5_word2vec_cbow.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "D7tqLMoKF6uq"
   },
   "source": [
    "Deep Learning\n",
    "=============\n",
    "\n",
    "Assignment 5\n",
    "------------\n",
    "\n",
    "The goal of this assignment is to train a Word2Vec skip-gram model over [Text8](http://mattmahoney.net/dc/textdata) data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "cellView": "both",
    "colab": {
     "autoexec": {
      "startup": false,
      "wait_interval": 0
     }
    },
    "colab_type": "code",
    "collapsed": true,
    "id": "0K1ZyLn04QZf"
   },
   "outputs": [],
   "source": [
    "# These are all the modules we'll be using later. Make sure you can import them\n",
    "# before proceeding further.\n",
    "%matplotlib inline\n",
    "from __future__ import print_function\n",
    "import collections\n",
    "import math\n",
    "import numpy as np\n",
    "import os\n",
    "import random\n",
    "import tensorflow as tf\n",
    "import zipfile\n",
    "from matplotlib import pylab\n",
    "from six.moves import range\n",
    "from six.moves.urllib.request import urlretrieve\n",
    "from sklearn.manifold import TSNE"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "aCjPJE944bkV"
   },
   "source": [
    "Download the data from the source website if necessary."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "cellView": "both",
    "colab": {
     "autoexec": {
      "startup": false,
      "wait_interval": 0
     },
     "output_extras": [
      {
       "item_id": 1
      }
     ]
    },
    "colab_type": "code",
    "collapsed": false,
    "executionInfo": {
     "elapsed": 14640,
     "status": "ok",
     "timestamp": 1445964482948,
     "user": {
      "color": "#1FA15D",
      "displayName": "Vincent Vanhoucke",
      "isAnonymous": false,
      "isMe": true,
      "permissionId": "05076109866853157986",
      "photoUrl": "//lh6.googleusercontent.com/-cCJa7dTDcgQ/AAAAAAAAAAI/AAAAAAAACgw/r2EZ_8oYer4/s50-c-k-no/photo.jpg",
      "sessionId": "2f1ffade4c9f20de",
      "userId": "102167687554210253930"
     },
     "user_tz": 420
    },
    "id": "RJ-o3UBUFtCw",
    "outputId": "c4ec222c-80b5-4298-e635-93ca9f79c3b7"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found and verified text8.zip\n"
     ]
    }
   ],
   "source": [
    "url = 'http://mattmahoney.net/dc/'\n",
    "\n",
    "def maybe_download(filename, expected_bytes):\n",
    "    \"\"\"Download a file if not present, and make sure it's the right size.\"\"\"\n",
    "    if not os.path.exists(filename):\n",
    "        filename, _ = urlretrieve(url + filename, filename)\n",
    "    statinfo = os.stat(filename)\n",
    "    if statinfo.st_size == expected_bytes:\n",
    "        print('Found and verified %s' % filename)\n",
    "    else:\n",
    "        print(statinfo.st_size)\n",
    "        raise Exception(\n",
    "          'Failed to verify ' + filename + '. Can you get to it with a browser?')\n",
    "    return filename\n",
    "\n",
    "filename = maybe_download('text8.zip', 31344016)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "Zqz3XiqI4mZT"
   },
   "source": [
    "Read the data into a string."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "cellView": "both",
    "colab": {
     "autoexec": {
      "startup": false,
      "wait_interval": 0
     },
     "output_extras": [
      {
       "item_id": 1
      }
     ]
    },
    "colab_type": "code",
    "collapsed": false,
    "executionInfo": {
     "elapsed": 28844,
     "status": "ok",
     "timestamp": 1445964497165,
     "user": {
      "color": "#1FA15D",
      "displayName": "Vincent Vanhoucke",
      "isAnonymous": false,
      "isMe": true,
      "permissionId": "05076109866853157986",
      "photoUrl": "//lh6.googleusercontent.com/-cCJa7dTDcgQ/AAAAAAAAAAI/AAAAAAAACgw/r2EZ_8oYer4/s50-c-k-no/photo.jpg",
      "sessionId": "2f1ffade4c9f20de",
      "userId": "102167687554210253930"
     },
     "user_tz": 420
    },
    "id": "Mvf09fjugFU_",
    "outputId": "e3a928b4-1645-4fe8-be17-fcf47de5716d"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Data size 17005207\n"
     ]
    }
   ],
   "source": [
    "def read_data(filename):\n",
    "    \"\"\"Extract the first file enclosed in a zip file as a list of words\"\"\"\n",
    "    with zipfile.ZipFile(filename) as f:\n",
