90 Testing Accuracy:84.4%\r",
      "Progress:9.0% Speed(reviews/sec):3139. #Correct:77 #Tested:91 Testing Accuracy:84.6%\r",
      "Progress:9.1% Speed(reviews/sec):3149. #Correct:78 #Tested:92 Testing Accuracy:84.7%\r",
      "Progress:9.2% Speed(reviews/sec):3168. #Correct:79 #Tested:93 Testing Accuracy:84.9%\r",
      "Progress:9.3% Speed(reviews/sec):3183. #Correct:80 #Tested:94 Testing Accuracy:85.1%\r",
      "Progress:9.4% Speed(reviews/sec):3201. #Correct:81 #Tested:95 Testing Accuracy:85.2%\r",
      "Progress:9.5% Speed(reviews/sec):3212. #Correct:82 #Tested:96 Testing Accuracy:85.4%\r",
      "Progress:9.6% Speed(reviews/sec):3223. #Correct:83 #Tested:97 Testing Accuracy:85.5%\r",
      "Progress:9.7% Speed(reviews/sec):3231. #Correct:84 #Tested:98 Testing Accuracy:85.7%\r",
      "Progress:9.8% Speed(reviews/sec):3250. #Correct:85 #Tested:99 Testing Accuracy:85.8%\r",
      "Progress:9.9% Speed(reviews/sec):3250. #Correct:86 #Tested:100 Testing Accuracy:86.0%\r",
      "Progress:10.0% Speed(reviews/sec):3265. #Correct:87 #Tested:101 Testing Accuracy:86.1%\r",
      "Progress:10.1% Speed(reviews/sec):3280. #Correct:88 #Tested:102 Testing Accuracy:86.2%\r",
      "Progress:10.2% Speed(reviews/sec):3295. #Correct:89 #Tested:103 Testing Accuracy:86.4%\r",
      "Progress:10.3% Speed(reviews/sec):3281. #Correct:89 #Tested:104 Testing Accuracy:85.5%\r",
      "Progress:10.4% Speed(reviews/sec):3297. #Correct:90 #Tested:105 Testing Accuracy:85.7%\r",
      "Progress:10.5% Speed(reviews/sec):3318. #Correct:91 #Tested:106 Testing Accuracy:85.8%\r",
      "Progress:10.6% Speed(reviews/sec):3327. #Correct:92 #Tested:107 Testing Accuracy:85.9%\r",
      "Progress:10.7% Speed(reviews/sec):3335. #Correct:93 #Tested:108 Testing Accuracy:86.1%\r",
      "Progress:10.8% Speed(reviews/sec):3311. #Correct:94 #Tested:109 Testing Accuracy:86.2%\r",
      "Progress:10.9% Speed(reviews/sec):3325. #Correct:94 #Tested:110 Testing Accuracy:85.4%\r",
      "Progress:11.0% Speed(reviews/sec):3335. #Correct:95 #Tested:111 Testing Accuracy:85.5%\r",
      "Progress:11.1% Speed(reviews/sec):3347. #Correct:96 #Tested:112 Testing Accuracy:85.7%\r",
      "Progress:11.2% Speed(reviews/sec):3356. #Correct:97 #Tested:113 Testing Accuracy:85.8%\r",
      "Progress:11.3% Speed(reviews/sec):3358. #Correct:98 #Tested:114 Testing Accuracy:85.9%\r",
      "Progress:11.4% Speed(reviews/sec):3360. #Correct:99 #Tested:115 Testing Accuracy:86.0%\r",
      "Progress:11.5% Speed(reviews/sec):3370. #Correct:100 #Tested:116 Testing Accuracy:86.2%\r",
      "Progress:11.6% Speed(reviews/sec):3383. #Correct:101 #Tested:117 Testing Accuracy:86.3%\r",
      "Progress:11.7% Speed(reviews/sec):3389. #Correct:101 #Tested:118 Testing Accuracy:85.5%\r",
      "Progress:11.8% Speed(reviews/sec):3402. #Correct:102 #Tested:119 Testing Accuracy:85.7%\r",
      "Progress:11.9% Speed(reviews/sec):3409. #Correct:103 #Tested:120 Testing Accuracy:85.8%\r",
      "Progress:12.0% Speed(reviews/sec):3420. #Correct:103 #Tested:121 Testing Accuracy:85.1%\r",
      "Progress:12.1% Speed(reviews/sec):3432. #Correct:104 #Tested:122 Testing Accuracy:85.2%\r",
      "Progress:12.2% Speed(reviews/sec):3449. #Correct:105 #Tested:123 Testing Accuracy:85.3%\r",
      "Progress:12.3% Speed(reviews/sec):3449. #Correct:106 #Tested:124 Testing Accuracy:85.4%\r",
      "Progress:12.4% Speed(reviews/sec):3456. #Correct:107 #Tested:125 Testing Accuracy:85.6%\r",
      "Progress:12.5% Speed(reviews/sec):3465. #Correct:108 #Tested:126 Testing Accuracy:85.7%\r",
      "Progress:12.6% Speed(reviews/sec):3262. #Correct:109 #Tested:127 Testing Accuracy:85.8%\r",
      "Progress:12.7% Speed(reviews/sec):3256. #Correct:110 #Tested:128 Testing Accuracy:85.9%\r",
      "Progress:12.8% Speed(reviews/sec):3270. #Correct:111 #Tested:129 Testing Accuracy:86.0%\r",
      "Progress:12.9% Speed(reviews/sec):3283. #Correct:112 #Tested:130 Testing Accuracy:86.1%\r",
      "Progress:13.0% Speed(reviews/sec):3296. #Correct:112 #Tested:131 Testing Accuracy:85.4%\r",
      "Progress:13.1% Speed(reviews/sec):3312. #Correct:113 #Tested:132 Testing Accuracy:85.6%\r",
      "Progress:13.2% Speed(reviews/sec):3324. #Correct:114 #Tested:133 Testing Accuracy:85.7%\r",
      "Progress:13.3% Speed(reviews/sec):3339. #Correct:115 #Tested:134 Testing Accuracy:85.8%\r",
      "Progress:13.4% Speed(reviews/sec):3346. #Correct:116 #Tested:135 Testing Accuracy:85.9%\r",
      "Progress:13.5% Speed(reviews/sec):3354. #Correct:116 #Tested:136 Testing Accuracy:85.2%\r",
      "Progress:13.6% Speed(reviews/sec):3370. #Correct:117 #Tested:137 Testing Accuracy:85.4%\r",
      "Progress:13.7% Speed(reviews/sec):3383. #Correct:118 #Tested:138 Testing Accuracy:85.5%\r",
      "Progress:13.8% Speed(reviews/sec):3396. #Correct:119 #Tested:139 Testing Accuracy:85.6%\r",
      "Progress:13.9% Speed(reviews/sec):3309. #Correct:120 #Tested:140 Testing Accuracy:85.7%\r",
      "Progress:14.0% Speed(reviews/sec):3324. #Correct:121 #Tested:141 Testing Accuracy:85.8%\r",
      "Progress:14.1% Speed(reviews/sec):3333. #Correct:122 #Tested:142 Testing Accuracy:85.9%\r",
      "Progress:14.2% Speed(reviews/sec):3342. #Correct:123 #Tested:143 Testing Accuracy:86.0%\r",
      "Progress:14.3% Speed(reviews/sec):3354. #Correct:124 #Tested:144 Testing Accuracy:86.1%\r",
      "Progress:14.4% Speed(reviews/sec):3333. #Correct:125 #Tested:145 Testing Accuracy:86.2%\r",
      "Progress:14.5% Speed(reviews/sec):3344. #Correct:126 #Tested:146 Testing Accuracy:86.3%\r",
      "Progress:14.6% Speed(reviews/sec):3345. #Correct:127 #Tested:147 Testing Accuracy:86.3%\r",
      "Progress:14.7% Speed(reviews/sec):3356. #Correct:128 #Tested:148 Testing Accuracy:86.4%\r",
      "Progress:14.8% Speed(reviews/sec):3369. #Correct:129 #Tested:149 Testing Accuracy:86.5%\r",
      "Progress:14.9% Speed(reviews/sec):3382. #Correct:130 #Tested:150 Testing Accuracy:86.6%\r",
      "Progress:15.0% Speed(reviews/sec):3394. #Correct:131 #Tested:151 Testing Accuracy:86.7%\r",
      "Progress:15.1% Speed(reviews/sec):3406. #Correct:132 #Tested:152 Testing Accuracy:86.8%\r",
      "Progress:15.2% Speed(reviews/sec):3407. #Correct:133 #Tested:153 Testing Accuracy:86.9%\r",
      "Progress:15.3% Speed(reviews/sec):3315. #Correct:134 #Tested:154 Testing Accuracy:87.0%\r",
      "Progress:15.4% Speed(reviews/sec):3323. #Correct:135 #Tested:155 Testing Accuracy:87.0%\r",
      "Progress:15.5% Speed(reviews/sec):3330. #Correct:136 #Tested:156 Testing Accuracy:87.1%\r",
      "Progress:15.6% Speed(reviews/sec):3329. #Correct:137 #Tested:157 Testing Accuracy:87.2%\r",
      "Progress:15.7% Speed(reviews/sec):3335. #Correct:138 #Tested:158 Testing Accuracy:87.3%\r",
      "Progress:15.8% Speed(reviews/sec):3344. #Correct:139 #Tested:159 Testing Accuracy:87.4%\r",
      "Progress:15.9% Speed(reviews/sec):3352. #Correct:140 #Tested:160 Testing Accuracy:87.5%\r",
      "Progress:16.0% Speed(reviews/sec):3358. #Correct:141 #Tested:161 Testing Accuracy:87.5%\r",
      "Progress:16.1% Speed(reviews/sec):3363. #Correct:141 #Tested:162 Testing Accuracy:87.0%\r",
      "Progress:16.2% Speed(reviews/sec):3375. #Correct:142 #Tested:163 Testing Accuracy:87.1%\r",
      "Progress:16.3% Speed(reviews/sec):3387. #Correct:143 #Tested:164 Testing Accuracy:87.1%\r",
      "Progress:16.4% Speed(reviews/sec):3400. #Correct:144 #Tested:165 Testing Accuracy:87.2%\r",
      "Progress:16.5% Speed(reviews/sec):3403. #Correct:145 #Tested:166 Testing Accuracy:87.3%\r",
      "Progress:16.6% Speed(reviews/sec):3352. #Correct:145 #Tested:167 Testing Accuracy:86.8%\r",
      "Progress:16.7% Speed(reviews/sec):3338. #Correct:146 #Tested:168 Testing Accuracy:86.9%\r",
      "Progress:16.8% Speed(reviews/sec):3347. #Correct:147 #Tested:169 Testing Accuracy:86.9%\r",
      "Progress:16.9% Speed(reviews/sec):3356. #Correct:148 #Tested:170 Testing Accuracy:87.0%\r",
      "Progress:17.0% Speed(reviews/sec):3364. #Correct:149 #Tested:171 Testing Accuracy:87.1%\r",
      "Progress:17.1% Speed(reviews/sec):3366. #Correct:150 #Tested:172 Testing Accuracy:87.2%\r",
      "Progress:17.2% Speed(reviews/sec):3367. #Correct:151 #Tested:173 Testing Accuracy:87.2%\r",
      "Progress:17.3% Speed(reviews/sec):3375. #Correct:151 #Tested:174 Testing Accuracy:86.7%\r",
      "Progress:17.4% Speed(reviews/sec):3382. #Correct:152 #Tested:175 Testing Accuracy:86.8%\r",
      "Progress:17.5% Speed(reviews/sec):3393. #Correct:153 #Tested:176 Testing Accuracy:86.9%\r",
      "Progress:17.6% Speed(reviews/sec):3405. #Correct:154 #Tested:177 Testing Accuracy:87.0%\r",
      "Progress:17.7% Speed(reviews/sec):3408. #Correct:155 #Tested:178 Testing Accuracy:87.0%\r",
      "Progress:17.8% Speed(reviews/sec):3419. #Correct:156 #Tested:179 Testing Accuracy:87.1%\r",
      "Progress:17.9% Speed(reviews/sec):3430. #Correct:157 #Tested:180 Testing Accuracy:87.2%\r",
      "Progress:18.0% Speed(reviews/sec):3439. #Correct:158 #Tested:181 Testing Accuracy:87.2%\r",
      "Progress:18.1% Speed(reviews/sec):3446. #Correct:159 #Tested:182 Testing Accuracy:87.3%\r",
      "Progress:18.2% Speed(reviews/sec):3450. #Correct:159 #Tested:183 Testing Accuracy:86.8%\r",
      "Progress:18.3% Speed(reviews/sec):3457. #Correct:160 #Tested:184 Testing Accuracy:86.9%\r",
      "Progress:18.4% Speed(reviews/sec):3453. #Correct:161 #Tested:185 Testing Accuracy:87.0%\r",
      "Progress:18.5% Speed(reviews/sec):3458. #Correct:162 #Tested:186 Testing Accuracy:87.0%\r",
      "Progress:18.6% Speed(reviews/sec):3468. #Correct:162 #Tested:187 Testing Accuracy:86.6%\r",
      "Progress:18.7% Speed(reviews/sec):3471. #Correct:163 #Tested:188 Testing Accuracy:86.7%\r",
      "Progress:18.8% Speed(reviews/sec):3474. #Correct:164 #Tested:189 Testing Accuracy:86.7%\r",
      "Progress:18.9% Speed(reviews/sec):3482. #Correct:165 #Tested:190 Testing Accuracy:86.8%\r",
      "Progress:19.0% Speed(reviews/sec):3494. #Correct:166 #Tested:191 Testing Accuracy:86.9%\r",
      "Progress:19.1% Speed(reviews/sec):3504. #Correct:167 #Tested:192 Testing Accuracy:86.9%\r",
      "Progress:19.2% Speed(reviews/sec):3432. #Correct:168 #Tested:193 Testing Accuracy:87.0%\r",
      "Progress:19.3% Speed(reviews/sec):3429. #Correct:169 #Tested:194 Testing Accuracy:87.1%\r",
      "Progress:19.4% Speed(reviews/sec):3439. #Correct:170 #Tested:195 Testing Accuracy:87.1%\r",
      "Progress:19.5% Speed(reviews/sec):3438. #Correct:170 #Tested:196 Testing Accuracy:86.7%\r",
      "Progress:19.6% Speed(reviews/sec):3440. #Correct:171 #Tested:197 Testing Accuracy:86.8%\r",
      "Progress:19.7% Speed(reviews/sec):3448. #Correct:172 #Tested:198 Testing Accuracy:86.8%\r",
      "Progress:19.8% Speed(reviews/sec):3459. #Correct:173 #Tested:199 Testing Accuracy:86.9%\r",
      "Progress:19.9% Speed(reviews/sec):3465. #Correct:174 #Tested:200 Testing Accuracy:87.0%\r",
      "Progress:20.0% Speed(reviews/sec):3477. #Correct:175 #Tested:201 Testing Accuracy:87.0%\r",
      "Progress:20.1% Speed(reviews/sec):3476. #Correct:176 #Tested:202 Testing Accuracy:87.1%\r",
      "Progress:20.2% Speed(reviews/sec):3482. #Correct:177 #Tested:203 Testing Accuracy:87.1%\r",
      "Progress:20.3% Speed(reviews/sec):3484. #Correct:178 #Tested:204 Testing Accuracy:87.2%\r",
      "Progress:20.4% Speed(reviews/sec):3493. #Correct:179 #Tested:205 Testing Accuracy:87.3%\r",
      "Progress:20.5% Speed(reviews/sec):3482. #Correct:180 #Tested:206 Testing Accuracy:87.3%\r",
      "Progress:20.6% Speed(reviews/sec):3487. #Correct:181 #Tested:207 Testing Accuracy:87.4%\r",
      "Progress:20.7% Speed(reviews/sec):3493. #Correct:182 #Tested:208 Testing Accuracy:87.5%\r",
      "Progress:20.8% Speed(reviews/sec):3501. #Correct:183 #Tested:209 Testing Accuracy:87.5%\r",
      "Progress:20.9% Speed(reviews/sec):3492. #Correct:184 #Tested:210 Testing Accuracy:87.6%\r",
      "Progress:21.0% Speed(reviews/sec):3499. #Correct:184 #Tested:211 Testing Accuracy:87.2%\r",
      "Progress:21.1% Speed(reviews/sec):3503. #Correct:185 #Tested:212 Testing Accuracy:87.2%\r",
      "Progress:21.2% Speed(reviews/sec):3511. #Correct:186 #Tested:213 Testing Accuracy:87.3%\r",
      "Progress:21.3% Speed(reviews/sec):3498. #Correct:187 #Tested:214 Testing Accuracy:87.3%\r",
      "Progress:21.4% Speed(reviews/sec):3509. #Correct:188 #Tested:215 Testing Accuracy:87.4%\r",
      "Progress:21.5% Speed(reviews/sec):3507. #Correct:189 #Tested:216 Testing Accuracy:87.5%\r",
      "Progress:21.6% Speed(reviews/sec):3516. #Correct:190 #Tested:217 Testing Accuracy:87.5%\r",
      "Progress:21.7% Speed(reviews/sec):3520. #Correct:190 #Tested:218 Testing Accuracy:87.1%\r",
      "Progress:21.8% Speed(reviews/sec):3524. #Correct:191 #Tested:219 Testing Accuracy:87.2%\r",
      "Progress:21.9% Speed(reviews/sec):3536. #Correct:192 #Tested:220 Testing Accuracy:87.2%\r",
      "Progress:22.0% Speed(reviews/sec):3536. #Correct:193 #Tested:221 Testing Accuracy:87.3%\r",
      "Progress:22.1% Speed(reviews/sec):3543. #Correct:194 #Tested:222 Testing Accuracy:87.3%\r",
      "Progress:22.2% Speed(reviews/sec):3430. #Correct:195 #Tested:223 Testing Accuracy:87.4%\r",
      "Progress:22.3% Speed(reviews/sec):3433. #Correct:196 #Tested:224 Testing Accuracy:87.5%\r",
      "Progress:22.4% Speed(reviews/sec):3426. #Correct:197 #Tested:225 Testing Accuracy:87.5%\r",
      "Progress:22.5% Speed(reviews/sec):3430. #Correct:198 #Tested:226 Testing Accuracy:87.6%\r",
      "Progress:22.6% Speed(reviews/sec):3420. #Correct:199 #Tested:227 Testing Accuracy:87.6%\r",
      "Progress:22.7% Speed(reviews/sec):3427. #Correct:200 #Tested:228 Testing Accuracy:87.7%\r",
      "Progress:22.8% Speed(reviews/sec):3433. #Correct:201 #Tested:229 Testing Accuracy:87.7%\r",
      "Progress:22.9% Speed(reviews/sec):3438. #Correct:202 #Tested:230 Testing Accuracy:87.8%\r",
      "Progress:23.0% Speed(reviews/sec):3439. #Correct:203 #Tested:231 Testing Accuracy:87.8%\r",
      "Progress:23.1% Speed(reviews/sec):3396. #Correct:204 #Tested:232 Testing Accuracy:87.9%\r",
      "Progress:23.2% Speed(reviews/sec):3405. #Correct:205 #Tested:233 Testing Accuracy:87.9%\r",
      "Progress:23.3% Speed(reviews/sec):3412. #Correct:206 #Tested:234 Testing Accuracy:88.0%\r",
      "Progress:23.4% Speed(reviews/sec):3406. #Correct:207 #Tested:235 Testing Accuracy:88.0%\r",
      "Progress:23.5% Speed(reviews/sec):3414. #Correct:208 #Tested:236 Testing Accuracy:88.1%\r",
      "Progress:23.6% Speed(reviews/sec):3393. #Correct:208 #Tested:237 Testing Accuracy:87.7%\r",
