6.1.3 从原始文本到词嵌入.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# 处理 IMDB 原始数据的标签\n",
    "import os\n",
    "imdb_dir = '/mnt/workspace/jupyter_notebook/deep-learning-with-python-notes/aclImdb'\n",
    "train_dir = os.path.join(imdb_dir, 'train')\n",
    "labels = []\n",
    "texts = []\n",
    "for label_type in ['neg', 'pos']:\n",
    "    dir_name = os.path.join(train_dir, label_type)\n",
    "    for fname in os.listdir(dir_name):\n",
    "        if fname[-4:] == '.txt':\n",
    "            f = open(os.path.join(dir_name, fname))\n",
    "            texts.append(f.read())\n",
    "            f.close()\n",
    "        if label_type == 'neg':\n",
    "            labels.append(0)\n",
    "        else:\n",
    "            labels.append(1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 88582 unique tokens.\n",
      "Shape of data tensor: (25000, 100)\n",
      "Shape of label tensor: (25000,)\n"
     ]
    }
   ],
   "source": [
    "# 对 IMDB 原始数据的文本进行分词\n",
    "from keras.preprocessing.text import Tokenizer\n",
    "from keras.preprocessing.sequence import pad_sequences\n",
    "import numpy as np\n",
    "\n",
    "maxlen = 100 # 在100个单词后截断评论\n",
    "training_samples = 200 # 在200个样本上训练\n",
    "validation_samples = 10000 # 在10000个样本上验证\n",
    "max_words = 10000 # 只考虑数据集钱10000个最常见的单词\n",
    "\n",
    "tokenizer = Tokenizer(num_words=max_words)\n",
    "tokenizer.fit_on_texts(texts)\n",
    "sequences = tokenizer.texts_to_sequences(texts)\n",
    "\n",
    "word_index = tokenizer.word_index\n",
    "print('Found %s unique tokens.' % len(word_index))\n",
    "\n",
    "data = pad_sequences(sequences, maxlen=maxlen)\n",
    "\n",
    "labels = np.asarray(labels)\n",
    "print('Shape of data tensor:', data.shape)\n",
    "print('Shape of label tensor:', labels.shape)\n",
    "\n",
    "indices = np.arange(data.shape[0])\n",
    "np.random.shuffle(indices)\n",
    "data = data[indices]\n",
    "labels = labels[indices]\n",
    "\n",
    "x_train = data[:training_samples]\n",
    "y_train = labels[:training_samples]\n",
    "x_val = data[training_samples: training_samples + validation_samples]\n",
    "y_val = labels[training_samples: training_samples + validation_samples]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 400000 word vectors.\n"
     ]
    }
   ],
   "source": [
    "# 　解析 GloVe 词嵌入文件\n",
    "glove_dir = '/home/fc/Downloads/glove.6B'\n",
    "embeddings_index = {}\n",
    "f = open(os.path.join(glove_dir, 'glove.6B.100d.txt'))\n",
    "for line in f:\n",
    "    values = line.split()\n",
    "    word = values[0]\n",
    "    coefs = np.asarray(values[1:], dtype='float32')\n",
    "    embeddings_index[word] = coefs\n",
    "f.close()\n",
    "print('Found %s word vectors.' % len(embeddings_index))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# 　准备 GloVe 词嵌入矩阵\n",
    "embedding_dim = 100\n",
    "embedding_matrix = np.zeros((max_words, embedding_dim))\n",
    "for word, i in word_index.items():\n",
    "    if i < max_words:\n",
    "        embedding_vector = embeddings_index.get(word)\n",
    "        if embedding_vector is not None:\n",
    "            embedding_matrix[i] = embedding_vector "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From /home/fc/anaconda3/lib/python3.6/site-packages/tensorflow/python/framework/op_def_library.py:263: colocate_with (from tensorflow.python.framework.ops) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Colocations handled automatically by placer.\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "embedding_1 (Embedding)      (None, 100, 100)          1000000   \n",
      "_________________________________________________________________\n",