    "        data = tf.compat.as_str(f.read(f.namelist()[0])).split()\n",
    "    return data\n",
    "  \n",
    "words = read_data(filename)\n",
    "print('Data size %d' % len(words))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "Zdw6i4F8glpp"
   },
   "source": [
    "Build the dictionary and replace rare words with UNK token."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "cellView": "both",
    "colab": {
     "autoexec": {
      "startup": false,
      "wait_interval": 0
     },
     "output_extras": [
      {
       "item_id": 1
      }
     ]
    },
    "colab_type": "code",
    "collapsed": false,
    "executionInfo": {
     "elapsed": 28849,
     "status": "ok",
     "timestamp": 1445964497178,
     "user": {
      "color": "#1FA15D",
      "displayName": "Vincent Vanhoucke",
      "isAnonymous": false,
      "isMe": true,
      "permissionId": "05076109866853157986",
      "photoUrl": "//lh6.googleusercontent.com/-cCJa7dTDcgQ/AAAAAAAAAAI/AAAAAAAACgw/r2EZ_8oYer4/s50-c-k-no/photo.jpg",
      "sessionId": "2f1ffade4c9f20de",
      "userId": "102167687554210253930"
     },
     "user_tz": 420
    },
    "id": "gAL1EECXeZsD",
    "outputId": "3fb4ecd1-df67-44b6-a2dc-2291730970b2"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Most common words (+UNK) [['UNK', 418391], ('the', 1061396), ('of', 593677), ('and', 416629), ('one', 411764)]\n",
      "Sample data [5239, 3084, 12, 6, 195, 2, 3137, 46, 59, 156]\n"
     ]
    }
   ],
   "source": [
    "vocabulary_size = 50000\n",
    "\n",
    "def build_dataset(words):\n",
    "    count = [['UNK', -1]]\n",
    "    count.extend(collections.Counter(words).most_common(vocabulary_size - 1))\n",
    "    dictionary = dict()\n",
    "    for word, _ in count:\n",
    "        dictionary[word] = len(dictionary)\n",
    "    data = list()\n",
    "    unk_count = 0\n",
    "    for word in words:\n",
    "        if word in dictionary:\n",
    "            index = dictionary[word]\n",
    "        else:\n",
    "            index = 0  # dictionary['UNK']\n",
    "            unk_count = unk_count + 1\n",
    "        data.append(index)\n",
    "    count[0][1] = unk_count\n",
    "    reverse_dictionary = dict(zip(dictionary.values(), dictionary.keys())) \n",
    "    return data, count, dictionary, reverse_dictionary\n",
    "\n",
    "data, count, dictionary, reverse_dictionary = build_dataset(words)\n",
    "print('Most common words (+UNK)', count[:5])\n",
    "print('Sample data', data[:10])\n",
    "del words  # Hint to reduce memory."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "lFwoyygOmWsL"
   },
   "source": [
    "Function to generate a training batch for the skip-gram model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "cellView": "both",
    "colab": {
     "autoexec": {
      "startup": false,
      "wait_interval": 0
     },
     "output_extras": [
      {
       "item_id": 1
      }
     ]
    },
    "colab_type": "code",
    "collapsed": false,
    "executionInfo": {
     "elapsed": 113,
     "status": "ok",
     "timestamp": 1445964901989,
     "user": {
      "color": "#1FA15D",
      "displayName": "Vincent Vanhoucke",
      "isAnonymous": false,
      "isMe": true,
      "permissionId": "05076109866853157986",
      "photoUrl": "//lh6.googleusercontent.com/-cCJa7dTDcgQ/AAAAAAAAAAI/AAAAAAAACgw/r2EZ_8oYer4/s50-c-k-no/photo.jpg",
      "sessionId": "2f1ffade4c9f20de",
      "userId": "102167687554210253930"
     },
     "user_tz": 420
    },
    "id": "w9APjA-zmfjV",
    "outputId": "67cccb02-cdaf-4e47-d489-43bcc8d57bb8"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "data: ['anarchism', 'originated', 'as', 'a', 'term', 'of', 'abuse', 'first']\n",
      "\n",
      "with num_skips = 8 and skip_window = 1:\n",
      "    batch: ['anarchism', 'as', 'a', 'originated', 'as', 'term', 'of', 'a']\n",
      "    labels: ['originated', 'as', 'a', 'term']\n",
      "\n",
      "with num_skips = 4 and skip_window = 2:\n",
      "    batch: ['originated', 'a', 'anarchism', 'term', 'as', 'originated', 'term', 'of']\n",
      "    labels: ['as', 'a']\n"
     ]
    }
   ],
   "source": [
    "data_index = 0\n",
    "\n",
    "def generate_batch_cbow(batch_size, skip_window):\n",
    "    global data_index\n",
    "    context_window = 2 * skip_window  # calculate the context_window - this is the total number of words around the target\n",