      "Progress:23.7% Speed(reviews/sec):3398. #Correct:209 #Tested:238 Testing Accuracy:87.8%\r",
      "Progress:23.8% Speed(reviews/sec):3406. #Correct:210 #Tested:239 Testing Accuracy:87.8%\r",
      "Progress:23.9% Speed(reviews/sec):3412. #Correct:211 #Tested:240 Testing Accuracy:87.9%\r",
      "Progress:24.0% Speed(reviews/sec):3414. #Correct:212 #Tested:241 Testing Accuracy:87.9%\r",
      "Progress:24.1% Speed(reviews/sec):3420. #Correct:212 #Tested:242 Testing Accuracy:87.6%\r",
      "Progress:24.2% Speed(reviews/sec):3425. #Correct:213 #Tested:243 Testing Accuracy:87.6%\r",
      "Progress:24.3% Speed(reviews/sec):3425. #Correct:214 #Tested:244 Testing Accuracy:87.7%\r",
      "Progress:24.4% Speed(reviews/sec):3417. #Correct:215 #Tested:245 Testing Accuracy:87.7%\r",
      "Progress:24.5% Speed(reviews/sec):3423. #Correct:216 #Tested:246 Testing Accuracy:87.8%\r",
      "Progress:24.6% Speed(reviews/sec):3431. #Correct:217 #Tested:247 Testing Accuracy:87.8%\r",
      "Progress:24.7% Speed(reviews/sec):3437. #Correct:218 #Tested:248 Testing Accuracy:87.9%\r",
      "Progress:24.8% Speed(reviews/sec):3447. #Correct:219 #Tested:249 Testing Accuracy:87.9%\r",
      "Progress:24.9% Speed(reviews/sec):3455. #Correct:220 #Tested:250 Testing Accuracy:88.0%\r",
      "Progress:25.0% Speed(reviews/sec):3462. #Correct:221 #Tested:251 Testing Accuracy:88.0%\r",
      "Progress:25.1% Speed(reviews/sec):3471. #Correct:222 #Tested:252 Testing Accuracy:88.0%\r",
      "Progress:25.2% Speed(reviews/sec):3479. #Correct:223 #Tested:253 Testing Accuracy:88.1%\r",
      "Progress:25.3% Speed(reviews/sec):3485. #Correct:224 #Tested:254 Testing Accuracy:88.1%\r",
      "Progress:25.4% Speed(reviews/sec):3493. #Correct:225 #Tested:255 Testing Accuracy:88.2%\r",
      "Progress:25.5% Speed(reviews/sec):3500. #Correct:226 #Tested:256 Testing Accuracy:88.2%\r",
      "Progress:25.6% Speed(reviews/sec):3453. #Correct:227 #Tested:257 Testing Accuracy:88.3%\r",
      "Progress:25.7% Speed(reviews/sec):3459. #Correct:228 #Tested:258 Testing Accuracy:88.3%\r",
      "Progress:25.8% Speed(reviews/sec):3468. #Correct:229 #Tested:259 Testing Accuracy:88.4%\r",
      "Progress:25.9% Speed(reviews/sec):3476. #Correct:229 #Tested:260 Testing Accuracy:88.0%\r",
      "Progress:26.0% Speed(reviews/sec):3484. #Correct:230 #Tested:261 Testing Accuracy:88.1%\r",
      "Progress:26.1% Speed(reviews/sec):3493. #Correct:231 #Tested:262 Testing Accuracy:88.1%\r",
      "Progress:26.2% Speed(reviews/sec):3492. #Correct:232 #Tested:263 Testing Accuracy:88.2%\r",
      "Progress:26.3% Speed(reviews/sec):3492. #Correct:233 #Tested:264 Testing Accuracy:88.2%\r",
      "Progress:26.4% Speed(reviews/sec):3494. #Correct:234 #Tested:265 Testing Accuracy:88.3%\r",
      "Progress:26.5% Speed(reviews/sec):3486. #Correct:234 #Tested:266 Testing Accuracy:87.9%\r",
      "Progress:26.6% Speed(reviews/sec):3490. #Correct:235 #Tested:267 Testing Accuracy:88.0%\r",
      "Progress:26.7% Speed(reviews/sec):3494. #Correct:235 #Tested:268 Testing Accuracy:87.6%\r",
      "Progress:26.8% Speed(reviews/sec):3498. #Correct:236 #Tested:269 Testing Accuracy:87.7%\r",
      "Progress:26.9% Speed(reviews/sec):3467. #Correct:236 #Tested:270 Testing Accuracy:87.4%\r",
      "Progress:27.0% Speed(reviews/sec):3468. #Correct:236 #Tested:271 Testing Accuracy:87.0%\r",
      "Progress:27.1% Speed(reviews/sec):3468. #Correct:237 #Tested:272 Testing Accuracy:87.1%\r",
      "Progress:27.2% Speed(reviews/sec):3464. #Correct:238 #Tested:273 Testing Accuracy:87.1%\r",
      "Progress:27.3% Speed(reviews/sec):3470. #Correct:239 #Tested:274 Testing Accuracy:87.2%\r",
      "Progress:27.4% Speed(reviews/sec):3472. #Correct:240 #Tested:275 Testing Accuracy:87.2%\r",
      "Progress:27.5% Speed(reviews/sec):3462. #Correct:241 #Tested:276 Testing Accuracy:87.3%\r",
      "Progress:27.6% Speed(reviews/sec):3466. #Correct:242 #Tested:277 Testing Accuracy:87.3%\r",
      "Progress:27.7% Speed(reviews/sec):3473. #Correct:243 #Tested:278 Testing Accuracy:87.4%\r",
      "Progress:27.8% Speed(reviews/sec):3480. #Correct:244 #Tested:279 Testing Accuracy:87.4%\r",
      "Progress:27.9% Speed(reviews/sec):3487. #Correct:245 #Tested:280 Testing Accuracy:87.5%\r",
      "Progress:28.0% Speed(reviews/sec):3493. #Correct:246 #Tested:281 Testing Accuracy:87.5%\r",
      "Progress:28.1% Speed(reviews/sec):3501. #Correct:247 #Tested:282 Testing Accuracy:87.5%\r",
      "Progress:28.2% Speed(reviews/sec):3510. #Correct:248 #Tested:283 Testing Accuracy:87.6%\r",
      "Progress:28.3% Speed(reviews/sec):3518. #Correct:249 #Tested:284 Testing Accuracy:87.6%\r",
      "Progress:28.4% Speed(reviews/sec):3499. #Correct:250 #Tested:285 Testing Accuracy:87.7%\r",
      "Progress:28.5% Speed(reviews/sec):3503. #Correct:251 #Tested:286 Testing Accuracy:87.7%\r",
      "Progress:28.6% Speed(reviews/sec):3507. #Correct:252 #Tested:287 Testing Accuracy:87.8%\r",
      "Progress:28.7% Speed(reviews/sec):3513. #Correct:253 #Tested:288 Testing Accuracy:87.8%\r",
      "Progress:28.8% Speed(reviews/sec):3517. #Correct:254 #Tested:289 Testing Accuracy:87.8%\r",
      "Progress:28.9% Speed(reviews/sec):3523. #Correct:255 #Tested:290 Testing Accuracy:87.9%\r",
      "Progress:29.0% Speed(reviews/sec):3528. #Correct:256 #Tested:291 Testing Accuracy:87.9%\r",
      "Progress:29.1% Speed(reviews/sec):3533. #Correct:257 #Tested:292 Testing Accuracy:88.0%\r",
      "Progress:29.2% Speed(reviews/sec):3539. #Correct:258 #Tested:293 Testing Accuracy:88.0%\r",
      "Progress:29.3% Speed(reviews/sec):3536. #Correct:259 #Tested:294 Testing Accuracy:88.0%\r",
      "Progress:29.4% Speed(reviews/sec):3542. #Correct:260 #Tested:295 Testing Accuracy:88.1%\r",
      "Progress:29.5% Speed(reviews/sec):3548. #Correct:261 #Tested:296 Testing Accuracy:88.1%\r",
      "Progress:29.6% Speed(reviews/sec):3555. #Correct:262 #Tested:297 Testing Accuracy:88.2%\r",
      "Progress:29.7% Speed(reviews/sec):3555. #Correct:263 #Tested:298 Testing Accuracy:88.2%\r",
      "Progress:29.8% Speed(reviews/sec):3558. #Correct:264 #Tested:299 Testing Accuracy:88.2%\r",
      "Progress:29.9% Speed(reviews/sec):3561. #Correct:265 #Tested:300 Testing Accuracy:88.3%\r",
      "Progress:30.0% Speed(reviews/sec):3564. #Correct:266 #Tested:301 Testing Accuracy:88.3%\r",
      "Progress:30.1% Speed(reviews/sec):3559. #Correct:266 #Tested:302 Testing Accuracy:88.0%\r",
      "Progress:30.2% Speed(reviews/sec):3505. #Correct:267 #Tested:303 Testing Accuracy:88.1%\r",
      "Progress:30.3% Speed(reviews/sec):3508. #Correct:268 #Tested:304 Testing Accuracy:88.1%\r",
      "Progress:30.4% Speed(reviews/sec):3515. #Correct:269 #Tested:305 Testing Accuracy:88.1%\r",
      "Progress:30.5% Speed(reviews/sec):3520. #Correct:269 #Tested:306 Testing Accuracy:87.9%\r",
      "Progress:30.6% Speed(reviews/sec):3503. #Correct:270 #Tested:307 Testing Accuracy:87.9%\r",
      "Progress:30.7% Speed(reviews/sec):3507. #Correct:270 #Tested:308 Testing Accuracy:87.6%\r",
      "Progress:30.8% Speed(reviews/sec):3508. #Correct:271 #Tested:309 Testing Accuracy:87.7%\r",
      "Progress:30.9% Speed(reviews/sec):3504. #Correct:272 #Tested:310 Testing Accuracy:87.7%\r",
      "Progress:31.0% Speed(reviews/sec):3510. #Correct:273 #Tested:311 Testing Accuracy:87.7%\r",
      "Progress:31.1% Speed(reviews/sec):3513. #Correct:274 #Tested:312 Testing Accuracy:87.8%\r",
      "Progress:31.2% Speed(reviews/sec):3505. #Correct:275 #Tested:313 Testing Accuracy:87.8%\r",
      "Progress:31.3% Speed(reviews/sec):3507. #Correct:276 #Tested:314 Testing Accuracy:87.8%\r",
      "Progress:31.4% Speed(reviews/sec):3514. #Correct:277 #Tested:315 Testing Accuracy:87.9%\r",
      "Progress:31.5% Speed(reviews/sec):3517. #Correct:278 #Tested:316 Testing Accuracy:87.9%\r",
      "Progress:31.6% Speed(reviews/sec):3524. #Correct:279 #Tested:317 Testing Accuracy:88.0%\r",
      "Progress:31.7% Speed(reviews/sec):3527. #Correct:279 #Tested:318 Testing Accuracy:87.7%\r",
      "Progress:31.8% Speed(reviews/sec):3532. #Correct:280 #Tested:319 Testing Accuracy:87.7%\r",
      "Progress:31.9% Speed(reviews/sec):3538. #Correct:281 #Tested:320 Testing Accuracy:87.8%\r",
      "Progress:32.0% Speed(reviews/sec):3544. #Correct:282 #Tested:321 Testing Accuracy:87.8%\r",
      "Progress:32.1% Speed(reviews/sec):3551. #Correct:282 #Tested:322 Testing Accuracy:87.5%\r",
      "Progress:32.2% Speed(reviews/sec):3558. #Correct:283 #Tested:323 Testing Accuracy:87.6%\r",
      "Progress:32.3% Speed(reviews/sec):3474. #Correct:284 #Tested:324 Testing Accuracy:87.6%\r",
      "Progress:32.4% Speed(reviews/sec):3472. #Correct:285 #Tested:325 Testing Accuracy:87.6%\r",
      "Progress:32.5% Speed(reviews/sec):3477. #Correct:286 #Tested:326 Testing Accuracy:87.7%\r",
      "Progress:32.6% Speed(reviews/sec):3482. #Correct:287 #Tested:327 Testing Accuracy:87.7%\r",
      "Progress:32.7% Speed(reviews/sec):3483. #Correct:287 #Tested:328 Testing Accuracy:87.5%\r",
      "Progress:32.8% Speed(reviews/sec):3485. #Correct:288 #Tested:329 Testing Accuracy:87.5%\r",
      "Progress:32.9% Speed(reviews/sec):3491. #Correct:289 #Tested:330 Testing Accuracy:87.5%\r",
      "Progress:33.0% Speed(reviews/sec):3496. #Correct:290 #Tested:331 Testing Accuracy:87.6%\r",
      "Progress:33.1% Speed(reviews/sec):3502. #Correct:291 #Tested:332 Testing Accuracy:87.6%\r",
      "Progress:33.2% Speed(reviews/sec):3504. #Correct:292 #Tested:333 Testing Accuracy:87.6%\r",
      "Progress:33.3% Speed(reviews/sec):3509. #Correct:293 #Tested:334 Testing Accuracy:87.7%\r",
      "Progress:33.4% Speed(reviews/sec):3476. #Correct:294 #Tested:335 Testing Accuracy:87.7%\r",
      "Progress:33.5% Speed(reviews/sec):3480. #Correct:295 #Tested:336 Testing Accuracy:87.7%\r",
      "Progress:33.6% Speed(reviews/sec):3484. #Correct:296 #Tested:337 Testing Accuracy:87.8%\r",
      "Progress:33.7% Speed(reviews/sec):3491. #Correct:297 #Tested:338 Testing Accuracy:87.8%\r",
      "Progress:33.8% Speed(reviews/sec):3496. #Correct:297 #Tested:339 Testing Accuracy:87.6%\r",
      "Progress:33.9% Speed(reviews/sec):3501. #Correct:298 #Tested:340 Testing Accuracy:87.6%\r",
      "Progress:34.0% Speed(reviews/sec):3499. #Correct:298 #Tested:341 Testing Accuracy:87.3%\r",
      "Progress:34.1% Speed(reviews/sec):3504. #Correct:299 #Tested:342 Testing Accuracy:87.4%\r",
      "Progress:34.2% Speed(reviews/sec):3496. #Correct:300 #Tested:343 Testing Accuracy:87.4%\r",
      "Progress:34.3% Speed(reviews/sec):3501. #Correct:301 #Tested:344 Testing Accuracy:87.5%\r",
      "Progress:34.4% Speed(reviews/sec):3501. #Correct:302 #Tested:345 Testing Accuracy:87.5%\r",
      "Progress:34.5% Speed(reviews/sec):3507. #Correct:303 #Tested:346 Testing Accuracy:87.5%\r",
      "Progress:34.6% Speed(reviews/sec):3513. #Correct:303 #Tested:347 Testing Accuracy:87.3%\r",
      "Progress:34.7% Speed(reviews/sec):3517. #Correct:304 #Tested:348 Testing Accuracy:87.3%\r",
      "Progress:34.8% Speed(reviews/sec):3521. #Correct:305 #Tested:349 Testing Accuracy:87.3%\r",
      "Progress:34.9% Speed(reviews/sec):3521. #Correct:306 #Tested:350 Testing Accuracy:87.4%\r",
      "Progress:35.0% Speed(reviews/sec):3527. #Correct:307 #Tested:351 Testing Accuracy:87.4%\r",
      "Progress:35.1% Speed(reviews/sec):3532. #Correct:308 #Tested:352 Testing Accuracy:87.5%\r",
      "Progress:35.2% Speed(reviews/sec):3534. #Correct:309 #Tested:353 Testing Accuracy:87.5%\r",
      "Progress:35.3% Speed(reviews/sec):3541. #Correct:309 #Tested:354 Testing Accuracy:87.2%\r",
      "Progress:35.4% Speed(reviews/sec):3547. #Correct:310 #Tested:355 Testing Accuracy:87.3%\r",
      "Progress:35.5% Speed(reviews/sec):3545. #Correct:311 #Tested:356 Testing Accuracy:87.3%\r",
      "Progress:35.6% Speed(reviews/sec):3549. #Correct:311 #Tested:357 Testing Accuracy:87.1%\r",
      "Progress:35.7% Speed(reviews/sec):3547. #Correct:311 #Tested:358 Testing Accuracy:86.8%\r",
      "Progress:35.8% Speed(reviews/sec):3552. #Correct:312 #Tested:359 Testing Accuracy:86.9%\r",
      "Progress:35.9% Speed(reviews/sec):3505. #Correct:313 #Tested:360 Testing Accuracy:86.9%\r",
      "Progress:36.0% Speed(reviews/sec):3506. #Correct:314 #Tested:361 Testing Accuracy:86.9%\r",
      "Progress:36.1% Speed(reviews/sec):3503. #Correct:315 #Tested:362 Testing Accuracy:87.0%\r",
      "Progress:36.2% Speed(reviews/sec):3497. #Correct:316 #Tested:363 Testing Accuracy:87.0%\r",
      "Progress:36.3% Speed(reviews/sec):3501. #Correct:317 #Tested:364 Testing Accuracy:87.0%\r",
      "Progress:36.4% Speed(reviews/sec):3505. #Correct:317 #Tested:365 Testing Accuracy:86.8%\r",
      "Progress:36.5% Speed(reviews/sec):3508. #Correct:318 #Tested:366 Testing Accuracy:86.8%\r",
      "Progress:36.6% Speed(reviews/sec):3512. #Correct:319 #Tested:367 Testing Accuracy:86.9%\r",
      "Progress:36.7% Speed(reviews/sec):3518. #Correct:320 #Tested:368 Testing Accuracy:86.9%\r",
      "Progress:36.8% Speed(reviews/sec):3521. #Correct:320 #Tested:369 Testing Accuracy:86.7%\r",
      "Progress:36.9% Speed(reviews/sec):3526. #Correct:320 #Tested:370 Testing Accuracy:86.4%\r",
      "Progress:37.0% Speed(reviews/sec):3525. #Correct:321 #Tested:371 Testing Accuracy:86.5%\r",
      "Progress:37.1% Speed(reviews/sec):3523. #Correct:322 #Tested:372 Testing Accuracy:86.5%\r",
      "Progress:37.2% Speed(reviews/sec):3528. #Correct:322 #Tested:373 Testing Accuracy:86.3%\r",
      "Progress:37.3% Speed(reviews/sec):3533. #Correct:322 #Tested:374 Testing Accuracy:86.0%\r",
      "Progress:37.4% Speed(reviews/sec):3533. #Correct:323 #Tested:375 Testing Accuracy:86.1%\r",
      "Progress:37.5% Speed(reviews/sec):3537. #Correct:324 #Tested:376 Testing Accuracy:86.1%\r",
      "Progress:37.6% Speed(reviews/sec):3542. #Correct:325 #Tested:377 Testing Accuracy:86.2%\r",
      "Progress:37.7% Speed(reviews/sec):3546. #Correct:326 #Tested:378 Testing Accuracy:86.2%\r",
      "Progress:37.8% Speed(reviews/sec):3551. #Correct:326 #Tested:379 Testing Accuracy:86.0%\r",
      "Progress:37.9% Speed(reviews/sec):3557. #Correct:327 #Tested:380 Testing Accuracy:86.0%\r",
      "Progress:38.0% Speed(reviews/sec):3512. #Correct:328 #Tested:381 Testing Accuracy:86.0%\r",
      "Progress:38.1% Speed(reviews/sec):3517. #Correct:329 #Tested:382 Testing Accuracy:86.1%\r",
      "Progress:38.2% Speed(reviews/sec):3515. #Correct:330 #Tested:383 Testing Accuracy:86.1%\r",