      "flatten_1 (Flatten)          (None, 10000)             0         \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 32)                320032    \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 1)                 33        \n",
      "=================================================================\n",
      "Total params: 1,320,065\n",
      "Trainable params: 1,320,065\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "# 模型定义\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Embedding, Flatten, Dense\n",
    "model = Sequential()\n",
    "model.add(Embedding(max_words, embedding_dim, input_length=maxlen))\n",
    "model.add(Flatten())\n",
    "model.add(Dense(32, activation='relu'))\n",
    "model.add(Dense(1, activation='sigmoid'))\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# 将预训练的词嵌入加载到 Embedding 层中\n",
    "model.layers[0].set_weights([embedding_matrix])\n",
    "model.layers[0].trainable = False"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From /home/fc/anaconda3/lib/python3.6/site-packages/tensorflow/python/ops/math_ops.py:3066: to_int32 (from tensorflow.python.ops.math_ops) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Use tf.cast instead.\n",
      "Train on 200 samples, validate on 10000 samples\n",
      "Epoch 1/10\n",
      "200/200 [==============================] - 2s 8ms/step - loss: 1.5980 - acc: 0.5650 - val_loss: 0.9860 - val_acc: 0.4939\n",
      "Epoch 2/10\n",
      "200/200 [==============================] - 0s 2ms/step - loss: 0.6996 - acc: 0.6500 - val_loss: 0.8721 - val_acc: 0.5093\n",
      "Epoch 3/10\n",
      "200/200 [==============================] - 0s 2ms/step - loss: 0.3261 - acc: 0.8950 - val_loss: 2.4184 - val_acc: 0.5062\n",
      "Epoch 4/10\n",
      "200/200 [==============================] - 0s 2ms/step - loss: 0.3349 - acc: 0.9100 - val_loss: 1.0310 - val_acc: 0.5150\n",
      "Epoch 5/10\n",
      "200/200 [==============================] - 0s 2ms/step - loss: 0.2321 - acc: 0.9000 - val_loss: 0.7978 - val_acc: 0.5530\n",
      "Epoch 6/10\n",
      "200/200 [==============================] - 0s 2ms/step - loss: 0.0834 - acc: 1.0000 - val_loss: 0.8088 - val_acc: 0.5553\n",
      "Epoch 7/10\n",
      "200/200 [==============================] - 0s 2ms/step - loss: 0.0460 - acc: 1.0000 - val_loss: 1.0532 - val_acc: 0.5318\n",
      "Epoch 8/10\n",
      "200/200 [==============================] - 0s 2ms/step - loss: 0.3084 - acc: 0.9150 - val_loss: 0.7622 - val_acc: 0.5624\n",
      "Epoch 9/10\n",
      "200/200 [==============================] - 0s 2ms/step - loss: 0.0286 - acc: 1.0000 - val_loss: 0.7936 - val_acc: 0.5661\n",
      "Epoch 10/10\n",
      "200/200 [==============================] - 0s 2ms/step - loss: 0.0214 - acc: 1.0000 - val_loss: 1.0202 - val_acc: 0.5199\n"
     ]
    }
   ],
   "source": [
    "# 训练与评估\n",
    "model.compile(optimizer='rmsprop',\n",
    "              loss='binary_crossentropy',\n",
    "              metrics=['acc'])\n",
    "history = model.fit(x_train, y_train,\n",
    "                    epochs=10,\n",
    "                    batch_size=32,\n",
    "                    validation_data=(x_val, y_val))\n",
    "model.save_weights('pre_trained_glove_model.h5')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXcAAAEICAYAAACktLTqAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nu4hIBP0/RM7NzyUehagAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f2d506fc8d0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXcAAAEICAYAAACktLTqAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4wLCBo\ndHRwOi8vbWF0cGxvdGxpYi5vcmcvpW3flQAAIABJREFUeJzt3Xl8VOXVwPHfAQJhBwE3EAJ1BWQJ\nEVCQVRHIuNaqiAsuRay2Lu1bKVq3t7xu1AW1vmKrtZqCflxaXwTUQRTRiiZssoigBgwghij7GnLe\nP56ZkIQsk2Rm7p3J+X4++czMnTv3ntxJzjxz7nOfR1QVY4wxyaWe1wEYY4yJPkvuxhiThCy5G2NM\nErLkbowxSciSuzHGJCFL7sYYk4QsuZtyiUh9EdkpIh2jua6XROR4EYl6318ROUtEcks8Xi0iZ0ay\nbg329VcRmVTT11ey3T+JyN+jvV3jnQZeB2CiQ0R2lnjYBNgHHAw9vkFVs6qzPVU9CDSL9rp1gaqe\nFI3tiMj1wBWqOqTEtq+PxrZN8rPkniRUtTi5hlqG16tqsKL1RaSBqhbGIzZjTPxZWaaOCH3tfkVE\npovIDuAKETldRD4Vka0isklEpopISmj9BiKiIpIWevxy6PnZIrJDRP4jIp2ru27o+VEi8pWIbBOR\nJ0XkYxEZV0HckcR4g4isFZGfRGRqidfWF5HHRKRARL4GRlZyfO4SkRlllj0tIo+G7l8vIqtCv8/X\noVZ1RdvKE5EhoftNROSlUGwrgD7l7Peb0HZXiMh5oeWnAk8BZ4ZKXltKHNt7S7x+Quh3LxCRf4nI\nMZEcm6qIyAWheLaKyPsiclKJ5yaJyEYR2S4iX5b4XfuLyKLQ8s0i8kik+zMxoKr2k2Q/QC5wVpll\nfwL2A+fiPtQbA6cB/XDf4LoAXwE3h9ZvACiQFnr8MrAFyABSgFeAl2uw7pHADuD80HO3AweAcRX8\nLpHE+G+gJZAG/Bj+3YGbgRVAB6ANMN/9yZe7ny7ATqBpiW3/AGSEHp8bWkeAYcAeoEfoubOA3BLb\nygOGhO5PAT4AWgOdgJVl1r0EOCb0nlweiuGo0HPXAx+UifNl4N7Q/RGhGHsBqcBfgPcjOTbl/P5/\nAv4eun9KKI5hofdoUui4pwDdgHXA0aF1OwNdQvc/B8aE7jcH+nn9v1CXf6zlXrcsUNX/U9UiVd2j\nqp+r6kJVLVTVb4BpwOBKXv+aqmar6gEgC5dUqrtuAFiiqv8OPfcY7oOgXBHG+ICqblPVXFwiDe/r\nEuAxVc1T1QLgwUr28w2wHPehA3A2sFVVs0PP/5+qfqPO+8BcoNyTpmVcAvxJVX9S1XW41njJ/b6q\nqptC78k/cR/MGRFsF2As8FdVXaKqe4GJwGAR6VBinYqOTWUuA95S1fdD79GDQAvch2wh7oOkW6i0\n923o2IH7kD5BRNqo6g5VXRjh72FiwJJ73fJdyQcicrKIvC0i34vIduB+oG0lr/++xP3dVH4StaJ1\njy0Zh6oqrqVbrghjjGhfuBZnZf4JjAndvxz3oRSOIyAiC0XkRxHZims1V3aswo6pLAYRGSciS0Pl\nj63AyRFuF9zvV7w9Vd0O/AS0L7FOdd6zirZbhHuP2qvqauC3uPfhh1CZ7+jQqtcAXYHVIvKZiIyO\n8PcwMWDJvW4p2w3wWVxr9XhVbQHcjSs7xNImXJkEABERSiejsmoT4ybguBKPq+qq+QpwVqjlez4u\n2SMijYHXgAdwJZNWwLsRxvF9RTGISBfgGeBGoE1ou1+W2G5V3TY34ko94e01x5V/NkQQV3W2Ww/3\nnm0AUNWXVXUAriRTH3dcUNXVqnoZrvT2Z+B1EUmtZSymhiy5123NgW3ALhE5BbghDvucCaSLyLki\n0gC4BWgXoxhfBW4VkfYi0ga4o7KVVXUzsAB4AVitqmtCTzUCGgL5wEERCQDDqxHDJBFpJe46gJtL\nPNcMl8DzcZ9z1+Na7mGbgQ7hE8jlmA5cJyI9RKQRLsl+pKoVfhOqRszniciQ0L7/C3eeZKGInCIi\nQ0P72xP6OYj7Ba4Ukbahlv620O9WVMtYTA1Zcq/bfgtcjfvHfRbXco2pUAK9FHgUKAB+BizG9cuP\ndozP4GrjX+BO9r0WwWv+iTtB+s8SMW8FbgPexJ2UvBj3IRWJe3DfIHKB2cA/Smx3GTAV+Cy0zslA\nyTr1e8AaYLOIlCyvhF8/B1ceeTP0+o64OnytqOoK3DF/BvfBMxI4L1R/bwQ8jDtP8j3um8JdoZeO\nBlaJ6401BbhUVffXNh5TM+JKnsZ4Q0Tq48oAF6vqR17HY0yysJa7iTsRGSkiLUNf7f+I64Hxmcdh\nGZNULLkbLwwEvsF9tR8JXKCqFZVljDE1YGUZY4xJQtZyN8aYJFTlwGEichzuDP/RuG5N01T1iTLr\nDMFd5vxtaNEbqnp/Zdtt27atpqWl1SBkY4ypu3JycraoamXdh4HIRoUsBH6rqotCF0nkiMh7qrqy\nzHofqWog0gDT0tLIzs6OdHVjjDGAiFR1pTUQQVkmNO7FotD9HcAqKr+i0BhjjMeqVXMXN6Rrb0pf\naBF2emiMjNki0i0KsRljjKmhiCfrEJFmwOvAraEBikpaBHRS1Z2hwYL+BZxQzjbGA+MBOnb09Yxs\nxhiT0CLqChkaX2Im8I6qPhrB+rm4cbArHMo1IyNDreZuTPwcOHCAvLw89u7d63UoJgKpqal06NCB\nlJTSQwuJSI6qVjksdCS9ZQT4G7CqosQeGvJzs6qqiPTFlXsKIvkFjDHxkZeXR/PmzUlLS8P9Wxu/\nUlUKCgrIy8ujc+fOVb+gHJGUZQYAVwJfiMiS0LJJhIYuVdX/xQ2kdKOIFOJGibtM7eooY3xl7969\nltgThIjQpk0b8vPza7yNKpO7qi6ginGrVfUpyswwY4zxH0vsiaO275VdoZrgcnLgww+9jsIY4zeW\n3BPchAlw6aVQZFMiGJ8rKCigV69e9OrVi6OPPpr27dsXP96/P7Jh36+55hpWr15d6TpPP/00WVlZ\nla4TqYEDB7JkyZKqV/ShiLtCGv/58UfXcld1t6ed5nVEJplkZcGdd8L69dCxI0yeDGNrMRVImzZt\nihPlvffeS7Nmzfjd735Xah1VRVWpV6/8ducLL7xQ5X5uuummmgeZRKzlnsDmzXOJHeDtt72NxSSX\nrCwYPx7WrXN/Y+vWucdRahCXsnbtWrp3786ECRNIT09n06ZNjB8/noyMDLp168b99x8apircki4s\nLKRVq1ZMnDiRnj17cvrpp/PDDz8AcNddd/H4448Xrz9x4kT69u3LSSedxCeffALArl27+PnPf07P\nnj0ZM2YMGRkZVbbQX375ZU499VS6d+/OpEmTACgsLOTKK68sXj516lQAHnvsMbp27UrPnj254oor\non7MImHJPYEFg9CsmWuxz4x00jdjInDnnbB7d+llu3e75bGwcuVKrrvuOhYvXkz79u158MEHyc7O\nZunSpbz33nusXFl2KCvYtm0bgwcPZunSpZx++uk8//zz5W5bVfnss8945JFHij8onnzySY4++miW\nLl3KxIkTWbx4caXx5eXlcddddzFv3jwWL17Mxx9/zMyZM8nJyWHLli188cUXLF++nKuuugqAhx9+\nmCVLlrB06VKeesqbviaW3BNYMAhDhsCFF7qyzKZNXkdkksX69dVbXls/+9nPOK1EXXH69Omkp6eT\nnp7OqlWryk3ujRs3ZtSoUQD06dOH3Nzccrd90UUXHbbOggULuOyyywDo2bMn3bpVPmLKwoULGTZs\nGG3btiUlJYXLL7+c+fPnc/zxx7N69WpuueUW3nnnHVq2bAlAt27duOKKK8jKyjrsIqR4seSeoNat\ng7Vr4ayzIBAai3PWLG9jMsmjotFBYjVqSNOmTYvvr1mzhieeeIL333+fZcuWMXLkyHKvqm3YsGHx\n/fr161NYWFjuths1anTYOtW9DKei9du0acOyZcsYOHAgU6dO5YYbbgDgnXfeYcKECXz22WdkZGRw\n8ODBau0vGiy5J6i5c93t8OHQvTscd5yVZkz0TJ4MTZqUXtakiVsea9u3b6d58+a0aNGCTZs28c47\n70R9HwMHDuTVV18F4Isvvij3m0FJ/fv3Z968eRQUFFBYWMiMGTMYPHgw+fn5qCq/+MUvuO+++1i0\naBEHDx4kLy+PYcOG8cgjj5Cfn8/usjWuOLDeMgkqGISjjoJu3UDEtd7/8Q/YuxdSU72OziS6cK+Y\naPaWiVR6ejpdu3ale/fudOnShQEDBkR9H7/+9a+56qqr6NGjB+np6XTv3r24pFKeDh06cP/99zNk\nyBBUlXPPPZfMzEwWLVrEddddh6oiIjz00EMUFhZy+eWXs2PHDoqKirjjjjto3rx51H+Hqng2h6oN\nHFZzRUVw9NEwYgS8/LJbNmsWZGbCnDlwzjnexmf8adWqVZxyyileh+ELhYWFFBYWkpqaypo1axgx\nYgRr1qyhQQN/tXfLe8+iNnCY8Z/lyyE/39Xbw4YOhcaNXZdIS+7GVG7nzp0MHz6cwsJCVJVnn33W\nd4m9tpLrt6kjgkF3O3z4oWWNG7vHM2fCE0+4Uo0xpnytWrUiJyfH6zBiyk6oJqBgEE46yZ1ELSkQ\ngG+/hVWrvInLGOMfltwTzP79bqCwkiWZsNGj3a1drWqMseSeYD791F0pWF5yP+446NnTukQaYyy5\nJ5y5c6FePXdlankCAfj4Y/jpp7iGZYzxGUvuCSYYdGPJtGpV/vOZmXDwIMTgug9jamXIkCGHXZD0\n+OOP86tf/arS1zVr1gyAjRs3cvHFF1e47aq6Vj/++OOlLiYaPXo0W7dujST0St17771MmTKl1tuJ\nNkvuCWT7dli4sHQvmbL69oW2ba00Y/xnzJgxzJgxo9SyGTNmMGbMmIhef+yxx/Laa6/VeP9lk/us\nWbNoVVErKQlYck8gH37oWuXl1dvD6td3J1Znz3brGuMXF198MTNnzmTfvn0A5ObmsnHjRgYOHFjc\n7zw9PZ1TTz2Vf//734e9Pjc3l+7duwOwZ88eLrvsMnr06MGll17Knj17ite78cYbi4cLvueeewCY\nOnUqGzduZOjQoQwdOhSAtLQ0tmzZAsCjjz5K9+7d6d69e/Fwwbm5uZxyyin88pe/pFu3bowYMaLU\nfsqzZMkS+vfvT48ePbjwwgv5KVQfnTp1Kl27dqVHjx7FA5Z9+OGHxZOV9O7dmx07dtT42JbH+rkn\nkGDQ9Wc//fTK18vMdEMRfPopxODKbZMEbr0Voj3BUK9eEMqL5WrTpg19+/Zlzpw5nH/++cyYMYNL\nL70UESE1NZU333yTFi1asGXLFvr37895551X4TyizzzzDE2aNGHZsmUsW7aM9PT04ucmT57MEUcc\nwcGDBxk+fDjLli3jN7/5DY8++ijz5s2jbdu2pbaVk5PDCy+8wMKFC1FV+vXrx+DBg2ndujVr1qxh\n+vTpPPfcc1xyySW8/vrrlY7PftVVV/Hkk08yePBg7r77bu677z4ef/xxHnzwQb799lsaNWpUXAqa\nMmUKTz/9NAMGDGDnzp2kRnncEGu5J5BgEM48s+qxY0aMgAYNrDRj/KdkaaZkSUZVmTRpEj169OCs\ns85iw4YNbN68ucLtzJ8/vzjJ9ujRgx49ehQ/9+qrr5Kenk7v3r1ZsWJFlYOCLViwgAsvvJCmTZvS\nrFkzLrroIj766CMAOnfuTK9evYDKhxUGN7781q1bGTx4MABXX3018+fPL45x7NixvPzyy8VXwg4Y\nMIDbb7+dqVOnsnXr1qhfIWst9wSxcSOsXAnjxlW9bqtW7kNg5kx44IGYh2YSUGUt7Fi64IILuP32\n21m0aBF79uwpbnFnZWWRn59PTk4OKSkppKWllTvMb0nlteq//fZbpkyZwueff07r1q0ZN25cldup\nbHyt8HDB4IYMrqosU5G3336b+fPn89Zbb/Hf//3frFixgokTJ5KZmcmsWbPo378/wWCQk08+uUbb\nL4+13BNEeIjfyurtJWVmujFo1q2LXUzGVFezZs0YMmQI1157bakTqdu2bePII48kJSWFefPmsa6K\nP9xBgwYVT4K9fPlyli1bBrjhgps2bUrLli3ZvHkzs2fPLn5N8+bNy61rDxo0iH/961/s3r2bXbt2\n8eabb3LmmWdW+3dr2bIlrVu3Lm71v/TSSwwePJiioiK+++47hg4dysMPP8zWrVvZuXMnX3/9Naee\neip33HEHGRkZfPnll9XeZ2Ws5Z4ggkFo08ZdpBSJQAB+9zt3tWoVPc2MiasxY8Zw0UUXleo5M3bs\nWM4991wyMjLo1atXlS3YG2+8kWuuuYYePXrQq1cv+vbtC7hZlXr37k23bt0OGy54/PjxjBo1imOO\nOYZ58+YVL09PT2fcuHHF27j++uvp3bt3pSWYirz44otMmDCB3bt306VLF1544QUOHjzIFVdcwbZt\n21BVbrvtNlq1asUf//hH5s2bR/369enatWvxrFLRYkP+JgBVd/XpgAHwyiuRv+bEE+GEE2yGJuPY\nkL+JpzZD/lpZJgGsXg0bNkRekgE3KmRmJrz/PuzaFbvYjDH+ZMk9AZQ3xG8kAgHYt88leGNM3WLJ\nPQEEg9C5M3TpUr3XDRoEzZpZl0hziFdlWFN9tX2vLLn7XGEhzJtXvZJMWMOGrs/722+7Gryp21JT\nUykoKLAEnwBUlYKCglpd2GS9ZXwuO9uNKVOT5A6uNPPGG7B0qbuC0NRdHTp0IC8vj/z8fK9DMRFI\nTU2lQ4cONX69JXefC9fbhw2r2evDE3jMnGnJva5LSUmhc+fOXodh4sTKMj4XDELv3m6kx5o46ig3\nRLDNzmRM3WLJ3cd27YJPPql5SSYsEHBDBf/wQ3TiMsb4X5XJXUSOE5F5IrJKRFaIyC3lrCMiMlVE\n1orIMhFJL29bpnoWLIADB6KT3FXdMMDGmLohkpZ7IfBbVT0F6A/cJCJdy6wzCjgh9DMeeCaqUdZR\nwaDr8TJwYO2207s3HHOMlWaMqUuqTO6quklVF4Xu7wBWAe3LrHY+8A91PgVaicgxUY+2jgkG3ZAD\nTZrUbjvhq1XfeQf2749ObMYYf6tWzV1E0oDewMIyT7UHvivxOI/DPwAQkfEiki0i2dYdq3L5+W4y\nhepelVqRQMB1qVywIDrbM8b4W8TJXUSaAa8Dt6rq9rJPl/OSw66UUNVpqpqhqhnt2rWrXqR1THjI\ngNrW28OGD3clHivNGFM3RJTcRSQFl9izVPWNclbJA44r8bgDsLH24dVdwSC0bAl9+kRne82awdCh\nNhSBMXVFJL1lBPgbsEpVH61gtbeAq0K9ZvoD21R1UxTjrFNU4b33XDKO5sxbgQB89RWsWRO9bRpj\n/CmSlvsA4EpgmIgsCf2MFpEJIjIhtM4s4BtgLfAcYNND1MI337gZlKJVkgnLzHS3VpoxJvlV2S5U\n1QWUX1MvuY4CN0UrqLouPORAtJN7587Qtasrzdx6a3S3bYzxF7tC1YfmzoUOHdxMStEWCMCHH7qe\nM8aY5GXJ3WeKilxyP+ss1z892jIz3TDC770X/W0bY/zDkrvPLFkCP/4Y/ZJM2BlnQKtW1mvGmGRn\nyd1najvEb1UaNIBRo9yk2UVFsdmHMcZ7ltx9JhiEbt3cWDCxkpnpRojMzo7dPowx3rLk7iN798JH\nH8WuJBM2ciTUq2elGWOSmSV3H/nkE5fgY53c27RxtXdL7sYkL0vuPhIMQv36MHhw7PeVmQmLF8OG\nDbHflzEm/iy5+8jcudC/PzRvHvt9BQLudtas2O/LGBN/ltx94qef3AnOWJdkwrp1g06drDRjTLKy\n5O4TH3zguibGK7mHJ/AIBl2d3xiTXCy5+0Qw6Ibl7dcvfvsMBGD3bvfBYoxJLpbcfSIYdCdSU1Li\nt8+hQ90UflaaMSb5WHL3gfXr3Tjr0ZpSL1KpqW6fb7/txpA3xiQPS+4+MHeuu41Xvb2kQAByc2Hl\nyvjv2xgTO5bcfSAYhCOPhO7d47/v8AQeVpoxJrlYcveYamyH+K1K+/bQq5fNzmRMsrHk7rEVK2Dz\nZm9KMmGBAHz8sRtq2BiTHCy5eyw8xG+8T6aWFAi4PvZz5ngXgzEmuiy5eywYdNPpdezoXQynnQbt\n2llpxphkYsndQwcOuAuIvCzJgBv+d/RomD3bTcFnjEl8ltw9tHAh7NrlfXIHV5r56Sf4z3+8jsQY\nEw2W3D0UDLpW85AhXkcCZ5/tpuCz0owxycGSu4eCQejTB1q39joSaNkSBg2y/u7GJIuESu5ZWZCW\n5lq7aWnucaLavh0+/dQfJZmwQMB1zczN9ToSY0xtJUxyz8qC8eNh3Tp34c+6de5xoib4+fPh4EF/\nJffw1apWmjEm8SVMcr/zTjc8bUm7d7vliWjuXDdw1xlneB3JISeeCCecYKUZY5JBwiT39eurt9zv\ngkE480yX4P0kEIB581wvHmNM4kqY5F7RRT5eXvxTU99/D8uX+6skE5aZCfv2HRqp0hiTmBImuU+e\n7CaWKKlJE7c80Xg5xG9VzjzTTdBtpRljElvCJPexY2HaNDeps4i7nTbNLU80wSAccYQbjdFvGjaE\nc86xCTyMSXQJk9zBJfLcXDfIVW5uYiZ2VZfchw93XTr9KDMTNm6EJUu8jsQYU1NVphcReV5EfhCR\n5RU8P0REtonIktDP3dEPM3l89RXk5Xk7CmRVRo1y346sNGNM4oqk7fh3YGQV63ykqr1CP/fXPqzk\nFR7i14/19rCjjoK+fS25G5PIqkzuqjofsGkcoiQYdFfXdunidSSVy8yEzz93E4kYYxJPtKq+p4vI\nUhGZLSLdKlpJRMaLSLaIZOfn50dp14nj4EHXh9yrKfWqIxBw5wdmz/Y6EmNMTUQjuS8COqlqT+BJ\n4F8Vraiq01Q1Q1Uz2rVrF4VdJ5acHNi2zd8lmbBeveDYY600Y0yiqnVyV9XtqrozdH8WkCIibWsd\nWRIK19uHDfM2jkiIuNLMu+/C/v1eR2OMqa5aJ3cROVrEFRlEpG9omwW13W4yCgZdizhRvrQEArBj\nB3z0kdeRGGOqK5KukNOB/wAniUieiFwnIhNEZEJolYuB5SKyFJgKXKZql7+UtXs3fPxxYpRkwoYP\nh0aNrDRjTCJqUNUKqjqmiuefAp6KWkRJasECV95IpOTetCkMHeqS+2OPeR2NMaY6fHqNZPIJBiEl\nBQYO9DqS6gkEYO1ad/GVMSZxWHKPk2DQjd3etKnXkVRPeAIPK80Yk1gsucfBli2weHFilWTC0tKg\nWzdL7sYkGkvucTBvnrtNxOQOrjTz0Ueuj74xJjFYco+DYBBatICMDK8jqZlAAAoLXZ93Y0xisOQe\nB8Gg63XSoMq+Sf7Uvz+0bm2lGWMSiSX3GPvmG/eTqCUZcB9Ko0a5cWYOHvQ6GmNMJCy5x5ifp9Sr\njkAA8vPdSJHGGP+z5B5jwSC0bw8nneR1JLVzzjlu5igrzRiTGCy5x1BRkWu5Dx/u/yF+q3LEETBg\ngJtb1Rjjf5bcY2jpUigoSPySTFgg4OZVzcvzOhJjTFUsucdQuN7u5/lSqyN8taq13o3xP0vuMRQM\nQteubtKLZNC1q7ti1ZK7Mf5nyT1G9u2D+fOTpyQD7rxBIOA+tPbs8ToaY0xlLLnHyH/+4xJgMiV3\ncKWZPXsODalgjPEnS+4xEgxC/foweLDXkUTXkCHQpImVZozxO0vuMRIMQr9+bkyZZJKaCmef7fq7\n23xbxviXJfcY2LrVXcmZbCWZsMxMWL8eli/3OhJjTEUsucfABx+4C5iSNbmPHu1urTRjjH9Zco+B\nYNDVpfv18zqS2GjfHtLTbSgCY/zMknsMzJ3rTqQ2bOh1JLGTmel6BBUUeB2JMaY8ltyjLC8Pvvwy\neUsyYYGAKz3NmeN1JMaY8lhyj7JkGeK3KhkZcOSRVpoxxq8suUdZMOiSXvfuXkcSW/XquROrc+a4\nKfiMMf5iyT2KVF1yHz7cJb9kFwi4bp+ffOJ1JMaYsupACoqflSvh+++TvyQTdvbZkJJipRlj/MiS\nexQFg+62riT3Fi1g0CBL7sb4kSX3KAoG4YQToGNHryOJn0AAVq1yk4AbY/zDknuUHDjgrkxNlok5\nIhUIuFu7WtUYf7HkHiWffw47d9adkkzY8cfDiSdaacYYv7HkHiXBoJvMYuhQryOJv0DAfWvZudPr\nSEzYtm1w223QrRu8+67X0RgvWHKPkmAQ+vSBI47wOpL4CwRg//5DJ5SNd4qK4MUX3bepJ55wSf6c\nc1yi37vX6+hMPFWZ3EXkeRH5QUTKHeBVnKkislZElolIevTD9LedO904K3WtJBM2cCC0bAm//70N\nR+ClJUvgzDNh3Djo3Bk++wzWrIGbb4bHH4e+fW2YZj9Yvhw2bYr9fiJpuf8dGFnJ86OAE0I/44Fn\nah9WYpk/312lWVeTe0oKvP66u4hr1Ch35eqqVV5HVXf89BPcdJP75rhmDTz/vLuwLCMDGjeGJ590\nJ7w3b3bLpk61iVa8sGkT/PKX0LMn/OlPcdihqlb5A6QByyt47llgTInHq4Fjqtpmnz59NFncdptq\naqrqnj1eR+KtfftUH31UtWVL1fr1VW++WXXLFq+jSl4HD6o+95xq27aq9eq54/3jjxWv//33qqNH\nq4LqyJGqmzbFL9a6bMcO1XvuUW3SRDUlxeWL2vxfANkaSd6OaKXKk/tMYGCJx3OBjArWHQ9kA9kd\nO3as0S+2YYP7g16/vkYvj4lTT1U96yyvo/CP/HzVX/3KJfhWrVQfe8wlfhM9n3+u2rev+w8eOFB1\nyZLIXldUpPr0064x0rat6ltvxTbOuuzAAdVp01SPPtq9T5dcorp2be23G8/k/nY5yb1PVdusacv9\nhRdc1KB6yimqt96qOmeO6u7dNdpcrW3a5GJ58EFv9u9ny5ernnOOOz4nnOASSVGR11Eltvx81V/+\nUlXEJY2XXqrZMV2xQrVXL/feTJigumtX9GOtq4qKVN9+W7VrV3d8zzhD9ZNPorf9eCb3uJZliopU\nv/hCdcoU1bPPVm3UyP0WqamqI0ao/vnPLqnEK4lkZbn9Z2fHZ3+JaNYs1ZNPdsdp+HDVpUu9jijx\nFBaq/uUvqq1bu29Et9+uum3Wdo7zAAAOiklEQVRb7ba5d6/q737n3peTTlLNyYlOrHVZTo7qsGHu\nmB5/vOrrr0c/F8UzuWcCswEB+gOfRbLNaNXcd+1SnT3bteBPOeVQq75DB9Vrr1V95RXVgoKo7Kpc\n117r/uEKC2O3j2Swf7/qk0+qHnGEqw+PH6+6ebPXUSWGTz5R7