    "    assert batch_size % context_window == 0 # ensure the context window can be taken from the batch size\n",
    "    num_labels = batch_size / context_window # how many context windows fit in the batch\n",
    "    batch = np.ndarray(shape=(batch_size), dtype=np.int32)\n",
    "    labels = np.ndarray(shape=(num_labels, 1), dtype=np.int32)\n",
    "    span = 2 * skip_window + 1 # [ skip_window target skip_window ]\n",
    "    buffer = collections.deque(maxlen=span)\n",
    "    for _ in range(span):\n",
    "        buffer.append(data[data_index])\n",
    "        data_index = (data_index + 1) % len(data)\n",
    "    for i in range(batch_size // context_window):\n",
    "        target = skip_window  # target label at the center of the buffer\n",
    "        labels[i, 0] = buffer[skip_window]\n",
    "        targets_to_avoid = [ skip_window ]\n",
    "        for j in range(context_window):\n",
    "            while target in targets_to_avoid:\n",
    "                target = random.randint(0, span - 1)\n",
    "            targets_to_avoid.append(target)\n",
    "            batch[i * context_window + j] = buffer[target]\n",
    "        buffer.append(data[data_index])\n",
    "        data_index = (data_index + 1) % len(data)\n",
    "    return batch, labels\n",
    "\n",
    "print('data:', [reverse_dictionary[di] for di in data[:8]])\n",
    "\n",
    "for skip_window in [1,2]:\n",
    "    data_index = 0\n",
    "    batch, labels = generate_batch_cbow(batch_size=8, skip_window=skip_window)\n",
    "    print('\\nwith num_skips = %d and skip_window = %d:' % (8/skip_window, skip_window))\n",
    "    print('    batch:', [reverse_dictionary[bi] for bi in batch])\n",
    "    print('    labels:', [reverse_dictionary[li] for li in labels.reshape(8/(2*skip_window))])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "Ofd1MbBuwiva"
   },
   "source": [
    "Train a skip-gram model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "cellView": "both",
    "colab": {
     "autoexec": {
      "startup": false,
      "wait_interval": 0
     }
    },
    "colab_type": "code",
    "collapsed": true,
    "id": "8pQKsV4Vwlzy"
   },
   "outputs": [],
   "source": [
    "batch_size = 128\n",
    "embedding_size = 128 # Dimension of the embedding vector.\n",
    "skip_window = 1 # How many words to consider left and right.\n",
    "num_skips = 2 # How many times to reuse an input to generate a label.\n",
    "# We pick a random validation set to sample nearest neighbors. here we limit the\n",
    "# validation samples to the words that have a low numeric ID, which by\n",
    "# construction are also the most frequent. \n",
    "valid_size = 16 # Random set of words to evaluate similarity on.\n",
    "valid_window = 100 # Only pick dev samples in the head of the distribution.\n",
    "valid_examples = np.array(random.sample(range(valid_window), valid_size))\n",
    "num_sampled = 64 # Number of negative examples to sample.\n",
    "\n",
    "# General definition\n",
    "context_window = 2 * skip_window\n",
    "num_labels = batch_size / context_window\n",
    "\n",
    "graph = tf.Graph()\n",
    "\n",
    "with graph.as_default():\n",
    "\n",
    "    # Input data.\n",
    "    train_dataset = tf.placeholder(tf.int32, shape=[batch_size])\n",
    "    train_labels = tf.placeholder(tf.int32, shape=[num_labels, 1])\n",
    "    valid_dataset = tf.constant(valid_examples, dtype=tf.int32)\n",
    "  \n",
    "    # Variables.\n",
    "    embeddings = tf.Variable(\n",
    "        tf.random_uniform([vocabulary_size, embedding_size], -1.0, 1.0))\n",
    "    softmax_weights = tf.Variable(\n",
    "        tf.truncated_normal([vocabulary_size, embedding_size],\n",
    "                         stddev=1.0 / math.sqrt(embedding_size)))\n",
    "    softmax_biases = tf.Variable(tf.zeros([vocabulary_size]))\n",
    "  \n",
    "    # Model.\n",
    "    # Look up embeddings for inputs.\n",
    "    embed = tf.nn.embedding_lookup(embeddings, train_dataset)\n",
    "    seq_ids = np.array([[i]*context_window for i in range(num_sampled)]).flatten()\n",
    "    embed = tf.segment_sum(embed, seq_ids)\n",
    "    # Compute the softmax loss, using a sample of the negative labels each time.\n",
    "    loss = tf.reduce_mean(\n",