      "Progress:38.3% Speed(reviews/sec):3510. #Correct:331 #Tested:384 Testing Accuracy:86.1%\r",
      "Progress:38.4% Speed(reviews/sec):3513. #Correct:332 #Tested:385 Testing Accuracy:86.2%\r",
      "Progress:38.5% Speed(reviews/sec):3518. #Correct:333 #Tested:386 Testing Accuracy:86.2%\r",
      "Progress:38.6% Speed(reviews/sec):3518. #Correct:334 #Tested:387 Testing Accuracy:86.3%\r",
      "Progress:38.7% Speed(reviews/sec):3524. #Correct:335 #Tested:388 Testing Accuracy:86.3%\r",
      "Progress:38.8% Speed(reviews/sec):3528. #Correct:335 #Tested:389 Testing Accuracy:86.1%\r",
      "Progress:38.9% Speed(reviews/sec):3533. #Correct:336 #Tested:390 Testing Accuracy:86.1%\r",
      "Progress:39.0% Speed(reviews/sec):3538. #Correct:336 #Tested:391 Testing Accuracy:85.9%\r",
      "Progress:39.1% Speed(reviews/sec):3543. #Correct:337 #Tested:392 Testing Accuracy:85.9%\r",
      "Progress:39.2% Speed(reviews/sec):3548. #Correct:337 #Tested:393 Testing Accuracy:85.7%\r",
      "Progress:39.3% Speed(reviews/sec):3550. #Correct:338 #Tested:394 Testing Accuracy:85.7%\r",
      "Progress:39.4% Speed(reviews/sec):3516. #Correct:338 #Tested:395 Testing Accuracy:85.5%\r",
      "Progress:39.5% Speed(reviews/sec):3518. #Correct:339 #Tested:396 Testing Accuracy:85.6%\r",
      "Progress:39.6% Speed(reviews/sec):3521. #Correct:340 #Tested:397 Testing Accuracy:85.6%\r",
      "Progress:39.7% Speed(reviews/sec):3525. #Correct:341 #Tested:398 Testing Accuracy:85.6%\r",
      "Progress:39.8% Speed(reviews/sec):3524. #Correct:341 #Tested:399 Testing Accuracy:85.4%\r",
      "Progress:39.9% Speed(reviews/sec):3527. #Correct:342 #Tested:400 Testing Accuracy:85.5%\r",
      "Progress:40.0% Speed(reviews/sec):3526. #Correct:343 #Tested:401 Testing Accuracy:85.5%\r",
      "Progress:40.1% Speed(reviews/sec):3526. #Correct:344 #Tested:402 Testing Accuracy:85.5%\r",
      "Progress:40.2% Speed(reviews/sec):3522. #Correct:345 #Tested:403 Testing Accuracy:85.6%\r",
      "Progress:40.3% Speed(reviews/sec):3525. #Correct:345 #Tested:404 Testing Accuracy:85.3%\r",
      "Progress:40.4% Speed(reviews/sec):3529. #Correct:346 #Tested:405 Testing Accuracy:85.4%\r",
      "Progress:40.5% Speed(reviews/sec):3533. #Correct:347 #Tested:406 Testing Accuracy:85.4%\r",
      "Progress:40.6% Speed(reviews/sec):3529. #Correct:348 #Tested:407 Testing Accuracy:85.5%\r",
      "Progress:40.7% Speed(reviews/sec):3533. #Correct:349 #Tested:408 Testing Accuracy:85.5%\r",
      "Progress:40.8% Speed(reviews/sec):3532. #Correct:350 #Tested:409 Testing Accuracy:85.5%\r",
      "Progress:40.9% Speed(reviews/sec):3535. #Correct:351 #Tested:410 Testing Accuracy:85.6%\r",
      "Progress:41.0% Speed(reviews/sec):3536. #Correct:352 #Tested:411 Testing Accuracy:85.6%\r",
      "Progress:41.1% Speed(reviews/sec):3530. #Correct:353 #Tested:412 Testing Accuracy:85.6%\r",
      "Progress:41.2% Speed(reviews/sec):3479. #Correct:354 #Tested:413 Testing Accuracy:85.7%\r",
      "Progress:41.3% Speed(reviews/sec):3470. #Correct:355 #Tested:414 Testing Accuracy:85.7%\r",
      "Progress:41.4% Speed(reviews/sec):3469. #Correct:356 #Tested:415 Testing Accuracy:85.7%\r",
      "Progress:41.5% Speed(reviews/sec):3470. #Correct:357 #Tested:416 Testing Accuracy:85.8%\r",
      "Progress:41.6% Speed(reviews/sec):3470. #Correct:358 #Tested:417 Testing Accuracy:85.8%\r",
      "Progress:41.7% Speed(reviews/sec):3471. #Correct:359 #Tested:418 Testing Accuracy:85.8%\r",
      "Progress:41.8% Speed(reviews/sec):3473. #Correct:360 #Tested:419 Testing Accuracy:85.9%\r",
      "Progress:41.9% Speed(reviews/sec):3478. #Correct:360 #Tested:420 Testing Accuracy:85.7%\r",
      "Progress:42.0% Speed(reviews/sec):3474. #Correct:361 #Tested:421 Testing Accuracy:85.7%\r",
      "Progress:42.1% Speed(reviews/sec):3476. #Correct:362 #Tested:422 Testing Accuracy:85.7%\r",
      "Progress:42.2% Speed(reviews/sec):3476. #Correct:363 #Tested:423 Testing Accuracy:85.8%\r",
      "Progress:42.3% Speed(reviews/sec):3478. #Correct:364 #Tested:424 Testing Accuracy:85.8%\r",
      "Progress:42.4% Speed(reviews/sec):3476. #Correct:365 #Tested:425 Testing Accuracy:85.8%\r",
      "Progress:42.5% Speed(reviews/sec):3474. #Correct:366 #Tested:426 Testing Accuracy:85.9%\r",
      "Progress:42.6% Speed(reviews/sec):3469. #Correct:367 #Tested:427 Testing Accuracy:85.9%\r",
      "Progress:42.7% Speed(reviews/sec):3471. #Correct:368 #Tested:428 Testing Accuracy:85.9%\r",
      "Progress:42.8% Speed(reviews/sec):3474. #Correct:369 #Tested:429 Testing Accuracy:86.0%\r",
      "Progress:42.9% Speed(reviews/sec):3421. #Correct:370 #Tested:430 Testing Accuracy:86.0%\r",
      "Progress:43.0% Speed(reviews/sec):3411. #Correct:371 #Tested:431 Testing Accuracy:86.0%\r",
      "Progress:43.1% Speed(reviews/sec):3410. #Correct:372 #Tested:432 Testing Accuracy:86.1%\r",
      "Progress:43.2% Speed(reviews/sec):3408. #Correct:372 #Tested:433 Testing Accuracy:85.9%\r",
      "Progress:43.3% Speed(reviews/sec):3411. #Correct:373 #Tested:434 Testing Accuracy:85.9%\r",
      "Progress:43.4% Speed(reviews/sec):3413. #Correct:374 #Tested:435 Testing Accuracy:85.9%\r",
      "Progress:43.5% Speed(reviews/sec):3407. #Correct:375 #Tested:436 Testing Accuracy:86.0%\r",
      "Progress:43.6% Speed(reviews/sec):3409. #Correct:376 #Tested:437 Testing Accuracy:86.0%\r",
      "Progress:43.7% Speed(reviews/sec):3406. #Correct:377 #Tested:438 Testing Accuracy:86.0%\r",
      "Progress:43.8% Speed(reviews/sec):3409. #Correct:378 #Tested:439 Testing Accuracy:86.1%\r",
      "Progress:43.9% Speed(reviews/sec):3409. #Correct:379 #Tested:440 Testing Accuracy:86.1%\r",
      "Progress:44.0% Speed(reviews/sec):3407. #Correct:380 #Tested:441 Testing Accuracy:86.1%\r",
      "Progress:44.1% Speed(reviews/sec):3405. #Correct:381 #Tested:442 Testing Accuracy:86.1%\r",
      "Progress:44.2% Speed(reviews/sec):3408. #Correct:382 #Tested:443 Testing Accuracy:86.2%\r",
      "Progress:44.3% Speed(reviews/sec):3375. #Correct:383 #Tested:444 Testing Accuracy:86.2%\r",
      "Progress:44.4% Speed(reviews/sec):3372. #Correct:384 #Tested:445 Testing Accuracy:86.2%\r",
      "Progress:44.5% Speed(reviews/sec):3375. #Correct:385 #Tested:446 Testing Accuracy:86.3%\r",
      "Progress:44.6% Speed(reviews/sec):3379. #Correct:386 #Tested:447 Testing Accuracy:86.3%\r",
      "Progress:44.7% Speed(reviews/sec):3380. #Correct:387 #Tested:448 Testing Accuracy:86.3%\r",
      "Progress:44.8% Speed(reviews/sec):3384. #Correct:388 #Tested:449 Testing Accuracy:86.4%\r",
      "Progress:44.9% Speed(reviews/sec):3379. #Correct:388 #Tested:450 Testing Accuracy:86.2%\r",
      "Progress:45.0% Speed(reviews/sec):3366. #Correct:389 #Tested:451 Testing Accuracy:86.2%\r",
      "Progress:45.1% Speed(reviews/sec):3361. #Correct:389 #Tested:452 Testing Accuracy:86.0%\r",
      "Progress:45.2% Speed(reviews/sec):3364. #Correct:390 #Tested:453 Testing Accuracy:86.0%\r",
      "Progress:45.3% Speed(reviews/sec):3368. #Correct:391 #Tested:454 Testing Accuracy:86.1%\r",
      "Progress:45.4% Speed(reviews/sec):3373. #Correct:392 #Tested:455 Testing Accuracy:86.1%\r",
      "Progress:45.5% Speed(reviews/sec):3377. #Correct:393 #Tested:456 Testing Accuracy:86.1%\r",
      "Progress:45.6% Speed(reviews/sec):3382. #Correct:394 #Tested:457 Testing Accuracy:86.2%\r",
      "Progress:45.7% Speed(reviews/sec):3386. #Correct:395 #Tested:458 Testing Accuracy:86.2%\r",
      "Progress:45.8% Speed(reviews/sec):3390. #Correct:396 #Tested:459 Testing Accuracy:86.2%\r",
      "Progress:45.9% Speed(reviews/sec):3389. #Correct:397 #Tested:460 Testing Accuracy:86.3%\r",
      "Progress:46.0% Speed(reviews/sec):3392. #Correct:398 #Tested:461 Testing Accuracy:86.3%\r",
      "Progress:46.1% Speed(reviews/sec):3392. #Correct:398 #Tested:462 Testing Accuracy:86.1%\r",
      "Progress:46.2% Speed(reviews/sec):3387. #Correct:399 #Tested:463 Testing Accuracy:86.1%\r",
      "Progress:46.3% Speed(reviews/sec):3390. #Correct:400 #Tested:464 Testing Accuracy:86.2%\r",
      "Progress:46.4% Speed(reviews/sec):3394. #Correct:401 #Tested:465 Testing Accuracy:86.2%\r",
      "Progress:46.5% Speed(reviews/sec):3398. #Correct:402 #Tested:466 Testing Accuracy:86.2%\r",
      "Progress:46.6% Speed(reviews/sec):3402. #Correct:403 #Tested:467 Testing Accuracy:86.2%\r",
      "Progress:46.7% Speed(reviews/sec):3401. #Correct:404 #Tested:468 Testing Accuracy:86.3%\r",
      "Progress:46.8% Speed(reviews/sec):3402. #Correct:405 #Tested:469 Testing Accuracy:86.3%\r",
      "Progress:46.9% Speed(reviews/sec):3374. #Correct:405 #Tested:470 Testing Accuracy:86.1%\r",
      "Progress:47.0% Speed(reviews/sec):3377. #Correct:406 #Tested:471 Testing Accuracy:86.1%\r",
      "Progress:47.1% Speed(reviews/sec):3380. #Correct:406 #Tested:472 Testing Accuracy:86.0%\r",
      "Progress:47.2% Speed(reviews/sec):3383. #Correct:407 #Tested:473 Testing Accuracy:86.0%\r",
      "Progress:47.3% Speed(reviews/sec):3384. #Correct:408 #Tested:474 Testing Accuracy:86.0%\r",
      "Progress:47.4% Speed(reviews/sec):3387. #Correct:409 #Tested:475 Testing Accuracy:86.1%\r",
      "Progress:47.5% Speed(reviews/sec):3388. #Correct:410 #Tested:476 Testing Accuracy:86.1%\r",
      "Progress:47.6% Speed(reviews/sec):3384. #Correct:411 #Tested:477 Testing Accuracy:86.1%\r",
      "Progress:47.7% Speed(reviews/sec):3384. #Correct:411 #Tested:478 Testing Accuracy:85.9%\r",
      "Progress:47.8% Speed(reviews/sec):3381. #Correct:412 #Tested:479 Testing Accuracy:86.0%\r",
      "Progress:47.9% Speed(reviews/sec):3385. #Correct:413 #Tested:480 Testing Accuracy:86.0%\r",
      "Progress:48.0% Speed(reviews/sec):3365. #Correct:414 #Tested:481 Testing Accuracy:86.0%\r",
      "Progress:48.1% Speed(reviews/sec):3351. #Correct:415 #Tested:482 Testing Accuracy:86.0%\r",
      "Progress:48.2% Speed(reviews/sec):3354. #Correct:416 #Tested:483 Testing Accuracy:86.1%\r",
      "Progress:48.3% Speed(reviews/sec):3357. #Correct:417 #Tested:484 Testing Accuracy:86.1%\r",
      "Progress:48.4% Speed(reviews/sec):3360. #Correct:418 #Tested:485 Testing Accuracy:86.1%\r",
      "Progress:48.5% Speed(reviews/sec):3361. #Correct:419 #Tested:486 Testing Accuracy:86.2%\r",
      "Progress:48.6% Speed(reviews/sec):3365. #Correct:420 #Tested:487 Testing Accuracy:86.2%\r",
      "Progress:48.7% Speed(reviews/sec):3369. #Correct:420 #Tested:488 Testing Accuracy:86.0%\r",
      "Progress:48.8% Speed(reviews/sec):3372. #Correct:421 #Tested:489 Testing Accuracy:86.0%\r",
      "Progress:48.9% Speed(reviews/sec):3357. #Correct:421 #Tested:490 Testing Accuracy:85.9%\r",
      "Progress:49.0% Speed(reviews/sec):3351. #Correct:422 #Tested:491 Testing Accuracy:85.9%\r",
      "Progress:49.1% Speed(reviews/sec):3354. #Correct:423 #Tested:492 Testing Accuracy:85.9%\r",
      "Progress:49.2% Speed(reviews/sec):3353. #Correct:424 #Tested:493 Testing Accuracy:86.0%\r",
      "Progress:49.3% Speed(reviews/sec):3352. #Correct:425 #Tested:494 Testing Accuracy:86.0%\r",
      "Progress:49.4% Speed(reviews/sec):3354. #Correct:426 #Tested:495 Testing Accuracy:86.0%\r",
      "Progress:49.5% Speed(reviews/sec):3355. #Correct:427 #Tested:496 Testing Accuracy:86.0%\r",
      "Progress:49.6% Speed(reviews/sec):3357. #Correct:428 #Tested:497 Testing Accuracy:86.1%\r",
      "Progress:49.7% Speed(reviews/sec):3359. #Correct:428 #Tested:498 Testing Accuracy:85.9%\r",
      "Progress:49.8% Speed(reviews/sec):3356. #Correct:429 #Tested:499 Testing Accuracy:85.9%\r",
      "Progress:49.9% Speed(reviews/sec):3354. #Correct:430 #Tested:500 Testing Accuracy:86.0%\r",
      "Progress:50.0% Speed(reviews/sec):3358. #Correct:431 #Tested:501 Testing Accuracy:86.0%\r",
      "Progress:50.1% Speed(reviews/sec):3355. #Correct:432 #Tested:502 Testing Accuracy:86.0%\r",
      "Progress:50.2% Speed(reviews/sec):3353. #Correct:433 #Tested:503 Testing Accuracy:86.0%\r",
      "Progress:50.3% Speed(reviews/sec):3311. #Correct:434 #Tested:504 Testing Accuracy:86.1%\r",
      "Progress:50.4% Speed(reviews/sec):3304. #Correct:434 #Tested:505 Testing Accuracy:85.9%\r",
      "Progress:50.5% Speed(reviews/sec):3302. #Correct:435 #Tested:506 Testing Accuracy:85.9%\r",
      "Progress:50.6% Speed(reviews/sec):3281. #Correct:436 #Tested:507 Testing Accuracy:85.9%\r",
      "Progress:50.7% Speed(reviews/sec):3259. #Correct:437 #Tested:508 Testing Accuracy:86.0%\r",
      "Progress:50.8% Speed(reviews/sec):3261. #Correct:438 #Tested:509 Testing Accuracy:86.0%\r",
      "Progress:50.9% Speed(reviews/sec):3264. #Correct:439 #Tested:510 Testing Accuracy:86.0%\r",
      "Progress:51.0% Speed(reviews/sec):3261. #Correct:440 #Tested:511 Testing Accuracy:86.1%\r",
      "Progress:51.1% Speed(reviews/sec):3258. #Correct:441 #Tested:512 Testing Accuracy:86.1%\r",
      "Progress:51.2% Speed(reviews/sec):3259. #Correct:442 #Tested:513 Testing Accuracy:86.1%\r",
      "Progress:51.3% Speed(reviews/sec):3263. #Correct:443 #Tested:514 Testing Accuracy:86.1%\r",
      "Progress:51.4% Speed(reviews/sec):3264. #Correct:444 #Tested:515 Testing Accuracy:86.2%\r",
      "Progress:51.5% Speed(reviews/sec):3264. #Correct:445 #Tested:516 Testing Accuracy:86.2%\r",
      "Progress:51.6% Speed(reviews/sec):3263. #Correct:446 #Tested:517 Testing Accuracy:86.2%\r",
      "Progress:51.7% Speed(reviews/sec):3266. #Correct:447 #Tested:518 Testing Accuracy:86.2%\r",
      "Progress:51.8% Speed(reviews/sec):3268. #Correct:447 #Tested:519 Testing Accuracy:86.1%\r",
      "Progress:51.9% Speed(reviews/sec):3271. #Correct:448 #Tested:520 Testing Accuracy:86.1%\r",
      "Progress:52.0% Speed(reviews/sec):3274. #Correct:449 #Tested:521 Testing Accuracy:86.1%\r",
      "Progress:52.1% Speed(reviews/sec):3277. #Correct:450 #Tested:522 Testing Accuracy:86.2%\r",
      "Progress:52.2% Speed(reviews/sec):3274. #Correct:451 #Tested:523 Testing Accuracy:86.2%\r",
      "Progress:52.3% Speed(reviews/sec):3275. #Correct:452 #Tested:524 Testing Accuracy:86.2%\r",
      "Progress:52.4% Speed(reviews/sec):3277. #Correct:453 #Tested:525 Testing Accuracy:86.2%\r",
      "Progress:52.5% Speed(reviews/sec):3280. #Correct:454 #Tested:526 Testing Accuracy:86.3%\r",
      "Progress:52.6% Speed(reviews/sec):3283. #Correct:455 #Tested:527 Testing Accuracy:86.3%\r",
      "Progress:52.7% Speed(reviews/sec):3252. #Correct:455 #Tested:528 Testing Accuracy:86.1%\r",
      "Progress:52.8% Speed(reviews/sec):3250. #Correct:455 #Tested:529 Testing Accuracy:86.0%\r",
      "Progress:52.9% Speed(reviews/sec):3252. #Correct:456 #Tested:530 Testing Accuracy:86.0%\r",