d3b/V0PHeoaL9H03nuqxx7r6sEP\nPeRq+aZ61q1TvfJK9x61aaM6dWrsSpFRS+7AdGATcADIA64DJgATQs8L8DTwNfBFRfX2sj+xOqG6\nbp2ryV98sTuxBy6Z9O/vTmp88omrhUVDUZHqccep/vzn0dleXVBQoHrLLaoNGqi2aKH68MOuBWkO\n9/33quPGub/h9u1VZ8yI3TfSLVvc33H4A8RP57T8bOtW1YkTXQWhUSPVO+5wy2Ipqi33WPzEo7fM\ngQOqH3+sevfdqv36uSQP7iTfxRfX/sTsV1+57T3zTPRiriu+/FI1EHDHr0sX1ddes3p82IEDqk88\n4RonKSkuYezYEfv9FhWp/u1vqk2buv+RV1+N/T4TVfibaNu27m/4iitUc3Pjs29L7uUoKHBlmmuv\ndS2hsidmZ8+u3onZv/zFvX7NmtjFnOzefVe1e3d3HAcNsrrvhx+63lfgzil9+WX8Y1iz5lBPnKuv\nVt2+Pf4x+FVRkeobb7gOAuFvOfE+32bJvQpFRa52+ec/1/zE7EUXqXbqZC3O2jpwQPV//9e1gkRU\nr7lGdeNGr6OKrw0bVMeOdX+DHTvG5kRcdezfr3rXXe7bbpcu0e3tkag+/VR1wAD3HnXtqjpzpjfv\nkSX3aqruidnCQvfV9brrvIs52Wzdqvpf/+VKEU2bqk6e7F0X13jZv9/1/GrWTLVhQ5dQ/dQt8aOP\nXAOmfn3V++6L3vmqRPL1166POqgedZTqs896exwsuddSRSdm+/VzNfznnnPLpk/3OtLks2aN6oUX\nHmrFTp+enN+O5s491JAYPdq/5b2tWw99qzjjDJfs6oKCAnc1aUqKu7r07rv9UaKy5B5FFZ2YBevO\nF0vz5h260OaMM1QXLvQ6ouhYv/5QS7Bz58S5SjQry/Vwat5c9cUXk/MDV9X13poyxX0zr1fPfTvf\nsMHrqA6x5B5D4ROzM2Z4HUnyKyx0PTiOOkqLeyV8953XUdXM3r2qDzzgWoGpqa7MkWjjEX37rRvu\nAFQvvbTysWwSTVGR+5aYlqbF4+8sW+Z1VIez5G6SyvbtqpMmuRPfjRu7axZ27vQ6qsjNmaN64onu\nP+6CC1ySTFSFhe58SIMG7jqPDz7wOqLa+/BD1dNOc+9Pz56uF5dfRZrcbbIOkxCaN4fJk+HLL+G8\n8+C+++Ckk+Cll9wEFX6VmwsXXggjR7pC3uzZ8OabkJbmdWQ1V78+TJrkhhVOTXWzj/3hD27ClkSz\nerV7fwYPho0b4e9/h5wcOPtsryOLgkg+AWLxYy13UxsLFhxqaZ12mjsn4id79qjef78rvzRpovo/\n/5OcV+Lu2KF6/fXufejTx5t++TWxefOhkUubNXPfRPzUS6kyRNhyF7du/GVkZGh2drYn+zbJoagI\nsrJg4kTX6rr0UnjoIejUqfR6qnDgAOzbV/7P/v3RfW7/fli7FvLy4JJLYMoUOO44b45RvLzxhpuI\nYu9eeOwxd1/E66gOt3u3m5XqwQfd/RtugHvugSOP9DqyyIlIjqpmVLmeJXeT6HbtgkcegYcfhoMH\noV27w5NtNDVqdPhPw4alH7doAbfeCsOHR3fffrZhg5viLxiE88+H555z70W87dsHBQWwZUvp2++/\nh7/9zX3onn++S/Annxz/+GrLkrupc/Ly4NFH3aTQ5SXgihJxdZ5LSfFni9Qviopcy/gPf3CTxb/4\nIowYUfPt7dnjEnM4SZdN2CVvw/d37qx4e/36uUbAoEE1j8lrltyNMZ5ZuhQuvxxWrnTfYB54wCX+\nSJJzydvduyveR4sW0LYttGnjbkveL3vbtq37sGnUKH7HIFYsuRtjPLVnD/z+9/DUU66HzcGDFa/b\nqlX5CbmiZUcc4b5J1UWRJvcG8QjGGFP3NG4MTz4J554L773nknJFLeoGlomizg6pMSamRoyoXd3d\n1IxdxFQDWVnuIpR69dxtVpbXERljTGnWcq+mrCwYP/7QiZ5169xjgLFjvYvLGGNKspZ7Nd155+Fn\n8HfvdsuNMcYvLLlX0/r11VtujDFesOReTR07Vm+5McZ4wZJ7NU2eDE2alF7WpIlbbowxfmHJvZrG\njoVp09zgVCLudto0O5lqjPEX6y1TA2PHWjI3xvibtdyNMSYJWXI3xpgkZMndGGOSkCV3Y4xJQpbc\njTEmCVlyN8aYJGTJ3RhjkpAld2OMSUIRJXcRGSkiq0VkrYhMLOf5cSKSLyJLQj/XRz9UY4wxkary\nClURqQ88DZwN5AGfi8hbqrqyzKqvqOrNMYjRGGNMNUXScu8LrFXVb1R1PzADOD+2YRljjKmNSJJ7\ne+C7Eo/zQsvK+rmILBOR10TkuPI2JCLjRSRbRLLz8/NrEK4xxphIRJLcpZxlWubx/wFpqtoDCAIv\nlrchVZ2mqhmqmtGuXbvqRWqMMSZikST3PKBkS7wDsLHkCqpaoKr7Qg+fA/pEJzxjjDE1EUly/xw4\nQUQ6i0hD4DLgrZIriMgxJR6eB6yKXojGGGOqq8reMqpaKCI3A+8A9YHnVXWFiNwPZKvqW8BvROQ8\noBD4ERgXw5iNMcZUQVTLls/jIyMjQ7Ozsz3ZtzHGJCoRyVHVjKrWsytUjTEmCVlyN8aYJGTJ3Rhj\nkpAld2OMSUKW3I0xJglZcje1lpUFaWlQr567zcryOiJjjCX3BOaHpJqVBePHw7p1oOpux4+3BG+M\n1yy5Jyi/JNU774Tdu0sv273bLTfGDw2QusouYkpQaWkuoZfVqRPk5sYvjnr13IdLWSJQVBS/OIz/\nhBsgJT/8mzSBadNg7Fjv4kp0dhFTklu/vnrLY6Vjx+otN3WHfavzliX3BOWXpDp5smuNldSkiVtu\n6ja/NEDqKkvuCcovSXXsWPc1u1MnV4rp1Mm+dhvHLw2QusqSe4LyU1IdO9bV+YuK3K0ldgP+aYDU\nVVUO+Wv8a+xYS6TGv8J/m3fe6UoxHTu6xG5/s/FhLXeTNKzbnf/YtzrvWMvdJIWy3e7C/f7BEoqp\nm6zlbpKCdbszpjRL7iYpWLc7Y0qz5G6SgnW7M6Y0S+4mKVi3O2NKs+RukoKf+v0b4wfWW8YkDev3\nb8wh1nI3xpgkZMndGGOSkCV3Y4xJQpbcjTEmCVlyNybKbIwb4wfWW8aYKLIxboxfWMvdmCiyMW6M\nX1hyNyaK/DLGjZWGjCV3Y6LID2PchEtD69aB6qHSUF1O8HXxw86SuzFR5Icxbqw0VJqfPuzi+SFj\nyd2YKPLDGDd+KQ35hV8+7OL9ISOqWvVKIiOBJ4D6wF9V9cEyzzcC/gH0AQqAS1U1t7JtZmRkaHZ2\ndg3DNsZUJC3NJY6yOnVyU93VNfXquWRaloib/i9eovW+iEiOqmZUtV6VLXcRqQ88DYwCugJjRKRr\nmdWuA35S1eOBx4CHIg/VGBNNfigN+YkfzoNA/L9RRVKW6QusVdVvVHU/MAM4v8w65wMvhu6/BgwX\nEYlemMaYSPmhNOQnfvmwi/eHTCTJvT3wXYnHeaFl5a6jqoXANqBN2Q2JyHgRyRaR7Pz8/JpFbIyp\n0tix7qt+UZG7rauJHfzzYRfvD5lIknt5LfCyFaxI1kFVp6lqhqpmtGvXLpL4jDGm1vzwYRfvD5lI\nhh/IA44r8bgDsLGCdfJEpAHQEvgxKhEaY0ySiOeEMpG03D8HThCRziLSELgMeKvMOm8BV4fuXwy8\nr5F0wzHGGBMTVbbcVbVQRG4G3sF1hXxeVVeIyP1Atqq+BfwNeElE1uJa7JfFMmhjjDGVi2hUSFWd\nBcwqs+zuEvf3Ar+IbmjGGGNqyq5QNcaYJGTJ3RhjklBEww/EZMci+UA5F+MmlLbAFq+D8BE7HqXZ\n8TjEjkVptTkenVS1yr7kniX3ZCAi2ZGM8VBX2PEozY7HIXYsSovH8bCyjDHGJCFL7sYYk4QsudfO\nNK8D8Bk7HqXZ8TjEjkVpMT8eVnM3xpgkZC13Y4xJQpbcjTEmCVlyrwEROU5E5onIKhFZISK3eB2T\n10SkvogsFpGZXsfiNRFpJSKviciXob+R072OyUsiclvo/2S5iEwXkVSvY4onEXleRH4QkeUllh0h\nIu+JyJrQbeto79eSe80UAr9V1VOA/sBN5Uw9WNfcAqzyOgifeAKYo6onAz2pw8dFRNoDvwEyVLU7\nbvDBujaw4N+BkWWWTQTmquoJwNzQ46iy5F4DqrpJVReF7u/A/fOWnZ2qzhCRDkAm8FevY/GaiLQA\nBuFGSkVV96vqVm+j8lwDoHForocmHD4fRFJT1fkcPr9FyalJXwQuiPZ+LbnXkoikAb2Bhd5G4qnH\ngd8DcZxL3re6APnAC6Ey1V9FpKnXQXlFVTcAU4D1wCZgm6q+621UvnCUqm4C11gEjoz2Diy514KI\nNANeB25V1e1ex+MFEQkAP6hqjtex+EQDIB14RlV7A7uIwVfuRBGqJZ8PdAaOBZqKyBXeRlU3WHKv\nIRFJwSX2LFV9w+t4PDQAOE9EcoEZwDARednbkDyVB+Spavib3Gu4ZF9XnQV8q6r5qnoAeAM4w+OY\n/GCziBwDELr9Ido7sOReAyIiuJrqKlV91Ot4vKSqf1DVDqqahjtR9r6q1tmWmap+D3wnIieFFg0H\nVnoYktfWA/1FpEno/2Y4dfgEcwklpya9Gvh3tHcQ0UxM5jADgCuBL0RkSWjZpNCMVcb8GsgKzTn8\nDXCNx/F4RlUXishrwCJcL7PF1LGhCERkOjAEaCsiecA9wIPAqyJyHe4DMOoz2dnwA8YYk4SsLGOM\nMUnIkrsxxiQhS+7GGJOELLkbY0wSsuRujDFJyJK7McYkIUvuxhiThP4f1++n79Y/O+wAAAAASUVO\nRK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f2d1dd2eef0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# 绘制结果\n",
    "import matplotlib.pyplot as plt\n",
    "% matplotlib inline\n",
    "\n",
    "acc = history.history['acc']\n",
    "val_acc = history.history['val_acc']\n",
    "loss = history.history['loss']\n",
    "val_loss = history.history['val_loss']\n",
    "epochs = range(1, len(acc) + 1)\n",
    "plt.plot(epochs, acc, 'bo', label='Training acc')\n",
    "plt.plot(epochs, val_acc, 'b', label='Validation acc')\n",
    "plt.title('Training and validation accuracy')\n",
    "plt.legend()\n",
    "\n",
    "plt.figure()\n",
    "plt.plot(epochs, loss, 'bo', label='Training loss')\n",
    "plt.plot(epochs, val_loss, 'b', label='Validation loss')\n",
    "plt.title('Training and validation loss')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "embedding_2 (Embedding)      (None, 100, 100)          1000000   \n",
      "_________________________________________________________________\n",
      "flatten_2 (Flatten)          (None, 10000)             0         \n",
      "_________________________________________________________________\n",
      "dense_3 (Dense)              (None, 32)                320032    \n",
      "_________________________________________________________________\n",
      "dense_4 (Dense)              (None, 1)                 33        \n",
      "=================================================================\n",
      "Total params: 1,320,065\n",
      "Trainable params: 1,320,065\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "Train on 200 samples, validate on 10000 samples\n",
      "Epoch 1/10\n",
      "200/200 [==============================] - 1s 4ms/step - loss: 0.6965 - acc: 0.4500 - val_loss: 0.6927 - val_acc: 0.5103\n",
      "Epoch 2/10\n",
      "200/200 [==============================] - 1s 3ms/step - loss: 0.5200 - acc: 0.9900 - val_loss: 0.6961 - val_acc: 0.5068\n",
      "Epoch 3/10\n",
      "200/200 [==============================] - 1s 3ms/step - loss: 0.3026 - acc: 0.9950 - val_loss: 0.7034 - val_acc: 0.5099\n",
      "Epoch 4/10\n",
      "200/200 [==============================] - 1s 3ms/step - loss: 0.1414 - acc: 1.0000 - val_loss: 0.7008 - val_acc: 0.5240\n",
      "Epoch 5/10\n",
      "200/200 [==============================] - 1s 3ms/step - loss: 0.0685 - acc: 1.0000 - val_loss: 0.7022 - val_acc: 0.5247\n",
      "Epoch 6/10\n",
      "200/200 [==============================] - 1s 3ms/step - loss: 0.0359 - acc: 1.0000 - val_loss: 0.7062 - val_acc: 0.5268\n",