    "        tf.nn.sampled_softmax_loss(softmax_weights, softmax_biases, embed,\n",
    "                               train_labels, num_sampled, vocabulary_size))\n",
    "\n",
    "    # Optimizer.\n",
    "    optimizer = tf.train.AdagradOptimizer(1.0).minimize(loss)\n",
    "  \n",
    "    # Compute the similarity between minibatch examples and all embeddings.\n",
    "    # We use the cosine distance:\n",
    "    norm = tf.sqrt(tf.reduce_sum(tf.square(embeddings), 1, keep_dims=True))\n",
    "    normalized_embeddings = embeddings / norm\n",
    "    valid_embeddings = tf.nn.embedding_lookup(\n",
    "        normalized_embeddings, valid_dataset)\n",
    "    similarity = tf.matmul(valid_embeddings, tf.transpose(normalized_embeddings))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "cellView": "both",
    "colab": {
     "autoexec": {
      "startup": false,
      "wait_interval": 0
     },
     "output_extras": [
      {
       "item_id": 23
      },
      {
       "item_id": 48
      },
      {
       "item_id": 61
      }
     ]
    },
    "colab_type": "code",
    "collapsed": false,
    "executionInfo": {
     "elapsed": 436189,
     "status": "ok",
     "timestamp": 1445965429787,
     "user": {
      "color": "#1FA15D",
      "displayName": "Vincent Vanhoucke",
      "isAnonymous": false,
      "isMe": true,
      "permissionId": "05076109866853157986",
      "photoUrl": "//lh6.googleusercontent.com/-cCJa7dTDcgQ/AAAAAAAAAAI/AAAAAAAACgw/r2EZ_8oYer4/s50-c-k-no/photo.jpg",
      "sessionId": "2f1ffade4c9f20de",
      "userId": "102167687554210253930"
     },
     "user_tz": 420
    },
    "id": "1bQFGceBxrWW",
    "outputId": "5ebd6d9a-33c6-4bcd-bf6d-252b0b6055e4",
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Initialized\n",
      "Average loss at step 0: 7.783331\n",
      "Nearest to states: beheld, safi, interdependent, shots, claws, chlorophyll, band, u,\n",
      "Nearest to b: abstinence, marcius, amritsar, deposited, dflp, emits, tarnished, sever,\n",
      "Nearest to american: dizzy, indirect, seaton, hailstones, queuing, akc, crandall, passwords,\n",
      "Nearest to people: lansdowne, giordano, galloping, bulletproof, impeachment, pka, bangladesh, clusters,\n",
      "Nearest to he: tertius, auckland, experiences, captive, castration, hephaestion, intrusion, con,\n",
      "Nearest to world: outflow, letsie, posts, dotted, coven, leptons, tractates, burgesses,\n",
      "Nearest to will: pox, disapproved, vaughn, pogroms, affects, occupies, calwell, hinayana,\n",
      "Nearest to his: features, akan, bleach, radon, strived, province, lupus, favours,\n",
      "Nearest to that: inflammatory, pipers, eisner, eugenie, hofstadter, incomprehensible, zhu, sarajevo,\n",
      "Nearest to most: inasmuch, siberian, ballots, optical, modell, employed, rubenstein, detriment,\n",
      "Nearest to other: upscale, unary, canonised, hoard, fpss, functionalist, castell, hi,\n",
      "Nearest to only: tsim, ligases, inhumane, deserving, kola, tenniel, internalized, legends,\n",
      "Nearest to i: deceit, othniel, worse, heckman, measurements, exported, rustic, manner,\n",
      "Nearest to a: felony, egmont, parcels, betrayal, seaboard, sidelines, gingrich, oversees,\n",
      "Nearest to been: cabral, sandstorms, donut, floresiensis, reestablishment, orderly, heliports, disclaimed,\n",
      "Nearest to four: ribosome, ropes, malting, teleplay, worldly, carbines, menagerie, hanlon,\n",
      "Average loss at step 2000: 4.226807\n",
      "Average loss at step 4000: 3.636816\n",
      "Average loss at step 6000: 3.549703\n",
      "Average loss at step 8000: 3.387287\n",
      "Average loss at step 10000: 3.295756\n",
      "Nearest to states: shots, rebuked, countries, beheld, morgan, netted, handle, stabilizes,\n",
      "Nearest to b: d, juneau, tolson, abstinence, rockville, magellan, durian, parthia,\n",
      "Nearest to american: french, and, glycine, moods, nome, cassini, cai, thracian,\n",
      "Nearest to people: signet, haze, playstation, cases, pretence, commencing, wasteland, charterers,\n",
      "Nearest to he: she, it, they, there, who, we, exterminated, subscribed,\n",
      "Nearest to world: u, nec, letsie, shock, voted, region, fresnel, nss,\n",
      "Nearest to will: can, could, would, may, should, might, must, cannot,\n",
      "Nearest to his: their, her, its, the, my, s, aleut, merchandise,\n",