      "Progress:53.0% Speed(reviews/sec):3253. #Correct:457 #Tested:531 Testing Accuracy:86.0%\r",
      "Progress:53.1% Speed(reviews/sec):3255. #Correct:457 #Tested:532 Testing Accuracy:85.9%\r",
      "Progress:53.2% Speed(reviews/sec):3258. #Correct:458 #Tested:533 Testing Accuracy:85.9%\r",
      "Progress:53.3% Speed(reviews/sec):3260. #Correct:459 #Tested:534 Testing Accuracy:85.9%\r",
      "Progress:53.4% Speed(reviews/sec):3264. #Correct:460 #Tested:535 Testing Accuracy:85.9%\r",
      "Progress:53.5% Speed(reviews/sec):3267. #Correct:461 #Tested:536 Testing Accuracy:86.0%\r",
      "Progress:53.6% Speed(reviews/sec):3271. #Correct:461 #Tested:537 Testing Accuracy:85.8%\r",
      "Progress:53.7% Speed(reviews/sec):3275. #Correct:462 #Tested:538 Testing Accuracy:85.8%\r",
      "Progress:53.8% Speed(reviews/sec):3278. #Correct:463 #Tested:539 Testing Accuracy:85.8%\r",
      "Progress:53.9% Speed(reviews/sec):3278. #Correct:464 #Tested:540 Testing Accuracy:85.9%\r",
      "Progress:54.0% Speed(reviews/sec):3280. #Correct:465 #Tested:541 Testing Accuracy:85.9%\r",
      "Progress:54.1% Speed(reviews/sec):3283. #Correct:466 #Tested:542 Testing Accuracy:85.9%\r",
      "Progress:54.2% Speed(reviews/sec):3284. #Correct:467 #Tested:543 Testing Accuracy:86.0%\r",
      "Progress:54.3% Speed(reviews/sec):3287. #Correct:468 #Tested:544 Testing Accuracy:86.0%\r",
      "Progress:54.4% Speed(reviews/sec):3289. #Correct:468 #Tested:545 Testing Accuracy:85.8%\r",
      "Progress:54.5% Speed(reviews/sec):3290. #Correct:469 #Tested:546 Testing Accuracy:85.8%\r",
      "Progress:54.6% Speed(reviews/sec):3290. #Correct:469 #Tested:547 Testing Accuracy:85.7%\r",
      "Progress:54.7% Speed(reviews/sec):3292. #Correct:470 #Tested:548 Testing Accuracy:85.7%\r",
      "Progress:54.8% Speed(reviews/sec):3268. #Correct:471 #Tested:549 Testing Accuracy:85.7%\r",
      "Progress:54.9% Speed(reviews/sec):3269. #Correct:472 #Tested:550 Testing Accuracy:85.8%\r",
      "Progress:55.0% Speed(reviews/sec):3271. #Correct:473 #Tested:551 Testing Accuracy:85.8%\r",
      "Progress:55.1% Speed(reviews/sec):3272. #Correct:474 #Tested:552 Testing Accuracy:85.8%\r",
      "Progress:55.2% Speed(reviews/sec):3275. #Correct:475 #Tested:553 Testing Accuracy:85.8%\r",
      "Progress:55.3% Speed(reviews/sec):3278. #Correct:476 #Tested:554 Testing Accuracy:85.9%\r",
      "Progress:55.4% Speed(reviews/sec):3277. #Correct:477 #Tested:555 Testing Accuracy:85.9%\r",
      "Progress:55.5% Speed(reviews/sec):3280. #Correct:478 #Tested:556 Testing Accuracy:85.9%\r",
      "Progress:55.6% Speed(reviews/sec):3282. #Correct:479 #Tested:557 Testing Accuracy:85.9%\r",
      "Progress:55.7% Speed(reviews/sec):3284. #Correct:480 #Tested:558 Testing Accuracy:86.0%\r",
      "Progress:55.8% Speed(reviews/sec):3283. #Correct:480 #Tested:559 Testing Accuracy:85.8%\r",
      "Progress:55.9% Speed(reviews/sec):3285. #Correct:481 #Tested:560 Testing Accuracy:85.8%\r",
      "Progress:56.0% Speed(reviews/sec):3288. #Correct:482 #Tested:561 Testing Accuracy:85.9%\r",
      "Progress:56.1% Speed(reviews/sec):3290. #Correct:483 #Tested:562 Testing Accuracy:85.9%\r",
      "Progress:56.2% Speed(reviews/sec):3292. #Correct:484 #Tested:563 Testing Accuracy:85.9%\r",
      "Progress:56.3% Speed(reviews/sec):3295. #Correct:485 #Tested:564 Testing Accuracy:85.9%\r",
      "Progress:56.4% Speed(reviews/sec):3296. #Correct:486 #Tested:565 Testing Accuracy:86.0%\r",
      "Progress:56.5% Speed(reviews/sec):3300. #Correct:487 #Tested:566 Testing Accuracy:86.0%\r",
      "Progress:56.6% Speed(reviews/sec):3302. #Correct:488 #Tested:567 Testing Accuracy:86.0%\r",
      "Progress:56.7% Speed(reviews/sec):3305. #Correct:489 #Tested:568 Testing Accuracy:86.0%\r",
      "Progress:56.8% Speed(reviews/sec):3308. #Correct:490 #Tested:569 Testing Accuracy:86.1%\r",
      "Progress:56.9% Speed(reviews/sec):3310. #Correct:491 #Tested:570 Testing Accuracy:86.1%\r",
      "Progress:57.0% Speed(reviews/sec):3308. #Correct:492 #Tested:571 Testing Accuracy:86.1%\r",
      "Progress:57.1% Speed(reviews/sec):3278. #Correct:493 #Tested:572 Testing Accuracy:86.1%\r",
      "Progress:57.2% Speed(reviews/sec):3276. #Correct:493 #Tested:573 Testing Accuracy:86.0%\r",
      "Progress:57.3% Speed(reviews/sec):3278. #Correct:493 #Tested:574 Testing Accuracy:85.8%\r",
      "Progress:57.4% Speed(reviews/sec):3278. #Correct:494 #Tested:575 Testing Accuracy:85.9%\r",
      "Progress:57.5% Speed(reviews/sec):3280. #Correct:495 #Tested:576 Testing Accuracy:85.9%\r",
      "Progress:57.6% Speed(reviews/sec):3283. #Correct:496 #Tested:577 Testing Accuracy:85.9%\r",
      "Progress:57.7% Speed(reviews/sec):3284. #Correct:497 #Tested:578 Testing Accuracy:85.9%\r",
      "Progress:57.8% Speed(reviews/sec):3282. #Correct:498 #Tested:579 Testing Accuracy:86.0%\r",
      "Progress:57.9% Speed(reviews/sec):3283. #Correct:499 #Tested:580 Testing Accuracy:86.0%\r",
      "Progress:58.0% Speed(reviews/sec):3284. #Correct:500 #Tested:581 Testing Accuracy:86.0%\r",
      "Progress:58.1% Speed(reviews/sec):3286. #Correct:501 #Tested:582 Testing Accuracy:86.0%\r",
      "Progress:58.2% Speed(reviews/sec):3289. #Correct:502 #Tested:583 Testing Accuracy:86.1%\r",
      "Progress:58.3% Speed(reviews/sec):3291. #Correct:503 #Tested:584 Testing Accuracy:86.1%\r",
      "Progress:58.4% Speed(reviews/sec):3293. #Correct:504 #Tested:585 Testing Accuracy:86.1%\r",
      "Progress:58.5% Speed(reviews/sec):3296. #Correct:505 #Tested:586 Testing Accuracy:86.1%\r",
      "Progress:58.6% Speed(reviews/sec):3299. #Correct:506 #Tested:587 Testing Accuracy:86.2%\r",
      "Progress:58.7% Speed(reviews/sec):3302. #Correct:507 #Tested:588 Testing Accuracy:86.2%\r",
      "Progress:58.8% Speed(reviews/sec):3299. #Correct:508 #Tested:589 Testing Accuracy:86.2%\r",
      "Progress:58.9% Speed(reviews/sec):3298. #Correct:509 #Tested:590 Testing Accuracy:86.2%\r",
      "Progress:59.0% Speed(reviews/sec):3301. #Correct:510 #Tested:591 Testing Accuracy:86.2%\r",
      "Progress:59.1% Speed(reviews/sec):3303. #Correct:511 #Tested:592 Testing Accuracy:86.3%\r",
      "Progress:59.2% Speed(reviews/sec):3302. #Correct:511 #Tested:593 Testing Accuracy:86.1%\r",
      "Progress:59.3% Speed(reviews/sec):3304. #Correct:512 #Tested:594 Testing Accuracy:86.1%\r",
      "Progress:59.4% Speed(reviews/sec):3306. #Correct:513 #Tested:595 Testing Accuracy:86.2%\r",
      "Progress:59.5% Speed(reviews/sec):3309. #Correct:514 #Tested:596 Testing Accuracy:86.2%\r",
      "Progress:59.6% Speed(reviews/sec):3313. #Correct:515 #Tested:597 Testing Accuracy:86.2%\r",
      "Progress:59.7% Speed(reviews/sec):3314. #Correct:516 #Tested:598 Testing Accuracy:86.2%\r",
      "Progress:59.8% Speed(reviews/sec):3314. #Correct:516 #Tested:599 Testing Accuracy:86.1%\r",
      "Progress:59.9% Speed(reviews/sec):3315. #Correct:517 #Tested:600 Testing Accuracy:86.1%\r",
      "Progress:60.0% Speed(reviews/sec):3317. #Correct:517 #Tested:601 Testing Accuracy:86.0%\r",
      "Progress:60.1% Speed(reviews/sec):3320. #Correct:518 #Tested:602 Testing Accuracy:86.0%\r",
      "Progress:60.2% Speed(reviews/sec):3323. #Correct:519 #Tested:603 Testing Accuracy:86.0%\r",
      "Progress:60.3% Speed(reviews/sec):3326. #Correct:520 #Tested:604 Testing Accuracy:86.0%\r",
      "Progress:60.4% Speed(reviews/sec):3329. #Correct:521 #Tested:605 Testing Accuracy:86.1%\r",
      "Progress:60.5% Speed(reviews/sec):3327. #Correct:522 #Tested:606 Testing Accuracy:86.1%\r",
      "Progress:60.6% Speed(reviews/sec):3330. #Correct:522 #Tested:607 Testing Accuracy:85.9%\r",
      "Progress:60.7% Speed(reviews/sec):3333. #Correct:523 #Tested:608 Testing Accuracy:86.0%\r",
      "Progress:60.8% Speed(reviews/sec):3313. #Correct:524 #Tested:609 Testing Accuracy:86.0%\r",
      "Progress:60.9% Speed(reviews/sec):3316. #Correct:525 #Tested:610 Testing Accuracy:86.0%\r",
      "Progress:61.0% Speed(reviews/sec):3317. #Correct:525 #Tested:611 Testing Accuracy:85.9%\r",
      "Progress:61.1% Speed(reviews/sec):3320. #Correct:526 #Tested:612 Testing Accuracy:85.9%\r",
      "Progress:61.2% Speed(reviews/sec):3298. #Correct:527 #Tested:613 Testing Accuracy:85.9%\r",
      "Progress:61.3% Speed(reviews/sec):3300. #Correct:528 #Tested:614 Testing Accuracy:85.9%\r",
      "Progress:61.4% Speed(reviews/sec):3303. #Correct:528 #Tested:615 Testing Accuracy:85.8%\r",
      "Progress:61.5% Speed(reviews/sec):3305. #Correct:528 #Tested:616 Testing Accuracy:85.7%\r",
      "Progress:61.6% Speed(reviews/sec):3305. #Correct:529 #Tested:617 Testing Accuracy:85.7%\r",
      "Progress:61.7% Speed(reviews/sec):3308. #Correct:530 #Tested:618 Testing Accuracy:85.7%\r",
      "Progress:61.8% Speed(reviews/sec):3312. #Correct:531 #Tested:619 Testing Accuracy:85.7%\r",
      "Progress:61.9% Speed(reviews/sec):3314. #Correct:531 #Tested:620 Testing Accuracy:85.6%\r",
      "Progress:62.0% Speed(reviews/sec):3315. #Correct:531 #Tested:621 Testing Accuracy:85.5%\r",
      "Progress:62.1% Speed(reviews/sec):3318. #Correct:532 #Tested:622 Testing Accuracy:85.5%\r",
      "Progress:62.2% Speed(reviews/sec):3321. #Correct:532 #Tested:623 Testing Accuracy:85.3%\r",
      "Progress:62.3% Speed(reviews/sec):3324. #Correct:533 #Tested:624 Testing Accuracy:85.4%\r",
      "Progress:62.4% Speed(reviews/sec):3326. #Correct:533 #Tested:625 Testing Accuracy:85.2%\r",
      "Progress:62.5% Speed(reviews/sec):3323. #Correct:533 #Tested:626 Testing Accuracy:85.1%\r",
      "Progress:62.6% Speed(reviews/sec):3319. #Correct:533 #Tested:627 Testing Accuracy:85.0%\r",
      "Progress:62.7% Speed(reviews/sec):3321. #Correct:533 #Tested:628 Testing Accuracy:84.8%\r",
      "Progress:62.8% Speed(reviews/sec):3325. #Correct:533 #Tested:629 Testing Accuracy:84.7%\r",
      "Progress:62.9% Speed(reviews/sec):3324. #Correct:534 #Tested:630 Testing Accuracy:84.7%\r",
      "Progress:63.0% Speed(reviews/sec):3326. #Correct:534 #Tested:631 Testing Accuracy:84.6%\r",
      "Progress:63.1% Speed(reviews/sec):3324. #Correct:535 #Tested:632 Testing Accuracy:84.6%\r",
      "Progress:63.2% Speed(reviews/sec):3327. #Correct:535 #Tested:633 Testing Accuracy:84.5%\r",
      "Progress:63.3% Speed(reviews/sec):3329. #Correct:536 #Tested:634 Testing Accuracy:84.5%\r",
      "Progress:63.4% Speed(reviews/sec):3301. #Correct:536 #Tested:635 Testing Accuracy:84.4%\r",
      "Progress:63.5% Speed(reviews/sec):3301. #Correct:537 #Tested:636 Testing Accuracy:84.4%\r",
      "Progress:63.6% Speed(reviews/sec):3304. #Correct:537 #Tested:637 Testing Accuracy:84.3%\r",
      "Progress:63.7% Speed(reviews/sec):3305. #Correct:538 #Tested:638 Testing Accuracy:84.3%\r",
      "Progress:63.8% Speed(reviews/sec):3308. #Correct:539 #Tested:639 Testing Accuracy:84.3%\r",
      "Progress:63.9% Speed(reviews/sec):3312. #Correct:540 #Tested:640 Testing Accuracy:84.3%\r",
      "Progress:64.0% Speed(reviews/sec):3315. #Correct:540 #Tested:641 Testing Accuracy:84.2%\r",
      "Progress:64.1% Speed(reviews/sec):3317. #Correct:540 #Tested:642 Testing Accuracy:84.1%\r",
      "Progress:64.2% Speed(reviews/sec):3319. #Correct:541 #Tested:643 Testing Accuracy:84.1%\r",
      "Progress:64.3% Speed(reviews/sec):3321. #Correct:542 #Tested:644 Testing Accuracy:84.1%\r",
      "Progress:64.4% Speed(reviews/sec):3322. #Correct:543 #Tested:645 Testing Accuracy:84.1%\r",
      "Progress:64.5% Speed(reviews/sec):3323. #Correct:543 #Tested:646 Testing Accuracy:84.0%\r",
      "Progress:64.6% Speed(reviews/sec):3306. #Correct:544 #Tested:647 Testing Accuracy:84.0%\r",
      "Progress:64.7% Speed(reviews/sec):3308. #Correct:545 #Tested:648 Testing Accuracy:84.1%\r",
      "Progress:64.8% Speed(reviews/sec):3310. #Correct:546 #Tested:649 Testing Accuracy:84.1%\r",
      "Progress:64.9% Speed(reviews/sec):3312. #Correct:547 #Tested:650 Testing Accuracy:84.1%\r",
      "Progress:65.0% Speed(reviews/sec):3315. #Correct:547 #Tested:651 Testing Accuracy:84.0%\r",
      "Progress:65.1% Speed(reviews/sec):3317. #Correct:548 #Tested:652 Testing Accuracy:84.0%\r",
      "Progress:65.2% Speed(reviews/sec):3320. #Correct:549 #Tested:653 Testing Accuracy:84.0%\r",
      "Progress:65.3% Speed(reviews/sec):3321. #Correct:550 #Tested:654 Testing Accuracy:84.0%\r",
      "Progress:65.4% Speed(reviews/sec):3323. #Correct:550 #Tested:655 Testing Accuracy:83.9%\r",
      "Progress:65.5% Speed(reviews/sec):3325. #Correct:551 #Tested:656 Testing Accuracy:83.9%\r",
      "Progress:65.6% Speed(reviews/sec):3328. #Correct:551 #Tested:657 Testing Accuracy:83.8%\r",
      "Progress:65.7% Speed(reviews/sec):3330. #Correct:552 #Tested:658 Testing Accuracy:83.8%\r",
      "Progress:65.8% Speed(reviews/sec):3333. #Correct:553 #Tested:659 Testing Accuracy:83.9%\r",
      "Progress:65.9% Speed(reviews/sec):3335. #Correct:554 #Tested:660 Testing Accuracy:83.9%\r",
      "Progress:66.0% Speed(reviews/sec):3338. #Correct:555 #Tested:661 Testing Accuracy:83.9%\r",
      "Progress:66.1% Speed(reviews/sec):3340. #Correct:556 #Tested:662 Testing Accuracy:83.9%\r",
      "Progress:66.2% Speed(reviews/sec):3342. #Correct:557 #Tested:663 Testing Accuracy:84.0%\r",
      "Progress:66.3% Speed(reviews/sec):3324. #Correct:557 #Tested:664 Testing Accuracy:83.8%\r",
      "Progress:66.4% Speed(reviews/sec):3326. #Correct:558 #Tested:665 Testing Accuracy:83.9%\r",
      "Progress:66.5% Speed(reviews/sec):3329. #Correct:559 #Tested:666 Testing Accuracy:83.9%\r",
      "Progress:66.6% Speed(reviews/sec):3331. #Correct:560 #Tested:667 Testing Accuracy:83.9%\r",
      "Progress:66.7% Speed(reviews/sec):3333. #Correct:561 #Tested:668 Testing Accuracy:83.9%\r",
      "Progress:66.8% Speed(reviews/sec):3334. #Correct:562 #Tested:669 Testing Accuracy:84.0%\r",
      "Progress:66.9% Speed(reviews/sec):3337. #Correct:562 #Tested:670 Testing Accuracy:83.8%\r",
      "Progress:67.0% Speed(reviews/sec):3340. #Correct:563 #Tested:671 Testing Accuracy:83.9%\r",
      "Progress:67.1% Speed(reviews/sec):3343. #Correct:564 #Tested:672 Testing Accuracy:83.9%\r",
      "Progress:67.2% Speed(reviews/sec):3346. #Correct:565 #Tested:673 Testing Accuracy:83.9%\r",
      "Progress:67.3% Speed(reviews/sec):3348. #Correct:566 #Tested:674 Testing Accuracy:83.9%\r",
      "Progress:67.4% Speed(reviews/sec):3351. #Correct:567 #Tested:675 Testing Accuracy:84.0%\r",
      "Progress:67.5% Speed(reviews/sec):3354. #Correct:568 #Tested:676 Testing Accuracy:84.0%\r",