      "Epoch 7/10\n",
      "200/200 [==============================] - 1s 3ms/step - loss: 0.0204 - acc: 1.0000 - val_loss: 0.7123 - val_acc: 0.5275\n",
      "Epoch 8/10\n",
      "200/200 [==============================] - 1s 3ms/step - loss: 0.0118 - acc: 1.0000 - val_loss: 0.7152 - val_acc: 0.5252\n",
      "Epoch 9/10\n",
      "200/200 [==============================] - 1s 3ms/step - loss: 0.0070 - acc: 1.0000 - val_loss: 0.7219 - val_acc: 0.5269\n",
      "Epoch 10/10\n",
      "200/200 [==============================] - 1s 3ms/step - loss: 0.0043 - acc: 1.0000 - val_loss: 0.7281 - val_acc: 0.5305\n"
     ]
    }
   ],
   "source": [
    "# 在不使用预训练词嵌入的情况下，训练相同的模型\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Embedding, Flatten, Dense\n",
    "model = Sequential()\n",
    "model.add(Embedding(max_words, embedding_dim, input_length=maxlen))\n",
    "model.add(Flatten())\n",
    "model.add(Dense(32, activation='relu'))\n",
    "model.add(Dense(1, activation='sigmoid'))\n",
    "model.summary()\n",
    "model.compile(optimizer='rmsprop',\n",
    "             loss='binary_crossentropy',\n",
    "             metrics=['acc'])\n",
    "history = model.fit(x_train, y_train,\n",
    "             epochs=10,\n",
    "             batch_size=32,\n",
    "             validation_data=(x_val, y_val))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXcAAAEICAYAAACktLTqAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\noP8P4bR3lEjYtp8AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f2d6345eda0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXcAAAEICAYAAACktLTqAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nrkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f2d1c43aa20>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# 绘制结果\n",
    "import matplotlib.pyplot as plt\n",
    "% matplotlib inline\n",
    "\n",
    "acc = history.history['acc']\n",
    "val_acc = history.history['val_acc']\n",
    "loss = history.history['loss']\n",
    "val_loss = history.history['val_loss']\n",
    "epochs = range(1, len(acc) + 1)\n",
    "plt.plot(epochs, acc, 'bo', label='Training acc')\n",
    "plt.plot(epochs, val_acc, 'b', label='Validation acc')\n",
    "plt.title('Training and validation accuracy')\n",
    "plt.legend()\n",
    "\n",
    "plt.figure()\n",
    "plt.plot(epochs, loss, 'bo', label='Training loss')\n",
    "plt.plot(epochs, val_loss, 'b', label='Validation loss')\n",
    "plt.title('Training and validation loss')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# 对测试集数据进行分词\n",
    "test_dir = os.path.join(imdb_dir, 'test')\n",
    "labels = []\n",
    "texts = []\n",
    "for label_type in ['neg', 'pos']:\n",
    "    dir_name = os.path.join(test_dir, label_type)\n",
    "    for fname in sorted(os.listdir(dir_name)):\n",
    "        if fname[-4:] == '.txt':\n",
    "            f = open(os.path.join(dir_name, fname))\n",
    "            texts.append(f.read())\n",
    "            f.close()\n",
    "            if label_type == 'neg':\n",
    "                labels.append(0)\n",
    "            else:\n",
    "                labels.append(1)\n",
    "sequences = tokenizer.texts_to_sequences(texts)\n",
    "x_test = pad_sequences(sequences, maxlen=maxlen)\n",
    "y_test = np.asarray(labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "25000/25000 [==============================] - 1s 58us/step\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[1.0008599682617187, 0.52647999999999995]"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 在测试集上评估模型\n",
    "model.load_weights('pre_trained_glove_model.h5')\n",
    "model.evaluate(x_test, y_test)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}