      "Nearest to that: which, argos, eventually, rst, shout, opportunities, when, curtain,\n",
      "Nearest to most: more, schilling, pushing, modena, parti, paradigmatic, yugoslavia, wraith,\n",
      "Nearest to other: battling, some, ways, these, reverence, different, many, birthdate,\n",
      "Nearest to only: pursuant, replaced, tsim, tabasco, hansard, ddd, eagle, contingents,\n",
      "Nearest to i: we, they, grasso, sociopolitical, she, legality, mamre, you,\n",
      "Nearest to a: the, this, plated, historiographical, kazakstan, no, pongo, treblinka,\n",
      "Nearest to been: become, were, be, was, tendons, ruiz, hart, thoughtful,\n",
      "Nearest to four: six, seven, eight, three, zero, five, two, nine,\n",
      "Average loss at step 12000: 3.273646\n",
      "Average loss at step 14000: 3.242148\n",
      "Average loss at step 16000: 3.019836\n",
      "Average loss at step 18000: 3.055558\n",
      "Average loss at step 20000: 3.152527\n",
      "Nearest to states: nations, types, countries, netted, shots, dylan, pixar, poseidon,\n",
      "Nearest to b: d, juneau, rockville, limestone, dalmatian, normalize, eda, foundry,\n",
      "Nearest to american: british, italian, french, canadian, english, german, spanish, ancient,\n",
      "Nearest to people: cases, others, men, behaviours, sadc, holidays, buoyancy, times,\n",
      "Nearest to he: she, it, they, who, there, druyan, exterminated, bilirubin,\n",
      "Nearest to world: u, holy, voted, book, ang, vietnam, fresnel, outflow,\n",
      "Nearest to will: would, can, should, could, may, must, might, cannot,\n",
      "Nearest to his: her, their, my, its, the, this, our, whose,\n",
      "Nearest to that: which, but, however, what, rst, lanthanides, tallinn, gaul,\n",
      "Nearest to most: more, some, schilling, wraith, leopoldo, less, sidemen, lively,\n",
      "Nearest to other: many, different, various, some, reverence, woolwich, telepathy, confucianism,\n",
      "Nearest to only: garter, customs, cues, deserving, replaced, jethro, intercept, tilt,\n",
      "Nearest to i: ii, we, iii, lawrencium, haereses, spousal, mons, verona,\n",
      "Nearest to a: any, palme, another, very, nationalism, braddock, acyclic, firenze,\n",
      "Nearest to been: become, be, cliffordadams, already, tendons, were, boutros, was,\n",
      "Nearest to four: seven, five, six, eight, three, nine, two, zero,\n",
      "Average loss at step 22000: 3.119410\n",
      "Average loss at step 24000: 3.076549\n",
      "Average loss at step 26000: 3.057445\n",
      "Average loss at step 28000: 3.063464\n",
      "Average loss at step 30000: 3.088435\n",
      "Nearest to states: countries, nations, kingdom, netted, horn, state, amazingly, poseidon,\n",
      "Nearest to b: d, c, tolson, eda, limestone, chihuly, normalize, rave,\n",
      "Nearest to american: british, australian, canadian, italian, english, austrian, pianists, german,\n",
      "Nearest to people: others, scientists, cases, writers, children, secretive, men, jews,\n",
      "Nearest to he: she, it, they, originally, who, there, we, never,\n",
      "Nearest to world: holy, u, outflow, place, continent, vietnam, region, showings,\n",
      "Nearest to will: would, can, may, must, could, should, might, cannot,\n",
      "Nearest to his: their, her, its, my, the, our, your, whose,\n",
      "Nearest to that: which, what, this, lanthanides, dependant, wt, christof, if,\n",
      "Nearest to most: more, jewell, among, schilling, paradigmatic, lively, sponsors, many,\n",
      "Nearest to other: different, older, several, various, reverence, jirga, marcellinus, some,\n",
      "Nearest to only: november, crucible, still, always, desperation, joual, vous, directive,\n",
      "Nearest to i: we, you, ii, iii, g, they, jerome, zeeland,\n",
      "Nearest to a: no, the, another, plated, palme, suckling, generalizes, any,\n",
      "Nearest to been: become, were, was, be, come, already, thoughtful, refraining,\n",
      "Nearest to four: seven, eight, six, three, five, nine, zero, two,\n",
      "Average loss at step 32000: 3.081313\n",
      "Average loss at step 34000: 3.045467\n",
      "Average loss at step 36000: 2.995827\n",
      "Average loss at step 38000: 2.797691\n",
      "Average loss at step 40000: 2.982875\n",