      "Progress:67.6% Speed(reviews/sec):3357. #Correct:568 #Tested:677 Testing Accuracy:83.8%\r",
      "Progress:67.7% Speed(reviews/sec):3359. #Correct:568 #Tested:678 Testing Accuracy:83.7%\r",
      "Progress:67.8% Speed(reviews/sec):3362. #Correct:569 #Tested:679 Testing Accuracy:83.7%\r",
      "Progress:67.9% Speed(reviews/sec):3365. #Correct:570 #Tested:680 Testing Accuracy:83.8%\r",
      "Progress:68.0% Speed(reviews/sec):3348. #Correct:570 #Tested:681 Testing Accuracy:83.7%\r",
      "Progress:68.1% Speed(reviews/sec):3349. #Correct:571 #Tested:682 Testing Accuracy:83.7%\r",
      "Progress:68.2% Speed(reviews/sec):3352. #Correct:572 #Tested:683 Testing Accuracy:83.7%\r",
      "Progress:68.3% Speed(reviews/sec):3355. #Correct:573 #Tested:684 Testing Accuracy:83.7%\r",
      "Progress:68.4% Speed(reviews/sec):3357. #Correct:574 #Tested:685 Testing Accuracy:83.7%\r",
      "Progress:68.5% Speed(reviews/sec):3359. #Correct:575 #Tested:686 Testing Accuracy:83.8%\r",
      "Progress:68.6% Speed(reviews/sec):3362. #Correct:575 #Tested:687 Testing Accuracy:83.6%\r",
      "Progress:68.7% Speed(reviews/sec):3366. #Correct:576 #Tested:688 Testing Accuracy:83.7%\r",
      "Progress:68.8% Speed(reviews/sec):3363. #Correct:577 #Tested:689 Testing Accuracy:83.7%\r",
      "Progress:68.9% Speed(reviews/sec):3365. #Correct:578 #Tested:690 Testing Accuracy:83.7%\r",
      "Progress:69.0% Speed(reviews/sec):3367. #Correct:579 #Tested:691 Testing Accuracy:83.7%\r",
      "Progress:69.1% Speed(reviews/sec):3370. #Correct:580 #Tested:692 Testing Accuracy:83.8%\r",
      "Progress:69.2% Speed(reviews/sec):3370. #Correct:581 #Tested:693 Testing Accuracy:83.8%\r",
      "Progress:69.3% Speed(reviews/sec):3372. #Correct:582 #Tested:694 Testing Accuracy:83.8%\r",
      "Progress:69.4% Speed(reviews/sec):3374. #Correct:582 #Tested:695 Testing Accuracy:83.7%\r",
      "Progress:69.5% Speed(reviews/sec):3375. #Correct:583 #Tested:696 Testing Accuracy:83.7%\r",
      "Progress:69.6% Speed(reviews/sec):3378. #Correct:584 #Tested:697 Testing Accuracy:83.7%\r",
      "Progress:69.7% Speed(reviews/sec):3381. #Correct:585 #Tested:698 Testing Accuracy:83.8%\r",
      "Progress:69.8% Speed(reviews/sec):3384. #Correct:586 #Tested:699 Testing Accuracy:83.8%\r",
      "Progress:69.9% Speed(reviews/sec):3386. #Correct:587 #Tested:700 Testing Accuracy:83.8%\r",
      "Progress:70.0% Speed(reviews/sec):3388. #Correct:588 #Tested:701 Testing Accuracy:83.8%\r",
      "Progress:70.1% Speed(reviews/sec):3390. #Correct:589 #Tested:702 Testing Accuracy:83.9%\r",
      "Progress:70.2% Speed(reviews/sec):3392. #Correct:590 #Tested:703 Testing Accuracy:83.9%\r",
      "Progress:70.3% Speed(reviews/sec):3386. #Correct:591 #Tested:704 Testing Accuracy:83.9%\r",
      "Progress:70.4% Speed(reviews/sec):3387. #Correct:592 #Tested:705 Testing Accuracy:83.9%\r",
      "Progress:70.5% Speed(reviews/sec):3388. #Correct:593 #Tested:706 Testing Accuracy:83.9%\r",
      "Progress:70.6% Speed(reviews/sec):3390. #Correct:594 #Tested:707 Testing Accuracy:84.0%\r",
      "Progress:70.7% Speed(reviews/sec):3392. #Correct:595 #Tested:708 Testing Accuracy:84.0%\r",
      "Progress:70.8% Speed(reviews/sec):3394. #Correct:596 #Tested:709 Testing Accuracy:84.0%\r",
      "Progress:70.9% Speed(reviews/sec):3396. #Correct:597 #Tested:710 Testing Accuracy:84.0%\r",
      "Progress:71.0% Speed(reviews/sec):3397. #Correct:598 #Tested:711 Testing Accuracy:84.1%\r",
      "Progress:71.1% Speed(reviews/sec):3398. #Correct:599 #Tested:712 Testing Accuracy:84.1%\r",
      "Progress:71.2% Speed(reviews/sec):3400. #Correct:599 #Tested:713 Testing Accuracy:84.0%\r",
      "Progress:71.3% Speed(reviews/sec):3399. #Correct:600 #Tested:714 Testing Accuracy:84.0%\r",
      "Progress:71.4% Speed(reviews/sec):3399. #Correct:601 #Tested:715 Testing Accuracy:84.0%\r",
      "Progress:71.5% Speed(reviews/sec):3401. #Correct:602 #Tested:716 Testing Accuracy:84.0%\r",
      "Progress:71.6% Speed(reviews/sec):3403. #Correct:603 #Tested:717 Testing Accuracy:84.1%\r",
      "Progress:71.7% Speed(reviews/sec):3406. #Correct:604 #Tested:718 Testing Accuracy:84.1%\r",
      "Progress:71.8% Speed(reviews/sec):3407. #Correct:605 #Tested:719 Testing Accuracy:84.1%\r",
      "Progress:71.9% Speed(reviews/sec):3409. #Correct:606 #Tested:720 Testing Accuracy:84.1%\r",
      "Progress:72.0% Speed(reviews/sec):3410. #Correct:606 #Tested:721 Testing Accuracy:84.0%\r",
      "Progress:72.1% Speed(reviews/sec):3412. #Correct:607 #Tested:722 Testing Accuracy:84.0%\r",
      "Progress:72.2% Speed(reviews/sec):3414. #Correct:608 #Tested:723 Testing Accuracy:84.0%\r",
      "Progress:72.3% Speed(reviews/sec):3416. #Correct:609 #Tested:724 Testing Accuracy:84.1%\r",
      "Progress:72.4% Speed(reviews/sec):3419. #Correct:609 #Tested:725 Testing Accuracy:84.0%\r",
      "Progress:72.5% Speed(reviews/sec):3421. #Correct:610 #Tested:726 Testing Accuracy:84.0%\r",
      "Progress:72.6% Speed(reviews/sec):3420. #Correct:611 #Tested:727 Testing Accuracy:84.0%\r",
      "Progress:72.7% Speed(reviews/sec):3419. #Correct:612 #Tested:728 Testing Accuracy:84.0%\r",
      "Progress:72.8% Speed(reviews/sec):3422. #Correct:613 #Tested:729 Testing Accuracy:84.0%\r",
      "Progress:72.9% Speed(reviews/sec):3421. #Correct:614 #Tested:730 Testing Accuracy:84.1%\r",
      "Progress:73.0% Speed(reviews/sec):3383. #Correct:615 #Tested:731 Testing Accuracy:84.1%\r",
      "Progress:73.1% Speed(reviews/sec):3385. #Correct:616 #Tested:732 Testing Accuracy:84.1%\r",
      "Progress:73.2% Speed(reviews/sec):3385. #Correct:617 #Tested:733 Testing Accuracy:84.1%\r",
      "Progress:73.3% Speed(reviews/sec):3386. #Correct:618 #Tested:734 Testing Accuracy:84.1%\r",
      "Progress:73.4% Speed(reviews/sec):3386. #Correct:619 #Tested:735 Testing Accuracy:84.2%\r",
      "Progress:73.5% Speed(reviews/sec):3387. #Correct:620 #Tested:736 Testing Accuracy:84.2%\r",
      "Progress:73.6% Speed(reviews/sec):3390. #Correct:621 #Tested:737 Testing Accuracy:84.2%\r",
      "Progress:73.7% Speed(reviews/sec):3392. #Correct:621 #Tested:738 Testing Accuracy:84.1%\r",
      "Progress:73.8% Speed(reviews/sec):3394. #Correct:622 #Tested:739 Testing Accuracy:84.1%\r",
      "Progress:73.9% Speed(reviews/sec):3394. #Correct:623 #Tested:740 Testing Accuracy:84.1%"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\r",
      "Progress:74.0% Speed(reviews/sec):3391. #Correct:624 #Tested:741 Testing Accuracy:84.2%\r",
      "Progress:74.1% Speed(reviews/sec):3393. #Correct:625 #Tested:742 Testing Accuracy:84.2%\r",
      "Progress:74.2% Speed(reviews/sec):3395. #Correct:626 #Tested:743 Testing Accuracy:84.2%\r",
      "Progress:74.3% Speed(reviews/sec):3396. #Correct:626 #Tested:744 Testing Accuracy:84.1%\r",
      "Progress:74.4% Speed(reviews/sec):3398. #Correct:627 #Tested:745 Testing Accuracy:84.1%\r",
      "Progress:74.5% Speed(reviews/sec):3397. #Correct:627 #Tested:746 Testing Accuracy:84.0%\r",
      "Progress:74.6% Speed(reviews/sec):3398. #Correct:628 #Tested:747 Testing Accuracy:84.0%\r",
      "Progress:74.7% Speed(reviews/sec):3395. #Correct:628 #Tested:748 Testing Accuracy:83.9%\r",
      "Progress:74.8% Speed(reviews/sec):3380. #Correct:629 #Tested:749 Testing Accuracy:83.9%\r",
      "Progress:74.9% Speed(reviews/sec):3380. #Correct:630 #Tested:750 Testing Accuracy:84.0%\r",
      "Progress:75.0% Speed(reviews/sec):3382. #Correct:631 #Tested:751 Testing Accuracy:84.0%\r",
      "Progress:75.1% Speed(reviews/sec):3385. #Correct:632 #Tested:752 Testing Accuracy:84.0%\r",
      "Progress:75.2% Speed(reviews/sec):3383. #Correct:633 #Tested:753 Testing Accuracy:84.0%\r",
      "Progress:75.3% Speed(reviews/sec):3384. #Correct:634 #Tested:754 Testing Accuracy:84.0%\r",
      "Progress:75.4% Speed(reviews/sec):3383. #Correct:635 #Tested:755 Testing Accuracy:84.1%\r",
      "Progress:75.5% Speed(reviews/sec):3386. #Correct:635 #Tested:756 Testing Accuracy:83.9%\r",
      "Progress:75.6% Speed(reviews/sec):3387. #Correct:635 #Tested:757 Testing Accuracy:83.8%\r",
      "Progress:75.7% Speed(reviews/sec):3385. #Correct:636 #Tested:758 Testing Accuracy:83.9%\r",
      "Progress:75.8% Speed(reviews/sec):3380. #Correct:636 #Tested:759 Testing Accuracy:83.7%\r",
      "Progress:75.9% Speed(reviews/sec):3376. #Correct:637 #Tested:760 Testing Accuracy:83.8%\r",
      "Progress:76.0% Speed(reviews/sec):3377. #Correct:637 #Tested:761 Testing Accuracy:83.7%\r",
      "Progress:76.1% Speed(reviews/sec):3374. #Correct:638 #Tested:762 Testing Accuracy:83.7%\r",
      "Progress:76.2% Speed(reviews/sec):3370. #Correct:638 #Tested:763 Testing Accuracy:83.6%\r",
      "Progress:76.3% Speed(reviews/sec):3372. #Correct:639 #Tested:764 Testing Accuracy:83.6%\r",
      "Progress:76.4% Speed(reviews/sec):3350. #Correct:639 #Tested:765 Testing Accuracy:83.5%\r",
      "Progress:76.5% Speed(reviews/sec):3350. #Correct:639 #Tested:766 Testing Accuracy:83.4%\r",
      "Progress:76.6% Speed(reviews/sec):3351. #Correct:639 #Tested:767 Testing Accuracy:83.3%\r",
      "Progress:76.7% Speed(reviews/sec):3352. #Correct:639 #Tested:768 Testing Accuracy:83.2%\r",
      "Progress:76.8% Speed(reviews/sec):3354. #Correct:639 #Tested:769 Testing Accuracy:83.0%\r",
      "Progress:76.9% Speed(reviews/sec):3356. #Correct:640 #Tested:770 Testing Accuracy:83.1%\r",
      "Progress:77.0% Speed(reviews/sec):3355. #Correct:640 #Tested:771 Testing Accuracy:83.0%\r",
      "Progress:77.1% Speed(reviews/sec):3357. #Correct:641 #Tested:772 Testing Accuracy:83.0%\r",
      "Progress:77.2% Speed(reviews/sec):3359. #Correct:642 #Tested:773 Testing Accuracy:83.0%\r",
      "Progress:77.3% Speed(reviews/sec):3360. #Correct:643 #Tested:774 Testing Accuracy:83.0%\r",
      "Progress:77.4% Speed(reviews/sec):3362. #Correct:644 #Tested:775 Testing Accuracy:83.0%\r",
      "Progress:77.5% Speed(reviews/sec):3362. #Correct:645 #Tested:776 Testing Accuracy:83.1%\r",
      "Progress:77.6% Speed(reviews/sec):3360. #Correct:645 #Tested:777 Testing Accuracy:83.0%\r",
      "Progress:77.7% Speed(reviews/sec):3360. #Correct:645 #Tested:778 Testing Accuracy:82.9%\r",
      "Progress:77.8% Speed(reviews/sec):3358. #Correct:646 #Tested:779 Testing Accuracy:82.9%\r",
      "Progress:77.9% Speed(reviews/sec):3359. #Correct:647 #Tested:780 Testing Accuracy:82.9%\r",
      "Progress:78.0% Speed(reviews/sec):3356. #Correct:647 #Tested:781 Testing Accuracy:82.8%\r",
      "Progress:78.1% Speed(reviews/sec):3358. #Correct:648 #Tested:782 Testing Accuracy:82.8%\r",
      "Progress:78.2% Speed(reviews/sec):3358. #Correct:648 #Tested:783 Testing Accuracy:82.7%\r",
      "Progress:78.3% Speed(reviews/sec):3361. #Correct:649 #Tested:784 Testing Accuracy:82.7%\r",
      "Progress:78.4% Speed(reviews/sec):3362. #Correct:649 #Tested:785 Testing Accuracy:82.6%\r",
      "Progress:78.5% Speed(reviews/sec):3364. #Correct:650 #Tested:786 Testing Accuracy:82.6%\r",
      "Progress:78.6% Speed(reviews/sec):3366. #Correct:650 #Tested:787 Testing Accuracy:82.5%\r",
      "Progress:78.7% Speed(reviews/sec):3364. #Correct:651 #Tested:788 Testing Accuracy:82.6%\r",
      "Progress:78.8% Speed(reviews/sec):3366. #Correct:651 #Tested:789 Testing Accuracy:82.5%\r",
      "Progress:78.9% Speed(reviews/sec):3366. #Correct:652 #Tested:790 Testing Accuracy:82.5%\r",
      "Progress:79.0% Speed(reviews/sec):3364. #Correct:652 #Tested:791 Testing Accuracy:82.4%\r",
      "Progress:79.1% Speed(reviews/sec):3364. #Correct:653 #Tested:792 Testing Accuracy:82.4%\r",
      "Progress:79.2% Speed(reviews/sec):3364. #Correct:653 #Tested:793 Testing Accuracy:82.3%\r",
      "Progress:79.3% Speed(reviews/sec):3366. #Correct:653 #Tested:794 Testing Accuracy:82.2%\r",
      "Progress:79.4% Speed(reviews/sec):3367. #Correct:654 #Tested:795 Testing Accuracy:82.2%\r",
      "Progress:79.5% Speed(reviews/sec):3370. #Correct:655 #Tested:796 Testing Accuracy:82.2%\r",
      "Progress:79.6% Speed(reviews/sec):3372. #Correct:656 #Tested:797 Testing Accuracy:82.3%\r",
      "Progress:79.7% Speed(reviews/sec):3375. #Correct:657 #Tested:798 Testing Accuracy:82.3%\r",
      "Progress:79.8% Speed(reviews/sec):3378. #Correct:658 #Tested:799 Testing Accuracy:82.3%\r",
      "Progress:79.9% Speed(reviews/sec):3380. #Correct:659 #Tested:800 Testing Accuracy:82.3%\r",
      "Progress:80.0% Speed(reviews/sec):3383. #Correct:660 #Tested:801 Testing Accuracy:82.3%\r",
      "Progress:80.1% Speed(reviews/sec):3385. #Correct:661 #Tested:802 Testing Accuracy:82.4%\r",
      "Progress:80.2% Speed(reviews/sec):3388. #Correct:662 #Tested:803 Testing Accuracy:82.4%\r",
      "Progress:80.3% Speed(reviews/sec):3390. #Correct:663 #Tested:804 Testing Accuracy:82.4%\r",
      "Progress:80.4% Speed(reviews/sec):3393. #Correct:664 #Tested:805 Testing Accuracy:82.4%\r",
      "Progress:80.5% Speed(reviews/sec):3396. #Correct:664 #Tested:806 Testing Accuracy:82.3%\r",
      "Progress:80.6% Speed(reviews/sec):3397. #Correct:665 #Tested:807 Testing Accuracy:82.4%\r",
      "Progress:80.7% Speed(reviews/sec):3398. #Correct:666 #Tested:808 Testing Accuracy:82.4%\r",
      "Progress:80.8% Speed(reviews/sec):3401. #Correct:667 #Tested:809 Testing Accuracy:82.4%\r",
      "Progress:80.9% Speed(reviews/sec):3401. #Correct:668 #Tested:810 Testing Accuracy:82.4%\r",
      "Progress:81.0% Speed(reviews/sec):3402. #Correct:669 #Tested:811 Testing Accuracy:82.4%\r",
      "Progress:81.1% Speed(reviews/sec):3403. #Correct:670 #Tested:812 Testing Accuracy:82.5%\r",
      "Progress:81.2% Speed(reviews/sec):3405. #Correct:671 #Tested:813 Testing Accuracy:82.5%\r",
      "Progress:81.3% Speed(reviews/sec):3405. #Correct:672 #Tested:814 Testing Accuracy:82.5%\r",
      "Progress:81.4% Speed(reviews/sec):3405. #Correct:673 #Tested:815 Testing Accuracy:82.5%\r",
      "Progress:81.5% Speed(reviews/sec):3407. #Correct:674 #Tested:816 Testing Accuracy:82.5%\r",
      "Progress:81.6% Speed(reviews/sec):3409. #Correct:675 #Tested:817 Testing Accuracy:82.6%\r",
      "Progress:81.7% Speed(reviews/sec):3409. #Correct:676 #Tested:818 Testing Accuracy:82.6%\r",
      "Progress:81.8% Speed(reviews/sec):3410. #Correct:677 #Tested:819 Testing Accuracy:82.6%\r",
      "Progress:81.9% Speed(reviews/sec):3412. #Correct:677 #Tested:820 Testing Accuracy:82.5%\r",
      "Progress:82.0% Speed(reviews/sec):3410. #Correct:678 #Tested:821 Testing Accuracy:82.5%\r",
      "Progress:82.1% Speed(reviews/sec):3410. #Correct:678 #Tested:822 Testing Accuracy:82.4%\r",
