fast text.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using Theano backend.\n"
     ]
    }
   ],
   "source": [
    "'''This example demonstrates the use of fast text in https://arxiv.org/abs/1607.01759.\n",
    "NOTE: not using hieraltical softmax and bi-gram features\n",
    "Gets to 0.88 test accuracy after 6 epochs. (equal CNN in Keras example: https://github.com/fchollet/keras/blob/master/examples/imdb_cnn.py)\n",
    "10s/epoch on Intel i5 1.4Ghz CPU.\n",
    "1s/epoch on Tesla K40 GPU.\n",
    "'''\n",
    " \n",
    "import numpy as np\n",
    "np.random.seed(1337)\n",
    "\n",
    "from keras.preprocessing import sequence\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense, Lambda, Embedding\n",
    "from keras.datasets import imdb\n",
    "from keras import backend as K"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loading data...\n",
      "20000 train sequences\n",
      "5000 test sequences\n",
      "Pad sequences (samples x time)\n"
     ]
    }
   ],
   "source": [
    "# set parameters:\n",
    "max_features = 5000\n",
    "maxlen = 400\n",
    "batch_size = 32\n",
    "embedding_dims = 50\n",
    "\n",
    "print('Loading data...')\n",
    "(X_train, y_train), (X_test, y_test) = imdb.load_data(nb_words=max_features,\n",
    "                                                      test_split=0.2)\n",
    "\n",
    "print(len(X_train), 'train sequences')\n",
    "print(len(X_test), 'test sequences')\n",
    "\n",
    "print('Pad sequences (samples x time)')\n",
    "X_train = sequence.pad_sequences(X_train, maxlen=maxlen)\n",
    "X_test = sequence.pad_sequences(X_test, maxlen=maxlen)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def means(x):\n",
    "    return K.mean(x, axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 20000 samples, validate on 5000 samples\n",
      "Epoch 1/10\n",
      "20000/20000 [==============================] - 10s - loss: 0.6322 - acc: 0.7113 - val_loss: 0.5388 - val_acc: 0.7882\n",
      "Epoch 2/10\n",
      "20000/20000 [==============================] - 9s - loss: 0.4584 - acc: 0.8365 - val_loss: 0.4032 - val_acc: 0.8496\n",
      "Epoch 3/10\n",
      "20000/20000 [==============================] - 10s - loss: 0.3575 - acc: 0.8712 - val_loss: 0.3462 - val_acc: 0.8674\n",
      "Epoch 4/10\n",
      "20000/20000 [==============================] - 9s - loss: 0.3080 - acc: 0.8864 - val_loss: 0.3162 - val_acc: 0.8768\n",
      "Epoch 5/10\n",
      "20000/20000 [==============================] - 9s - loss: 0.2780 - acc: 0.8959 - val_loss: 0.3021 - val_acc: 0.8794\n",
      "Epoch 6/10\n",
      "20000/20000 [==============================] - 9s - loss: 0.2567 - acc: 0.9040 - val_loss: 0.2918 - val_acc: 0.8824\n",
      "Epoch 7/10\n",
      "20000/20000 [==============================] - 10s - loss: 0.2402 - acc: 0.9105 - val_loss: 0.2843 - val_acc: 0.8850\n",
      "Epoch 8/10\n",
      "20000/20000 [==============================] - 10s - loss: 0.2262 - acc: 0.9166 - val_loss: 0.2808 - val_acc: 0.8854\n",
      "Epoch 9/10\n",
      "20000/20000 [==============================] - 10s - loss: 0.2153 - acc: 0.9213 - val_loss: 0.2792 - val_acc: 0.8868\n",
      "Epoch 10/10\n",
      "20000/20000 [==============================] - 9s - loss: 0.2060 - acc: 0.9244 - val_loss: 0.2788 - val_acc: 0.8874\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x10e083f28>"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model = Sequential()\n",
    "model.add(Embedding(input_dim=5000, output_dim=embedding_dims, init='glorot_uniform'))\n",
    "model.add(Lambda(means, output_shape=(embedding_dims,)))\n",
    "\n",
    "# We project onto a single unit output layer, and squash it with a sigmoid:\n",
    "model.add(Dense(1, activation=\"sigmoid\"))\n",
    "\n",
    "model.compile(loss='binary_crossentropy',\n",
    "              optimizer='adam',\n",
    "              metrics=['accuracy'])\n",
    "\n",
    "model.fit(X_train, y_train,\n",
    "          batch_size=batch_size,\n",
    "          validation_data=(X_test, y_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "anaconda-cloud": {},
  "kernelspec": {
   "display_name": "Python [Root]",
   "language": "python",
   "name": "Python [Root]"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.5.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}