      "Nearest to states: nations, countries, kingdom, netted, archers, state, horn, secretion,\n",
      "Nearest to b: d, p, e, r, UNK, c, g, j,\n",
      "Nearest to american: canadian, russian, australian, indian, german, british, italian, yea,\n",
      "Nearest to people: men, scientists, cases, individuals, children, intercepted, person, animals,\n",
      "Nearest to he: she, it, they, newbie, i, who, there, we,\n",
      "Nearest to world: u, rod, bombed, continent, moe, album, oedipus, buckyballs,\n",
      "Nearest to will: would, can, must, could, should, may, cannot, might,\n",
      "Nearest to his: their, her, its, my, your, whose, the, our,\n",
      "Nearest to that: which, what, this, however, generally, blogging, krag, rst,\n",
      "Nearest to most: more, among, less, leopoldo, anacletus, paradigmatic, schilling, lively,\n",
      "Nearest to other: various, different, older, reverence, several, many, takeovers, numerous,\n",
      "Nearest to only: esteemed, cahokia, usually, toasts, option, always, not, santos,\n",
      "Nearest to i: we, you, g, iii, ii, he, she, pasta,\n",
      "Nearest to a: another, palme, the, any, adherence, every, coupled, numbers,\n",
      "Nearest to been: become, be, were, being, badly, never, are, was,\n",
      "Nearest to four: eight, six, seven, five, two, nine, three, one,\n",
      "Average loss at step 42000: 2.983240\n",
      "Average loss at step 44000: 2.987076\n",
      "Average loss at step 46000: 2.983530\n",
      "Average loss at step 48000: 2.881806\n",
      "Average loss at step 50000: 2.903051\n",
      "Nearest to states: nations, kingdom, countries, woodland, amazingly, state, horn, rickshaws,\n",
      "Nearest to b: c, d, gif, satires, sais, carrel, brussels, crystalline,\n",
      "Nearest to american: russian, indian, italian, canadian, australian, harpo, hedonist, swiss,\n",
      "Nearest to people: men, words, children, jews, animals, baptists, scientists, letters,\n",
      "Nearest to he: she, it, they, who, there, we, newbie, gore,\n",
      "Nearest to world: u, continent, game, oedipus, savant, season, rod, buckyballs,\n",
      "Nearest to will: would, could, must, may, can, should, might, cannot,\n",
      "Nearest to his: her, their, my, its, your, whose, our, the,\n",
      "Nearest to that: which, however, deoxyribonucleic, thief, weston, meantime, blogging, rst,\n",
      "Nearest to most: more, less, some, schilling, sponsors, many, among, futures,\n",
      "Nearest to other: different, older, many, some, various, reverence, those, multiple,\n",
      "Nearest to only: rask, now, praetorian, always, anglicanism, either, intractable, never,\n",
      "Nearest to i: we, you, g, p, c, they, meo, d,\n",
      "Nearest to a: another, the, kazakstan, generalizes, plated, any, postmodernist, beasts,\n",
      "Nearest to been: become, be, being, was, already, were, figured, entrenched,\n",
      "Nearest to four: six, seven, three, five, eight, nine, two, zero,\n",
      "Average loss at step 52000: 2.953902\n",
      "Average loss at step 54000: 2.930086\n",
      "Average loss at step 56000: 2.957082\n",
      "Average loss at step 58000: 2.877521\n",
      "Average loss at step 60000: 2.897687\n",
      "Nearest to states: nations, kingdom, countries, rebuked, hamlin, regions, state, netted,\n",
      "Nearest to b: c, p, d, khair, assailants, rescues, z, cl,\n",
      "Nearest to american: italian, russian, canadian, indian, australian, harpo, african, austrian,\n",
      "Nearest to people: men, women, cultures, scientists, tribes, individuals, jews, players,\n",
      "Nearest to he: she, it, they, originally, finally, who, never, eventually,\n",
      "Nearest to world: u, continent, buckyballs, savant, outflow, moe, european, rod,\n",
      "Nearest to will: could, must, would, can, may, should, might, cannot,\n",
      "Nearest to his: her, their, my, its, your, our, xxxii, favours,\n",
      "Nearest to that: which, what, this, lanthanides, it, gauntlets, defoe, strat,\n",
      "Nearest to most: some, many, groves, more, lack, scherzo, end, lively,\n",
      "Nearest to other: different, older, smokey, others, various, many, those, sala,\n",
      "Nearest to only: always, until, actually, appearances, still, never, missing, do,\n",
      "Nearest to i: we, you, ii, heaney, molten, g, reisman, intrinsic,\n",
      "Nearest to a: the, another, any, airtime, palme, kazakstan, retract, swine,\n",
      "Nearest to been: become, be, was, remained, disturbed, being, experienced, come,\n",
      "Nearest to four: six, five, eight, three, seven, nine, two, zero,\n",
      "Average loss at step 62000: 2.683755\n",
      "Average loss at step 64000: 2.699987\n",
      "Average loss at step 66000: 2.879155\n",
      "Average loss at step 68000: 2.888142\n",
      "Average loss at step 70000: 2.832167\n",
      "Nearest to states: nations, kingdom, countries, state, initialized, deposition, domination, robotics,\n",
      "Nearest to b: h, d, z, p, qualifies, j, sais, capitalistic,\n",
      "Nearest to american: italian, korean, australian, indian, harpo, america, swiss, canadian,\n",
      "Nearest to people: women, men, scientists, individuals, children, cultures, players, those,\n",
      "Nearest to he: she, it, they, there, badly, eventually, newbie, we,\n",
      "Nearest to world: u, supreme, philadelphia, continent, venereal, wachowski, savant, season,\n",
      "Nearest to will: would, may, could, can, might, must, cannot, should,\n",
      "Nearest to his: her, their, my, its, your, the, personal, our,\n",
      "Nearest to that: which, however, what, typically, this, lanthanides, where, there,\n",
      "Nearest to most: more, less, many, lively, particularly, some, sponsors, highly,\n",
      "Nearest to other: multiple, different, others, older, various, smokey, reverence, battling,\n",
      "Nearest to only: posteriori, always, polarizing, insured, ouster, usually, last, still,\n",
      "Nearest to i: we, heaney, ii, stockton, carpenters, meo, molten, intrinsic,\n",
      "Nearest to a: palme, another, quite, the, enough, generalizes, ghazni, stijl,\n",
      "Nearest to been: become, be, remained, experienced, was, already, grown, recently,\n",
      "Nearest to four: five, seven, six, eight, three, nine, zero, two,\n",
      "Average loss at step 72000: 2.863852\n",
      "Average loss at step 74000: 2.830623\n",
      "Average loss at step 76000: 2.820694\n",
      "Average loss at step 78000: 2.827307\n",
      "Average loss at step 80000: 2.852228\n",
      "Nearest to states: kingdom, nations, state, countries, woodland, poseidon, horn, robotics,\n",
      "Nearest to b: c, d, j, cali, rede, cl, dxf, stf,\n",
      "Nearest to american: british, french, italian, russian, astro, scottish, korean, indian,\n",
      "Nearest to people: men, women, children, baptists, individuals, others, jews, cultures,\n",
      "Nearest to he: she, it, they, there, we, probably, originally, who,\n",
      "Nearest to world: u, uk, vanquished, philadelphia, oedipus, posts, carols, kush,\n",
      "Nearest to will: would, can, could, must, cannot, should, may, might,\n",
      "Nearest to his: her, their, my, your, its, our, personal, the,\n",
      "Nearest to that: which, however, what, crystallize, lanthanides, strat, roving, adr,\n",
      "Nearest to most: some, more, less, many, among, especially, sponsors, particularly,\n",
      "Nearest to other: others, older, various, sala, fida, coping, smokey, domesticated,\n",
      "Nearest to only: last, always, polarizing, usually, rask, comparisons, eurofighter, either,\n",
      "Nearest to i: ii, we, slums, j, mamre, g, sa, heaney,\n",
      "Nearest to a: another, the, generalizes, palme, grafton, every, little, gcd,\n",
      "Nearest to been: become, be, remained, was, already, experienced, sioux, occurred,\n",
      "Nearest to four: six, five, eight, three, seven, nine, zero, two,\n",
      "Average loss at step 82000: 2.885642\n",
      "Average loss at step 84000: 2.889056\n",
      "Average loss at step 86000: 2.853636\n",
      "Average loss at step 88000: 2.809261\n",
      "Average loss at step 90000: 2.817901\n",
      "Nearest to states: nations, kingdom, state, countries, us, hallucinations, constitutions, curdled,\n",
      "Nearest to b: l, meo, c, stf, mez, h, lowercase, ramayana,\n",
      "Nearest to american: italian, british, russian, australian, canadian, korean, americans, indian,\n",
      "Nearest to people: women, children, baptists, students, person, players, americans, man,\n",
      "Nearest to he: she, it, they, there, originally, we, who, twh,\n",
      "Nearest to world: u, copula, uk, napoleonic, supreme, carols, acheson, savant,\n",