      "Progress:82.2% Speed(reviews/sec):3412. #Correct:678 #Tested:823 Testing Accuracy:82.3%\r",
      "Progress:82.3% Speed(reviews/sec):3414. #Correct:679 #Tested:824 Testing Accuracy:82.4%\r",
      "Progress:82.4% Speed(reviews/sec):3416. #Correct:680 #Tested:825 Testing Accuracy:82.4%\r",
      "Progress:82.5% Speed(reviews/sec):3416. #Correct:681 #Tested:826 Testing Accuracy:82.4%\r",
      "Progress:82.6% Speed(reviews/sec):3412. #Correct:682 #Tested:827 Testing Accuracy:82.4%\r",
      "Progress:82.7% Speed(reviews/sec):3414. #Correct:683 #Tested:828 Testing Accuracy:82.4%\r",
      "Progress:82.8% Speed(reviews/sec):3414. #Correct:684 #Tested:829 Testing Accuracy:82.5%\r",
      "Progress:82.9% Speed(reviews/sec):3413. #Correct:685 #Tested:830 Testing Accuracy:82.5%\r",
      "Progress:83.0% Speed(reviews/sec):3414. #Correct:686 #Tested:831 Testing Accuracy:82.5%\r",
      "Progress:83.1% Speed(reviews/sec):3416. #Correct:687 #Tested:832 Testing Accuracy:82.5%\r",
      "Progress:83.2% Speed(reviews/sec):3417. #Correct:688 #Tested:833 Testing Accuracy:82.5%\r",
      "Progress:83.3% Speed(reviews/sec):3418. #Correct:688 #Tested:834 Testing Accuracy:82.4%\r",
      "Progress:83.4% Speed(reviews/sec):3419. #Correct:689 #Tested:835 Testing Accuracy:82.5%\r",
      "Progress:83.5% Speed(reviews/sec):3420. #Correct:690 #Tested:836 Testing Accuracy:82.5%\r",
      "Progress:83.6% Speed(reviews/sec):3419. #Correct:691 #Tested:837 Testing Accuracy:82.5%\r",
      "Progress:83.7% Speed(reviews/sec):3390. #Correct:692 #Tested:838 Testing Accuracy:82.5%\r",
      "Progress:83.8% Speed(reviews/sec):3391. #Correct:692 #Tested:839 Testing Accuracy:82.4%\r",
      "Progress:83.9% Speed(reviews/sec):3390. #Correct:693 #Tested:840 Testing Accuracy:82.5%\r",
      "Progress:84.0% Speed(reviews/sec):3392. #Correct:694 #Tested:841 Testing Accuracy:82.5%\r",
      "Progress:84.1% Speed(reviews/sec):3393. #Correct:695 #Tested:842 Testing Accuracy:82.5%\r",
      "Progress:84.2% Speed(reviews/sec):3395. #Correct:696 #Tested:843 Testing Accuracy:82.5%\r",
      "Progress:84.3% Speed(reviews/sec):3396. #Correct:697 #Tested:844 Testing Accuracy:82.5%\r",
      "Progress:84.4% Speed(reviews/sec):3398. #Correct:698 #Tested:845 Testing Accuracy:82.6%\r",
      "Progress:84.5% Speed(reviews/sec):3399. #Correct:699 #Tested:846 Testing Accuracy:82.6%\r",
      "Progress:84.6% Speed(reviews/sec):3399. #Correct:699 #Tested:847 Testing Accuracy:82.5%\r",
      "Progress:84.7% Speed(reviews/sec):3401. #Correct:699 #Tested:848 Testing Accuracy:82.4%\r",
      "Progress:84.8% Speed(reviews/sec):3403. #Correct:700 #Tested:849 Testing Accuracy:82.4%\r",
      "Progress:84.9% Speed(reviews/sec):3401. #Correct:701 #Tested:850 Testing Accuracy:82.4%\r",
      "Progress:85.0% Speed(reviews/sec):3399. #Correct:702 #Tested:851 Testing Accuracy:82.4%\r",
      "Progress:85.1% Speed(reviews/sec):3400. #Correct:703 #Tested:852 Testing Accuracy:82.5%\r",
      "Progress:85.2% Speed(reviews/sec):3400. #Correct:703 #Tested:853 Testing Accuracy:82.4%\r",
      "Progress:85.3% Speed(reviews/sec):3401. #Correct:704 #Tested:854 Testing Accuracy:82.4%\r",
      "Progress:85.4% Speed(reviews/sec):3398. #Correct:705 #Tested:855 Testing Accuracy:82.4%\r",
      "Progress:85.5% Speed(reviews/sec):3399. #Correct:706 #Tested:856 Testing Accuracy:82.4%\r",
      "Progress:85.6% Speed(reviews/sec):3400. #Correct:707 #Tested:857 Testing Accuracy:82.4%\r",
      "Progress:85.7% Speed(reviews/sec):3400. #Correct:708 #Tested:858 Testing Accuracy:82.5%\r",
      "Progress:85.8% Speed(reviews/sec):3401. #Correct:709 #Tested:859 Testing Accuracy:82.5%\r",
      "Progress:85.9% Speed(reviews/sec):3401. #Correct:710 #Tested:860 Testing Accuracy:82.5%\r",
      "Progress:86.0% Speed(reviews/sec):3403. #Correct:711 #Tested:861 Testing Accuracy:82.5%\r",
      "Progress:86.1% Speed(reviews/sec):3400. #Correct:712 #Tested:862 Testing Accuracy:82.5%\r",
      "Progress:86.2% Speed(reviews/sec):3401. #Correct:712 #Tested:863 Testing Accuracy:82.5%\r",
      "Progress:86.3% Speed(reviews/sec):3398. #Correct:713 #Tested:864 Testing Accuracy:82.5%\r",
      "Progress:86.4% Speed(reviews/sec):3400. #Correct:713 #Tested:865 Testing Accuracy:82.4%\r",
      "Progress:86.5% Speed(reviews/sec):3399. #Correct:714 #Tested:866 Testing Accuracy:82.4%\r",
      "Progress:86.6% Speed(reviews/sec):3401. #Correct:714 #Tested:867 Testing Accuracy:82.3%\r",
      "Progress:86.7% Speed(reviews/sec):3403. #Correct:714 #Tested:868 Testing Accuracy:82.2%\r",
      "Progress:86.8% Speed(reviews/sec):3405. #Correct:715 #Tested:869 Testing Accuracy:82.2%\r",
      "Progress:86.9% Speed(reviews/sec):3402. #Correct:716 #Tested:870 Testing Accuracy:82.2%\r",
      "Progress:87.0% Speed(reviews/sec):3402. #Correct:717 #Tested:871 Testing Accuracy:82.3%\r",
      "Progress:87.1% Speed(reviews/sec):3401. #Correct:718 #Tested:872 Testing Accuracy:82.3%\r",
      "Progress:87.2% Speed(reviews/sec):3397. #Correct:719 #Tested:873 Testing Accuracy:82.3%\r",
      "Progress:87.3% Speed(reviews/sec):3387. #Correct:720 #Tested:874 Testing Accuracy:82.3%\r",
      "Progress:87.4% Speed(reviews/sec):3385. #Correct:721 #Tested:875 Testing Accuracy:82.4%\r",
      "Progress:87.5% Speed(reviews/sec):3385. #Correct:722 #Tested:876 Testing Accuracy:82.4%\r",
      "Progress:87.6% Speed(reviews/sec):3386. #Correct:722 #Tested:877 Testing Accuracy:82.3%\r",
      "Progress:87.7% Speed(reviews/sec):3336. #Correct:723 #Tested:878 Testing Accuracy:82.3%\r",
      "Progress:87.8% Speed(reviews/sec):3295. #Correct:724 #Tested:879 Testing Accuracy:82.3%\r",
      "Progress:87.9% Speed(reviews/sec):3265. #Correct:725 #Tested:880 Testing Accuracy:82.3%\r",
      "Progress:88.0% Speed(reviews/sec):3266. #Correct:726 #Tested:881 Testing Accuracy:82.4%\r",
      "Progress:88.1% Speed(reviews/sec):3268. #Correct:727 #Tested:882 Testing Accuracy:82.4%\r",
      "Progress:88.2% Speed(reviews/sec):3270. #Correct:728 #Tested:883 Testing Accuracy:82.4%\r",
      "Progress:88.3% Speed(reviews/sec):3272. #Correct:729 #Tested:884 Testing Accuracy:82.4%\r",
      "Progress:88.4% Speed(reviews/sec):3274. #Correct:730 #Tested:885 Testing Accuracy:82.4%\r",
      "Progress:88.5% Speed(reviews/sec):3276. #Correct:731 #Tested:886 Testing Accuracy:82.5%\r",
      "Progress:88.6% Speed(reviews/sec):3278. #Correct:731 #Tested:887 Testing Accuracy:82.4%\r",
      "Progress:88.7% Speed(reviews/sec):3280. #Correct:732 #Tested:888 Testing Accuracy:82.4%\r",
      "Progress:88.8% Speed(reviews/sec):3282. #Correct:732 #Tested:889 Testing Accuracy:82.3%\r",
      "Progress:88.9% Speed(reviews/sec):3282. #Correct:733 #Tested:890 Testing Accuracy:82.3%\r",
      "Progress:89.0% Speed(reviews/sec):3284. #Correct:734 #Tested:891 Testing Accuracy:82.3%\r",
      "Progress:89.1% Speed(reviews/sec):3284. #Correct:735 #Tested:892 Testing Accuracy:82.3%\r",
      "Progress:89.2% Speed(reviews/sec):3283. #Correct:735 #Tested:893 Testing Accuracy:82.3%\r",
      "Progress:89.3% Speed(reviews/sec):3284. #Correct:736 #Tested:894 Testing Accuracy:82.3%\r",
      "Progress:89.4% Speed(reviews/sec):3287. #Correct:737 #Tested:895 Testing Accuracy:82.3%\r",
      "Progress:89.5% Speed(reviews/sec):3289. #Correct:738 #Tested:896 Testing Accuracy:82.3%\r",
      "Progress:89.6% Speed(reviews/sec):3290. #Correct:739 #Tested:897 Testing Accuracy:82.3%\r",
      "Progress:89.7% Speed(reviews/sec):3292. #Correct:740 #Tested:898 Testing Accuracy:82.4%\r",
      "Progress:89.8% Speed(reviews/sec):3291. #Correct:741 #Tested:899 Testing Accuracy:82.4%\r",
      "Progress:89.9% Speed(reviews/sec):3291. #Correct:742 #Tested:900 Testing Accuracy:82.4%\r",
      "Progress:90.0% Speed(reviews/sec):3290. #Correct:743 #Tested:901 Testing Accuracy:82.4%\r",
      "Progress:90.1% Speed(reviews/sec):3284. #Correct:744 #Tested:902 Testing Accuracy:82.4%\r",
      "Progress:90.2% Speed(reviews/sec):3284. #Correct:745 #Tested:903 Testing Accuracy:82.5%\r",
      "Progress:90.3% Speed(reviews/sec):3286. #Correct:746 #Tested:904 Testing Accuracy:82.5%\r",
      "Progress:90.4% Speed(reviews/sec):3288. #Correct:747 #Tested:905 Testing Accuracy:82.5%\r",
      "Progress:90.5% Speed(reviews/sec):3290. #Correct:748 #Tested:906 Testing Accuracy:82.5%\r",
      "Progress:90.6% Speed(reviews/sec):3292. #Correct:748 #Tested:907 Testing Accuracy:82.4%\r",
      "Progress:90.7% Speed(reviews/sec):3294. #Correct:749 #Tested:908 Testing Accuracy:82.4%\r",
      "Progress:90.8% Speed(reviews/sec):3294. #Correct:749 #Tested:909 Testing Accuracy:82.3%\r",
      "Progress:90.9% Speed(reviews/sec):3290. #Correct:749 #Tested:910 Testing Accuracy:82.3%\r",
      "Progress:91.0% Speed(reviews/sec):3290. #Correct:750 #Tested:911 Testing Accuracy:82.3%\r",
      "Progress:91.1% Speed(reviews/sec):3291. #Correct:750 #Tested:912 Testing Accuracy:82.2%\r",
      "Progress:91.2% Speed(reviews/sec):3292. #Correct:751 #Tested:913 Testing Accuracy:82.2%\r",
      "Progress:91.3% Speed(reviews/sec):3291. #Correct:751 #Tested:914 Testing Accuracy:82.1%\r",
      "Progress:91.4% Speed(reviews/sec):3293. #Correct:752 #Tested:915 Testing Accuracy:82.1%\r",
      "Progress:91.5% Speed(reviews/sec):3266. #Correct:752 #Tested:916 Testing Accuracy:82.0%\r",
      "Progress:91.6% Speed(reviews/sec):3262. #Correct:752 #Tested:917 Testing Accuracy:82.0%\r",
      "Progress:91.7% Speed(reviews/sec):3262. #Correct:752 #Tested:918 Testing Accuracy:81.9%\r",
      "Progress:91.8% Speed(reviews/sec):3262. #Correct:753 #Tested:919 Testing Accuracy:81.9%\r",
      "Progress:91.9% Speed(reviews/sec):3262. #Correct:754 #Tested:920 Testing Accuracy:81.9%\r",
      "Progress:92.0% Speed(reviews/sec):3255. #Correct:754 #Tested:921 Testing Accuracy:81.8%\r",
      "Progress:92.1% Speed(reviews/sec):3254. #Correct:755 #Tested:922 Testing Accuracy:81.8%\r",
      "Progress:92.2% Speed(reviews/sec):3254. #Correct:756 #Tested:923 Testing Accuracy:81.9%\r",
      "Progress:92.3% Speed(reviews/sec):3255. #Correct:757 #Tested:924 Testing Accuracy:81.9%\r",
      "Progress:92.4% Speed(reviews/sec):3256. #Correct:758 #Tested:925 Testing Accuracy:81.9%\r",
      "Progress:92.5% Speed(reviews/sec):3257. #Correct:759 #Tested:926 Testing Accuracy:81.9%\r",
      "Progress:92.6% Speed(reviews/sec):3257. #Correct:760 #Tested:927 Testing Accuracy:81.9%\r",
      "Progress:92.7% Speed(reviews/sec):3245. #Correct:761 #Tested:928 Testing Accuracy:82.0%\r",
      "Progress:92.8% Speed(reviews/sec):3208. #Correct:761 #Tested:929 Testing Accuracy:81.9%\r",
      "Progress:92.9% Speed(reviews/sec):3190. #Correct:762 #Tested:930 Testing Accuracy:81.9%\r",
      "Progress:93.0% Speed(reviews/sec):3189. #Correct:763 #Tested:931 Testing Accuracy:81.9%\r",
      "Progress:93.1% Speed(reviews/sec):3186. #Correct:764 #Tested:932 Testing Accuracy:81.9%\r",
      "Progress:93.2% Speed(reviews/sec):3187. #Correct:765 #Tested:933 Testing Accuracy:81.9%\r",
      "Progress:93.3% Speed(reviews/sec):3149. #Correct:766 #Tested:934 Testing Accuracy:82.0%\r",
      "Progress:93.4% Speed(reviews/sec):3145. #Correct:767 #Tested:935 Testing Accuracy:82.0%\r",
      "Progress:93.5% Speed(reviews/sec):3146. #Correct:768 #Tested:936 Testing Accuracy:82.0%\r",
      "Progress:93.6% Speed(reviews/sec):3147. #Correct:768 #Tested:937 Testing Accuracy:81.9%\r",
      "Progress:93.7% Speed(reviews/sec):3148. #Correct:769 #Tested:938 Testing Accuracy:81.9%\r",
      "Progress:93.8% Speed(reviews/sec):3150. #Correct:769 #Tested:939 Testing Accuracy:81.8%\r",
      "Progress:93.9% Speed(reviews/sec):3127. #Correct:770 #Tested:940 Testing Accuracy:81.9%\r",
      "Progress:94.0% Speed(reviews/sec):3128. #Correct:771 #Tested:941 Testing Accuracy:81.9%\r",
      "Progress:94.1% Speed(reviews/sec):3122. #Correct:771 #Tested:942 Testing Accuracy:81.8%\r",
      "Progress:94.2% Speed(reviews/sec):3118. #Correct:771 #Tested:943 Testing Accuracy:81.7%\r",
      "Progress:94.3% Speed(reviews/sec):3118. #Correct:772 #Tested:944 Testing Accuracy:81.7%\r",
      "Progress:94.4% Speed(reviews/sec):3093. #Correct:773 #Tested:945 Testing Accuracy:81.7%\r",
      "Progress:94.5% Speed(reviews/sec):3082. #Correct:774 #Tested:946 Testing Accuracy:81.8%\r",
      "Progress:94.6% Speed(reviews/sec):3081. #Correct:775 #Tested:947 Testing Accuracy:81.8%\r",
      "Progress:94.7% Speed(reviews/sec):3079. #Correct:776 #Tested:948 Testing Accuracy:81.8%\r",
      "Progress:94.8% Speed(reviews/sec):3076. #Correct:777 #Tested:949 Testing Accuracy:81.8%\r",
      "Progress:94.9% Speed(reviews/sec):3077. #Correct:778 #Tested:950 Testing Accuracy:81.8%\r",
      "Progress:95.0% Speed(reviews/sec):3067. #Correct:779 #Tested:951 Testing Accuracy:81.9%\r",
      "Progress:95.1% Speed(reviews/sec):3061. #Correct:779 #Tested:952 Testing Accuracy:81.8%\r",
      "Progress:95.2% Speed(reviews/sec):3053. #Correct:779 #Tested:953 Testing Accuracy:81.7%\r",
      "Progress:95.3% Speed(reviews/sec):3053. #Correct:780 #Tested:954 Testing Accuracy:81.7%\r",
      "Progress:95.4% Speed(reviews/sec):3024. #Correct:781 #Tested:955 Testing Accuracy:81.7%\r",
      "Progress:95.5% Speed(reviews/sec):3021. #Correct:782 #Tested:956 Testing Accuracy:81.7%\r",
      "Progress:95.6% Speed(reviews/sec):3021. #Correct:783 #Tested:957 Testing Accuracy:81.8%\r",
      "Progress:95.7% Speed(reviews/sec):3022. #Correct:784 #Tested:958 Testing Accuracy:81.8%\r",
      "Progress:95.8% Speed(reviews/sec):3021. #Correct:785 #Tested:959 Testing Accuracy:81.8%\r",
      "Progress:95.9% Speed(reviews/sec):3023. #Correct:786 #Tested:960 Testing Accuracy:81.8%\r",
      "Progress:96.0% Speed(reviews/sec):3022. #Correct:787 #Tested:961 Testing Accuracy:81.8%\r",
      "Progress:96.1% Speed(reviews/sec):3020. #Correct:788 #Tested:962 Testing Accuracy:81.9%\r",
      "Progress:96.2% Speed(reviews/sec):3022. #Correct:789 #Tested:963 Testing Accuracy:81.9%\r",
      "Progress:96.3% Speed(reviews/sec):3023. #Correct:790 #Tested:964 Testing Accuracy:81.9%\r",
      "Progress:96.4% Speed(reviews/sec):3025. #Correct:791 #Tested:965 Testing Accuracy:81.9%\r",
      "Progress:96.5% Speed(reviews/sec):3027. #Correct:792 #Tested:966 Testing Accuracy:81.9%\r",
      "Progress:96.6% Speed(reviews/sec):3029. #Correct:793 #Tested:967 Testing Accuracy:82.0%\r",
      "Progress:96.7% Speed(reviews/sec):3031. #Correct:794 #Tested:968 Testing Accuracy:82.0%\r",
      "Progress:96.8% Speed(reviews/sec):3033. #Correct:795 #Tested:969 Testing Accuracy:82.0%\r",
      "Progress:96.9% Speed(reviews/sec):3034. #Correct:796 #Tested:970 Testing Accuracy:82.0%\r",
      "Progress:97.0% Speed(reviews/sec):3034. #Correct:797 #Tested:971 Testing Accuracy:82.0%\r",
      "Progress:97.1% Speed(reviews/sec):3034. #Correct:798 #Tested:972 Testing Accuracy:82.0%\r",
      "Progress:97.2% Speed(reviews/sec):3036. #Correct:798 #Tested:973 Testing Accuracy:82.0%\r",
      "Progress:97.3% Speed(reviews/sec):3037. #Correct:799 #Tested:974 Testing Accuracy:82.0%\r",
      "Progress:97.4% Speed(reviews/sec):3038. #Correct:800 #Tested:975 Testing Accuracy:82.0%\r",
      "Progress:97.5% Speed(reviews/sec):3039. #Correct:801 #Tested:976 Testing Accuracy:82.0%\r",
      "Progress:97.6% Speed(reviews/sec):3041. #Correct:802 #Tested:977 Testing Accuracy:82.0%\r",
      "Progress:97.7% Speed(reviews/sec):3041. #Correct:803 #Tested:978 Testing Accuracy:82.1%\r",
      "Progress:97.8% Speed(reviews/sec):3042. #Correct:804 #Tested:979 Testing Accuracy:82.1%\r",
      "Progress:97.9% Speed(reviews/sec):3043. #Correct:805 #Tested:980 Testing Accuracy:82.1%\r",
      "Progress:98.0% Speed(reviews/sec):3032. #Correct:806 #Tested:981 Testing Accuracy:82.1%\r",
      "Progress:98.1% Speed(reviews/sec):3029. #Correct:807 #Tested:982 Testing Accuracy:82.1%\r",
      "Progress:98.2% Speed(reviews/sec):3028. #Correct:808 #Tested:983 Testing Accuracy:82.1%\r",
      "Progress:98.3% Speed(reviews/sec):3025. #Correct:809 #Tested:984 Testing Accuracy:82.2%\r",
      "Progress:98.4% Speed(reviews/sec):3027. #Correct:809 #Tested:985 Testing Accuracy:82.1%\r",
      "Progress:98.5% Speed(reviews/sec):3029. #Correct:810 #Tested:986 Testing Accuracy:82.1%\r",
      "Progress:98.6% Speed(reviews/sec):3031. #Correct:811 #Tested:987 Testing Accuracy:82.1%\r",
      "Progress:98.7% Speed(reviews/sec):3032. #Correct:812 #Tested:988 Testing Accuracy:82.1%\r",
      "Progress:98.8% Speed(reviews/sec):3034. #Correct:813 #Tested:989 Testing Accuracy:82.2%\r",
      "Progress:98.9% Speed(reviews/sec):3033. #Correct:814 #Tested:990 Testing Accuracy:82.2%\r",
      "Progress:99.0% Speed(reviews/sec):3033. #Correct:815 #Tested:991 Testing Accuracy:82.2%\r",
      "Progress:99.1% Speed(reviews/sec):3034. #Correct:816 #Tested:992 Testing Accuracy:82.2%\r",
      "Progress:99.2% Speed(reviews/sec):3036. #Correct:817 #Tested:993 Testing Accuracy:82.2%\r",
      "Progress:99.3% Speed(reviews/sec):3038. #Correct:818 #Tested:994 Testing Accuracy:82.2%\r",
      "Progress:99.4% Speed(reviews/sec):3037. #Correct:818 #Tested:995 Testing Accuracy:82.2%\r",
      "Progress:99.5% Speed(reviews/sec):3037. #Correct:819 #Tested:996 Testing Accuracy:82.2%\r",
      "Progress:99.6% Speed(reviews/sec):3039. #Correct:820 #Tested:997 Testing Accuracy:82.2%\r",
      "Progress:99.7% Speed(reviews/sec):3040. #Correct:821 #Tested:998 Testing Accuracy:82.2%\r",
      "Progress:99.8% Speed(reviews/sec):3039. #Correct:821 #Tested:999 Testing Accuracy:82.1%\r",
      "Progress:99.9% Speed(reviews/sec):3040. #Correct:822 #Tested:1000 Testing Accuracy:82.2%"
     ]
    }
   ],
   "source": [
    "mlp.test(reviews[-1000:],labels[-1000:])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 项目 6 到此结束。\n",
    "\n",
    "# 分析：权重怎么了？<a id='lesson_7'></a>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {},
   "outputs": [],
   "source": [
    "mlp_full = SentimentNetwork(reviews[:-1000],labels[:-1000],min_count=0,polarity_cutoff=0,learning_rate=0.01)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Progress:0.0% Speed(reviews/sec):0.0 #Correct:1 #Trained:1 Training Accuracy:100.%\n",
      "Progress:10.4% Speed(reviews/sec):1120. #Correct:1962 #Trained:2501 Training Accuracy:78.4%\n",
      "Progress:20.8% Speed(reviews/sec):1084. #Correct:4002 #Trained:5001 Training Accuracy:80.0%\n",
      "Progress:31.2% Speed(reviews/sec):1086. #Correct:6120 #Trained:7501 Training Accuracy:81.5%\n",
      "Progress:41.6% Speed(reviews/sec):1094. #Correct:8271 #Trained:10001 Training Accuracy:82.7%\n",
      "Progress:52.0% Speed(reviews/sec):1095. #Correct:10431 #Trained:12501 Training Accuracy:83.4%\n",
      "Progress:62.5% Speed(reviews/sec):1088. #Correct:12565 #Trained:15001 Training Accuracy:83.7%\n",
      "Progress:72.9% Speed(reviews/sec):1078. #Correct:14670 #Trained:17501 Training Accuracy:83.8%\n",
      "Progress:83.3% Speed(reviews/sec):1077. #Correct:16833 #Trained:20001 Training Accuracy:84.1%\n",
      "Progress:93.7% Speed(reviews/sec):1076. #Correct:19015 #Trained:22501 Training Accuracy:84.5%\n",
      "Progress:97.9% Speed(reviews/sec):1077. #Correct:19880 #Trained:23501 Training Accuracy:84.5%"
     ]
    }
   ],
   "source": [
    "mlp_full.train(reviews[:-1000],labels[:-1000])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 80,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Image(filename='sentiment_network_sparse.png')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_most_similar_words(focus = \"horrible\"):\n",
    "    most_similar = Counter()\n",
    "\n",
    "    for word in mlp_full.word2index.keys():\n",
    "        most_similar[word] = np.dot(mlp_full.weights_0_1[mlp_full.word2index[word]],mlp_full.weights_0_1[mlp_full.word2index[focus]])\n",
    "    \n",
    "    return most_similar.most_common()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[('excellent', 0.13672950757352476),\n",
       " ('perfect', 0.12548286087225946),\n",
       " ('amazing', 0.09182763392599974),\n",
       " ('today', 0.09022366269441423),\n",
       " ('wonderful', 0.0893559769622146),\n",
       " ('fun', 0.08750446667420686),\n",
       " ('great', 0.08714175888229203),\n",
       " ('best', 0.08581088561788065),\n",
       " ('liked', 0.07769762912384343),\n",
       " ('definitely', 0.07662878140696602),\n",
       " ('brilliant', 0.07342385876927905),\n",
       " ('loved', 0.07328542892812215),\n",
       " ('favorite', 0.07278113603616078),\n",
       " ('superb', 0.07173620717850507),\n",
       " ('fantastic', 0.0709221919162662),\n",
       " ('job', 0.06916061720763408),\n",
       " ('incredible', 0.06642407795261443),\n",
       " ('enjoyable', 0.06563256050288879),\n",
       " ('rare', 0.06481921266261509),\n",
       " ('highly', 0.06388945335097054),\n",
       " ('enjoyed', 0.06212754610181293),\n",
       " ('wonderfully', 0.06205517860409017),\n",
       " ('perfectly', 0.061093208811887394),\n",
       " ('fascinating', 0.0606635479374939),\n",
       " ('bit', 0.05965542704565305),\n",
       " ('gem', 0.05951085929615678),\n",
       " ('outstanding', 0.05886080814708302),\n",
       " ('beautiful', 0.05861393470316208),\n",
       " ('surprised', 0.058273314482562975),\n",
       " ('worth', 0.05765748423647123)]"
      ]
     },
     "execution_count": 90,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "get_most_similar_words(\"excellent\")[0:30]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[('worst', 0.16966107259049842),\n",
       " ('awful', 0.12026847019691243),\n",
       " ('waste', 0.11945367265311004),\n",
       " ('poor', 0.0927588875744355),\n",
       " ('terrible', 0.09142538719772794),\n",
       " ('dull', 0.08420927167822358),\n",
       " ('poorly', 0.08124154451604203),\n",
       " ('disappointment', 0.08006475962136872),\n",
       " ('fails', 0.0785997737233375),\n",
       " ('disappointing', 0.07733948548032334),\n",
       " ('boring', 0.07712785874801287),\n",
       " ('unfortunately', 0.07550244970585908),\n",
       " ('worse', 0.07060183536419468),\n",
       " ('mess', 0.0705642996235904),\n",
       " ('stupid', 0.06948482283254304),\n",
       " ('badly', 0.06688890366622857),\n",
       " ('annoying', 0.06568702190337415),\n",
       " ('bad', 0.06309381453757214),\n",
       " ('save', 0.06288059749586573),\n",
       " ('disappointed', 0.06269235381207289),\n",
       " ('wasted', 0.061387183028051275),\n",
       " ('supposed', 0.06098545295772515),\n",
       " ('horrible', 0.06012177233938014),\n",
       " ('laughable', 0.058698406285467644),\n",
       " ('crap', 0.05810452866788458),\n",
       " ('basically', 0.057218840369636155),\n",
       " ('nothing', 0.05715822004303418),\n",
       " ('ridiculous', 0.05690548106893144),\n",
       " ('lacks', 0.055766565889465436),\n",
       " ('lame', 0.05561600905811018)]"
      ]
     },
     "execution_count": 91,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "get_most_similar_words(\"terrible\")[0:30]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.colors as colors\n",
    "words_to_visualize = list()\n",
    "for word, ratio in pos_neg_ratios.most_common(500):\n",
    "    if(word in mlp_full.word2index.keys()):\n",
    "        words_to_visualize.append(word)\n",
    "    \n",
    "for word, ratio in list(reversed(pos_neg_ratios.most_common()))[0:500]:\n",
    "    if(word in mlp_full.word2index.keys()):\n",
    "        words_to_visualize.append(word)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "metadata": {},
   "outputs": [],
   "source": [
    "pos = 0\n",
    "neg = 0\n",
    "\n",
    "colors_list = list()\n",
    "vectors_list = list()\n",
    "for word in words_to_visualize:\n",
    "    if word in pos_neg_ratios.keys():\n",
    "        vectors_list.append(mlp_full.weights_0_1[mlp_full.word2index[word]])\n",
    "        if(pos_neg_ratios[word] > 0):\n",
    "            pos+=1\n",
    "            colors_list.append(\"#00ff00\")\n",
    "        else:\n",
    "            neg+=1\n",
    "            colors_list.append(\"#000000\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.manifold import TSNE\n",
    "tsne = TSNE(n_components=2, random_state=0)\n",
    "words_top_ted_tsne = tsne.fit_transform(vectors_list)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "\n",
       "\n",
       "\n",
       "\n",
       "\n",
       "\n",
       "  <div class=\"bk-root\" id=\"5e01973a-ccf9-405d-92fe-53fa8cf91795\" data-root-id=\"1178\"></div>\n"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/javascript": [
       "(function(root) {\n",
       "  function embed_document(root) {\n",
       "    \n",
       "  var docs_json = {\"a8bbecdd-d0a4-4e01-afd8-6742118f875e\":{\"roots\":{\"references\":[{\"attributes\":{\"below\":[{\"id\":\"1188\",\"type\":\"LinearAxis\"}],\"left\":[{\"id\":\"1193\",\"type\":\"LinearAxis\"}],\"renderers\":[{\"id\":\"1188\",\"type\":\"LinearAxis\"},{\"id\":\"1192\",\"type\":\"Grid\"},{\"id\":\"1193\",\"type\":\"LinearAxis\"},{\"id\":\"1197\",\"type\":\"Grid\"},{\"id\":\"1211\",\"type\":\"GlyphRenderer\"},{\"id\":\"1213\",\"type\":\"LabelSet\"}],\"title\":{\"id\":\"1177\",\"type\":\"Title\"},\"toolbar\":{\"id\":\"1202\",\"type\":\"Toolbar\"},\"toolbar_location\":\"above\",\"x_range\":{\"id\":\"1180\",\"type\":\"DataRange1d\"},\"x_scale\":{\"id\":\"1184\",\"type\":\"LinearScale\"},\"y_range\":{\"id\":\"1182\",\"type\":\"DataRange1d\"},\"y_scale\":{\"id\":\"1186\",\"type\":\"LinearScale\"}},\"id\":\"1178\",\"subtype\":\"Figure\",\"type\":\"Plot\"},{\"attributes\":{},\"id\":\"1199\",\"type\":\"WheelZoomTool\"},{\"attributes\":{\"callback\":null,\"data\":{\"color\":[\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#00ff00\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\",\"#000000\"],\"names\":[\"edie\",\"paulie\",\"felix\",\"polanski\",\"matthau\",\"victoria\",\"mildred\",\"gandhi\",\"flawless\",\"superbly\",\"perfection\",\"astaire\",\"captures\",\"voight\",\"wonderfully\",\"powell\",\"brosnan\",\"lily\",\"bakshi\",\"lincoln\",\"refreshing\",\"breathtaking\",\"bourne\",\"lemmon\",\"delightful\",\"flynn\",\"andrews\",\"homer\",\"beautifully\",\"soccer\",\"elvira\",\"underrated\",\"gripping\",\"superb\",\"delight\",\"welles\",\"sadness\",\"sinatra\",\"touching\",\"timeless\",\"macy\",\"unforgettable\",\"favorites\",\"stewart\",\"sullivan\",\"extraordinary\",\"hartley\",\"brilliantly\",\"friendship\",\"wonderful\",\"palma\",\"magnificent\",\"finest\",\"jackie\",\"ritter\",\"tremendous\",\"freedom\",\"fantastic\",\"terrific\",\"noir\",\"sidney\",\"outstanding\",\"pleasantly\",\"mann\",\"nancy\",\"marie\",\"marvelous\",\"excellent\",\"ruth\",\"stanwyck\",\"widmark\",\"splendid\",\"chan\",\"exceptional\",\"tender\",\"gentle\",\"poignant\",\"gem\",\"amazing\",\"chilling\",\"fisher\",\"davies\",\"captivating\",\"darker\",\"april\",\"kelly\",\"blake\",\"overlooked\",\"ralph\",\"bette\",\"hoffman\",\"cole\",\"shines\",\"powerful\",\"notch\",\"remarkable\",\"pitt\",\"winters\",\"vivid\",\"gritty\",\"giallo\",\"portrait\",\"innocence\",\"psychiatrist\",\"favorite\",\"ensemble\",\"stunning\",\"burns\",\"garbo\",\"barbara\",\"philip\",\"panic\",\"holly\",\"carol\",\"perfect\",\"appreciated\",\"favourite\",\"journey\",\"rural\",\"bond\",\"builds\",\"brilliant\",\"brooklyn\",\"von\",\"recommended\",\"unfolds\",\"daniel\",\"perfectly\",\"crafted\",\"prince\",\"troubled\",\"consequences\",\"haunting\",\"cinderella\",\"alexander\",\"emotions\",\"boxing\",\"subtle\",\"curtis\",\"rare\",\"loved\",\"daughters\",\"courage\",\"dentist\",\"highly\",\"nominated\",\"tony\",\"draws\",\"everyday\",\"contrast\",\"cried\",\"fabulous\",\"ned\",\"fay\",\"emma\",\"sensitive\",\"smooth\",\"dramas\",\"today\",\"helps\",\"inspiring\",\"jimmy\",\"awesome\",\"unique\",\"tragic\",\"intense\",\"stellar\",\"rival\",\"provides\",\"depression\",\"shy\",\"carrie\",\"blend\",\"hank\",\"diana\",\"adorable\",\"unexpected\",\"achievement\",\"bettie\",\"happiness\",\"glorious\",\"davis\",\"terrifying\",\"beauty\",\"ideal\",\"fears\",\"hong\",\"seasons\",\"fascinating\",\"carries\",\"satisfying\",\"definite\",\"touched\",\"greatest\",\"creates\",\"aunt\",\"walter\",\"spectacular\",\"portrayal\",\"ann\",\"enterprise\",\"musicals\",\"deeply\",\"incredible\",\"mature\",\"triumph\",\"margaret\",\"navy\",\"harry\",\"lucas\",\"sweet\",\"joey\",\"oscar\",\"balance\",\"warm\",\"ages\",\"guilt\",\"glover\",\"carrey\",\"learns\",\"unusual\",\"sons\",\"complex\",\"essence\",\"brazil\",\"widow\",\"solid\",\"beautiful\",\"holmes\",\"awe\",\"vhs\",\"eerie\",\"lonely\",\"grim\",\"sport\",\"debut\",\"destiny\",\"thrillers\",\"tears\",\"rose\",\"feelings\",\"ginger\",\"winning\",\"stanley\",\"cox\",\"paris\",\"heart\",\"hooked\",\"comfortable\",\"mgm\",\"masterpiece\",\"themes\",\"danny\",\"anime\",\"perry\",\"joy\",\"lovable\",\"mysteries\",\"hal\",\"louis\",\"charming\",\"urban\",\"allows\",\"impact\",\"italy\",\"gradually\",\"lifestyle\",\"spy\",\"treat\",\"subsequent\",\"kennedy\",\"loving\",\"surprising\",\"quiet\",\"winter\",\"reveals\",\"raw\",\"funniest\",\"pleased\",\"norman\",\"thief\",\"season\",\"secrets\",\"colorful\",\"highest\",\"compelling\",\"danes\",\"castle\",\"kudos\",\"great\",\"baseball\",\"subtitles\",\"bleak\",\"winner\",\"tragedy\",\"todd\",\"nicely\",\"arthur\",\"essential\",\"gorgeous\",\"fonda\",\"eastwood\",\"focuses\",\"enjoyed\",\"natural\",\"intensity\",\"witty\",\"rob\",\"worlds\",\"health\",\"magical\",\"deeper\",\"lucy\",\"moving\",\"lovely\",\"purple\",\"memorable\",\"sings\",\"craig\",\"modesty\",\"relate\",\"episodes\",\"strong\",\"smith\",\"tear\",\"apartment\",\"princess\",\"disagree\",\"kung\",\"adventure\",\"columbo\",\"jake\",\"adds\",\"hart\",\"strength\",\"realizes\",\"dave\",\"childhood\",\"forbidden\",\"tight\",\"surreal\",\"manager\",\"dancer\",\"studios\",\"con\",\"miike\",\"realistic\",\"explicit\",\"kurt\",\"traditional\",\"deals\",\"holds\",\"carl\",\"touches\",\"gene\",\"albert\",\"abc\",\"cry\",\"sides\",\"develops\",\"eyre\",\"dances\",\"oscars\",\"legendary\",\"hearted\",\"importance\",\"portraying\",\"impressed\",\"waters\",\"empire\",\"edge\",\"jean\",\"environment\",\"sentimental\",\"captured\",\"styles\",\"daring\",\"frank\",\"tense\",\"backgrounds\",\"matches\",\"gothic\",\"sharp\",\"achieved\",\"court\",\"steals\",\"rules\",\"colors\",\"reunion\",\"covers\",\"tale\",\"rain\",\"denzel\",\"stays\",\"blob\",\"maria\",\"conventional\",\"fresh\",\"midnight\",\"landscape\",\"animated\",\"titanic\",\"sunday\",\"spring\",\"cagney\",\"enjoyable\",\"immensely\",\"sir\",\"nevertheless\",\"driven\",\"performances\",\"memories\",\"nowadays\",\"simple\",\"golden\",\"leslie\",\"lovers\",\"relationship\",\"supporting\",\"che\",\"packed\",\"trek\",\"provoking\",\"strikes\",\"depiction\",\"emotional\",\"secretary\",\"influenced\",\"florida\",\"germany\",\"brings\",\"lewis\",\"elderly\",\"owner\",\"streets\",\"henry\",\"portrays\",\"bears\",\"china\",\"anger\",\"society\",\"available\",\"best\",\"bugs\",\"magic\",\"delivers\",\"verhoeven\",\"jim\",\"donald\",\"endearing\",\"relationships\",\"greatly\",\"charlie\",\"brad\",\"simon\",\"effectively\",\"march\",\"atmosphere\",\"influence\",\"genius\",\"emotionally\",\"ken\",\"identity\",\"sophisticated\",\"dan\",\"andrew\",\"india\",\"roy\",\"surprisingly\",\"sky\",\"romantic\",\"match\",\"meets\",\"cowboy\",\"wave\",\"bitter\",\"patient\",\"stylish\",\"britain\",\"affected\",\"beatty\",\"love\",\"paul\",\"andy\",\"performance\",\"patrick\",\"unlike\",\"brooks\",\"refuses\",\"award\",\"complaint\",\"ride\",\"dawson\",\"luke\",\"wells\",\"france\",\"sports\",\"handsome\",\"directs\",\"rebel\",\"boll\",\"uwe\",\"seagal\",\"unwatchable\",\"stinker\",\"mst\",\"incoherent\",\"unfunny\",\"waste\",\"blah\",\"horrid\",\"pointless\",\"atrocious\",\"redeeming\",\"prom\",\"drivel\",\"lousy\",\"worst\",\"laughable\",\"awful\",\"poorly\",\"wasting\",\"remotely\",\"existent\",\"boredom\",\"miserably\",\"sucks\",\"uninspired\",\"lame\",\"insult\",\"godzilla\",\"uninteresting\",\"gadget\",\"appalling\",\"unconvincing\",\"unintentional\",\"horrible\",\"amateurish\",\"pathetic\",\"idiotic\",\"stupidity\",\"cardboard\",\"wasted\",\"crap\",\"insulting\",\"tedious\",\"dreadful\",\"dire\",\"badly\",\"suck\",\"worse\",\"terrible\",\"embarrassing\",\"mess\",\"garbage\",\"pile\",\"stupid\",\"ashamed\",\"vampires\",\"worthless\",\"dull\",\"inept\",\"avoid\",\"wooden\",\"forgettable\",\"fulci\",\"crappy\",\"bat\",\"unbelievably\",\"whatsoever\",\"excuse\",\"rubbish\",\"ridiculous\",\"junk\",\"flop\",\"boring\",\"turkey\",\"shark\",\"topless\",\"ridiculously\",\"useless\",\"seed\",\"ripped\",\"embarrassed\",\"rambo\",\"costs\",\"hideous\",\"horrendous\",\"bother\",\"dumb\",\"disjointed\",\"plastic\",\"horribly\",\"fest\",\"ludicrous\",\"unintentionally\",\"obnoxious\",\"mildly\",\"bland\",\"mummy\",\"annoying\",\"amateur\",\"bad\",\"dinosaurs\",\"unless\",\"fails\",\"mediocre\",\"awake\",\"clichd\",\"clich\",\"meaningless\",\"disappointment\",\"zombies\",\"asleep\",\"miscast\",\"irritating\",\"utter\",\"disappointing\",\"screaming\",\"supposed\",\"kidding\",\"poor\",\"apes\",\"unbelievable\",\"fake\",\"dude\",\"dracula\",\"joke\",\"clumsy\",\"random\",\"cheap\",\"idiots\",\"devoid\",\"trite\",\"wannabe\",\"unbearable\",\"alright\",\"pretentious\",\"scooby\",\"sucked\",\"senseless\",\"bo\",\"bin\",\"coherent\",\"idiot\",\"toilet\",\"doo\",\"werewolf\",\"cabin\",\"generous\",\"offensive\",\"monkey\",\"painfully\",\"renting\",\"lazy\",\"disgusting\",\"blame\",\"walked\",\"seconds\",\"generic\",\"cheese\",\"sloppy\",\"huh\",\"retarded\",\"trash\",\"shelf\",\"ugly\",\"oh\",\"slightest\",\"explanation\",\"failed\",\"cringe\",\"blatant\",\"clue\",\"bored\",\"cgi\",\"sat\",\"paid\",\"warn\",\"painful\",\"nowhere\",\"bore\",\"absurd\",\"flies\",\"paint\",\"porn\",\"paper\",\"predictable\",\"pseudo\",\"repetitive\",\"outer\",\"brain\",\"sorry\",\"vampire\",\"motivation\",\"unrealistic\",\"wrestling\",\"overrated\",\"aliens\",\"halfway\",\"save\",\"santa\",\"security\",\"contrived\",\"lacks\",\"whale\",\"gore\",\"bunch\",\"hype\",\"flat\",\"noise\",\"below\",\"plain\",\"spending\",\"bothered\",\"annoyed\",\"sounded\",\"honestly\",\"minutes\",\"wreck\",\"lesbian\",\"chick\",\"dollar\",\"f\",\"secondly\",\"wanna\",\"rat\",\"errors\",\"shallow\",\"synopsis\",\"breasts\",\"gray\",\"yeah\",\"nonsense\",\"unnecessary\",\"swear\",\"grave\",\"ruined\",\"somebody\",\"elvis\",\"mindless\",\"terribly\",\"continuity\",\"hoping\",\"ha\",\"nudity\",\"endless\",\"decent\",\"torture\",\"rented\",\"disaster\",\"downright\",\"ok\",\"fat\",\"unpleasant\",\"figured\",\"rip\",\"throwing\",\"attempt\",\"weak\",\"slap\",\"jesus\",\"christian\",\"barely\",\"apparently\",\"implausible\",\"nothing\",\"clichs\",\"credibility\",\"bible\",\"explained\",\"presumably\",\"celluloid\",\"couldn\",\"money\",\"snake\",\"hollow\",\"load\",\"sake\",\"total\",\"priest\",\"supposedly\",\"consists\",\"zombie\",\"bomb\",\"ape\",\"bottom\",\"christ\",\"unfortunately\",\"bullets\",\"grade\",\"drags\",\"freak\",\"wolf\",\"fx\",\"offended\",\"script\",\"raped\",\"producers\",\"okay\",\"confusing\",\"stomach\",\"monster\",\"seriously\",\"alas\",\"promising\",\"knife\",\"substance\",\"premise\",\"threw\",\"k\",\"dear\",\"z\",\"write\",\"rental\",\"warned\",\"zero\",\"semi\",\"guess\",\"scientist\",\"logic\",\"vague\",\"slasher\",\"throw\",\"accents\",\"alien\",\"silly\",\"clown\",\"skip\",\"instead\",\"blank\",\"throat\",\"lab\",\"par\",\"gag\",\"execution\",\"nose\",\"hated\",\"effort\",\"shoot\",\"fill\",\"gratuitous\",\"burn\",\"none\",\"cameras\",\"assume\",\"stick\",\"reasonable\",\"failure\",\"pie\",\"rent\",\"dubbing\",\"weren\",\"truck\",\"stock\",\"thin\",\"daddy\",\"holy\",\"exercise\",\"pg\",\"arm\",\"tried\",\"suppose\",\"advice\",\"gonna\",\"disbelief\",\"derek\",\"mean\",\"merit\",\"looked\",\"channel\",\"gross\",\"stereotypical\",\"hoped\",\"lacking\",\"spent\",\"stiff\",\"overdone\",\"low\",\"romero\",\"hour\",\"blair\",\"saved\",\"damage\",\"reason\",\"intentions\",\"sentence\",\"hardcore\",\"makeup\",\"lack\",\"makers\",\"empty\",\"holes\",\"wouldn\",\"proof\",\"demon\",\"toys\",\"doll\",\"utterly\",\"originality\",\"bush\",\"saying\",\"cover\",\"meat\",\"forest\",\"deserve\",\"sum\",\"bucks\",\"hills\",\"watchable\",\"lacked\",\"handed\",\"mistake\",\"please\",\"whoever\",\"sadistic\",\"monsters\",\"screenwriter\",\"neither\",\"nuclear\",\"sequels\",\"flesh\",\"lying\",\"creature\",\"annie\",\"propaganda\",\"leonard\",\"thats\",\"racist\",\"convince\",\"asian\",\"why\",\"rex\",\"satan\",\"remake\",\"fail\",\"ah\",\"loser\",\"favor\",\"except\",\"flick\",\"freddy\",\"relies\",\"spare\",\"dialog\",\"lou\",\"dragged\",\"guy\",\"problem\",\"melting\",\"flash\",\"im\",\"least\",\"mouth\",\"sole\",\"hell\",\"jerk\",\"drink\",\"intent\",\"shower\",\"fifteen\",\"wasn\",\"thugs\",\"corpse\",\"virus\",\"idea\",\"budget\",\"minimal\",\"reasonably\",\"naked\",\"rick\",\"category\",\"cheesy\",\"judging\",\"half\",\"pregnant\",\"no\",\"millions\",\"stereotypes\",\"juvenile\",\"weekend\",\"convoluted\",\"laurel\",\"killings\",\"sequel\",\"hire\",\"somewhere\",\"frankly\",\"paying\",\"someone\",\"cant\",\"cash\",\"research\",\"dimensional\",\"walk\",\"editing\",\"conceived\",\"scare\",\"positive\",\"anything\"],\"x1\":{\"__ndarray__\":\"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\",\"dtype\":\"float32\",\"shape\":[1000]},\"x2\":{\"__ndarray__\":\"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\",\"dtype\":\"float32\",\"shape\":[1000]}},\"selected\":{\"id\":\"1234\",\"type\":\"Selection\"},\"selection_policy\":{\"id\":\"1235\",\"type\":\"UnionRenderers\"}},\"id\":\"1207\",\"type\":\"ColumnDataSource\"},{\"attributes\":{},\"id\":\"1200\",\"type\":\"ResetTool\"},{\"attributes\":{\"fill_color\":{\"field\":\"color\"},\"line_color\":{\"value\":\"#1f77b4\"},\"size\":{\"units\":\"screen\",\"value\":8},\"x\":{\"field\":\"x1\"},\"y\":{\"field\":\"x2\"}},\"id\":\"1209\",\"type\":\"Scatter\"},{\"attributes\":{\"plot\":null,\"text\":\"vector T-SNE for most polarized words\"},\"id\":\"1177\",\"type\":\"Title\"},{\"attributes\":{},\"id\":\"1201\",\"type\":\"SaveTool\"},{\"attributes\":{},\"id\":\"1230\",\"type\":\"BasicTickFormatter\"},{\"attributes\":{\"callback\":null},\"id\":\"1180\",\"type\":\"DataRange1d\"},{\"attributes\":{\"fill_alpha\":{\"value\":0.1},\"fill_color\":{\"value\":\"#1f77b4\"},\"line_alpha\":{\"value\":0.1},\"line_color\":{\"value\":\"#1f77b4\"},\"size\":{\"units\":\"screen\",\"value\":8},\"x\":{\"field\":\"x1\"},\"y\":{\"field\":\"x2\"}},\"id\":\"1210\",\"type\":\"Scatter\"},{\"attributes\":{\"active_drag\":\"auto\",\"active_inspect\":\"auto\",\"active_multi\":null,\"active_scroll\":\"auto\",\"active_tap\":\"auto\",\"tools\":[{\"id\":\"1198\",\"type\":\"PanTool\"},{\"id\":\"1199\",\"type\":\"WheelZoomTool\"},{\"id\":\"1200\",\"type\":\"ResetTool\"},{\"id\":\"1201\",\"type\":\"SaveTool\"}]},\"id\":\"1202\",\"type\":\"Toolbar\"},{\"attributes\":{\"callback\":null},\"id\":\"1182\",\"type\":\"DataRange1d\"},{\"attributes\":{\"data_source\":{\"id\":\"1207\",\"type\":\"ColumnDataSource\"},\"glyph\":{\"id\":\"1209\",\"type\":\"Scatter\"},\"hover_glyph\":null,\"muted_glyph\":null,\"nonselection_glyph\":{\"id\":\"1210\",\"type\":\"Scatter\"},\"selection_glyph\":null,\"view\":{\"id\":\"1212\",\"type\":\"CDSView\"}},\"id\":\"1211\",\"type\":\"GlyphRenderer\"},{\"attributes\":{},\"id\":\"1184\",\"type\":\"LinearScale\"},{\"attributes\":{\"source\":{\"id\":\"1207\",\"type\":\"ColumnDataSource\"}},\"id\":\"1212\",\"type\":\"CDSView\"},{\"attributes\":{},\"id\":\"1186\",\"type\":\"LinearScale\"},{\"attributes\":{\"formatter\":{\"id\":\"1230\",\"type\":\"BasicTickFormatter\"},\"plot\":{\"id\":\"1178\",\"subtype\":\"Figure\",\"type\":\"Plot\"},\"ticker\":{\"id\":\"1189\",\"type\":\"BasicTicker\"}},\"id\":\"1188\",\"type\":\"LinearAxis\"},{\"attributes\":{},\"id\":\"1189\",\"type\":\"BasicTicker\"},{\"attributes\":{\"plot\":{\"id\":\"1178\",\"subtype\":\"Figure\",\"type\":\"Plot\"},\"ticker\":{\"id\":\"1189\",\"type\":\"BasicTicker\"}},\"id\":\"1192\",\"type\":\"Grid\"},{\"attributes\":{},\"id\":\"1234\",\"type\":\"Selection\"},{\"attributes\":{},\"id\":\"1235\",\"type\":\"UnionRenderers\"},{\"attributes\":{},\"id\":\"1194\",\"type\":\"BasicTicker\"},{\"attributes\":{\"dimension\":1,\"plot\":{\"id\":\"1178\",\"subtype\":\"Figure\",\"type\":\"Plot\"},\"ticker\":{\"id\":\"1194\",\"type\":\"BasicTicker\"}},\"id\":\"1197\",\"type\":\"Grid\"},{\"attributes\":{\"formatter\":{\"id\":\"1232\",\"type\":\"BasicTickFormatter\"},\"plot\":{\"id\":\"1178\",\"subtype\":\"Figure\",\"type\":\"Plot\"},\"ticker\":{\"id\":\"1194\",\"type\":\"BasicTicker\"}},\"id\":\"1193\",\"type\":\"LinearAxis\"},{\"attributes\":{\"plot\":{\"id\":\"1178\",\"subtype\":\"Figure\",\"type\":\"Plot\"},\"source\":{\"id\":\"1207\",\"type\":\"ColumnDataSource\"},\"text\":{\"field\":\"names\"},\"text_align\":\"center\",\"text_color\":{\"value\":\"#555555\"},\"text_font_size\":{\"value\":\"8pt\"},\"x\":{\"field\":\"x1\"},\"y\":{\"field\":\"x2\"},\"y_offset\":{\"value\":6}},\"id\":\"1213\",\"type\":\"LabelSet\"},{\"attributes\":{},\"id\":\"1232\",\"type\":\"BasicTickFormatter\"},{\"attributes\":{},\"id\":\"1198\",\"type\":\"PanTool\"}],\"root_ids\":[\"1178\"]},\"title\":\"Bokeh Application\",\"version\":\"1.0.4\"}};\n",
       "  var render_items = [{\"docid\":\"a8bbecdd-d0a4-4e01-afd8-6742118f875e\",\"roots\":{\"1178\":\"5e01973a-ccf9-405d-92fe-53fa8cf91795\"}}];\n",
       "  root.Bokeh.embed.embed_items_notebook(docs_json, render_items);\n",
       "\n",
       "  }\n",
       "  if (root.Bokeh !== undefined) {\n",
       "    embed_document(root);\n",
       "  } else {\n",
       "    var attempts = 0;\n",
       "    var timer = setInterval(function(root) {\n",
       "      if (root.Bokeh !== undefined) {\n",
       "        embed_document(root);\n",
       "        clearInterval(timer);\n",
       "      }\n",
       "      attempts++;\n",
       "      if (attempts > 100) {\n",
       "        console.log(\"Bokeh: ERROR: Unable to run BokehJS code because BokehJS library is missing\");\n",
       "        clearInterval(timer);\n",
       "      }\n",
       "    }, 10, root)\n",
       "  }\n",
       "})(window);"
      ],
      "application/vnd.bokehjs_exec.v0+json": ""
     },
     "metadata": {
      "application/vnd.bokehjs_exec.v0+json": {
       "id": "1178"
      }
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "p = figure(tools=\"pan,wheel_zoom,reset,save\",\n",
    "           toolbar_location=\"above\",\n",
    "           title=\"vector T-SNE for most polarized words\")\n",
    "\n",
    "source = ColumnDataSource(data=dict(x1=words_top_ted_tsne[:,0],\n",
    "                                    x2=words_top_ted_tsne[:,1],\n",
    "                                    names=words_to_visualize,\n",
    "                                    color=colors_list))\n",
    "\n",
    "p.scatter(x=\"x1\", y=\"x2\", size=8, source=source, fill_color=\"color\")\n",
    "\n",
    "word_labels = LabelSet(x=\"x1\", y=\"x2\", text=\"names\", y_offset=6,\n",
    "                  text_font_size=\"8pt\", text_color=\"#555555\",\n",
    "                  source=source, text_align='center')\n",
    "p.add_layout(word_labels)\n",
    "\n",
    "show(p)\n",
    "\n",
    "# green indicates positive words, black indicates negative words"
   ]
  }
 ],
 "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": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}