      "Nearest to will: must, would, can, could, cannot, may, might, should,\n",
      "Nearest to his: her, their, its, my, your, our, the, whose,\n",
      "Nearest to that: which, however, lanthanides, abruptly, what, nevado, vai, relaxing,\n",
      "Nearest to most: many, some, less, more, particularly, groves, all, among,\n",
      "Nearest to other: various, sala, coping, multiple, industrial, individual, specific, older,\n",
      "Nearest to only: last, even, kristina, polarizing, no, always, chiefly, standardised,\n",
      "Nearest to i: g, ii, iii, vii, you, we, irreplaceable, iv,\n",
      "Nearest to a: the, every, another, supermarkets, any, generalizes, palme, airtime,\n",
      "Nearest to been: become, remained, was, already, previously, be, being, experienced,\n",
      "Nearest to four: seven, five, three, eight, six, two, nine, zero,\n",
      "Average loss at step 92000: 2.862007\n",
      "Average loss at step 94000: 2.687684\n",
      "Average loss at step 96000: 2.832134\n",
      "Average loss at step 98000: 2.690173\n",
      "Average loss at step 100000: 2.815958\n",
      "Nearest to states: nations, kingdom, countries, state, us, pedagogy, companies, andalusia,\n",
      "Nearest to b: d, pardo, landholdings, rey, vah, geologist, durian, kenneth,\n",
      "Nearest to american: british, italian, canadian, russian, americans, australian, swiss, french,\n",
      "Nearest to people: children, women, countries, men, players, others, individuals, cultures,\n",
      "Nearest to he: she, it, they, there, who, we, never, nobita,\n",
      "Nearest to world: u, oedipus, uk, napoleonic, season, country, finest, eurosceptic,\n",
      "Nearest to will: would, can, must, could, cannot, should, may, might,\n",
      "Nearest to his: her, their, my, your, our, its, the, whose,\n",
      "Nearest to that: which, what, however, nevertheless, nevado, strat, rst, cortex,\n",
      "Nearest to most: more, less, highly, both, especially, lively, heir, some,\n",
      "Nearest to other: others, older, smokey, various, individual, coping, sala, rather,\n",
      "Nearest to only: polarizing, last, best, ouster, normally, chiefly, no, reciprocate,\n",
      "Nearest to i: ii, we, you, iii, they, g, vii, antonin,\n",
      "Nearest to a: another, the, plated, any, palme, braddock, enough, historiographical,\n",
      "Nearest to been: become, remained, be, previously, already, was, were, recently,\n",
      "Nearest to four: seven, five, six, eight, nine, two, three, zero,\n"
     ]
    }
   ],
   "source": [
    "num_steps = 100001\n",
    "\n",
    "with tf.Session(graph=graph) as session:\n",
    "    tf.initialize_all_variables().run()\n",
    "    print('Initialized')\n",
    "    average_loss = 0\n",
    "    for step in range(num_steps):\n",
    "        batch_data, batch_labels = generate_batch_cbow(batch_size, skip_window)\n",
    "        feed_dict = {train_dataset : batch_data, train_labels : batch_labels}\n",
    "        _, l = session.run([optimizer, loss], feed_dict=feed_dict)\n",
    "        average_loss += l\n",
    "        if step % 2000 == 0:\n",
    "            if step > 0:\n",
    "                average_loss = average_loss / 2000\n",
    "            # The average loss is an estimate of the loss over the last 2000 batches.\n",
    "            print('Average loss at step %d: %f' % (step, average_loss))\n",
    "            average_loss = 0\n",
    "        # note that this is expensive (~20% slowdown if computed every 500 steps)\n",
    "        if step % 10000 == 0:\n",
    "            sim = similarity.eval()\n",
    "            for i in range(valid_size):\n",
    "                valid_word = reverse_dictionary[valid_examples[i]]\n",
    "                top_k = 8 # number of nearest neighbors\n",
    "                nearest = (-sim[i, :]).argsort()[1:top_k+1]\n",
    "                log = 'Nearest to %s:' % valid_word\n",
    "                for k in range(top_k):\n",
    "                    close_word = reverse_dictionary[nearest[k]]\n",
    "                    log = '%s %s,' % (log, close_word)\n",
    "                print(log)\n",
    "    final_embeddings = normalized_embeddings.eval()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "cellView": "both",
    "colab": {
     "autoexec": {
      "startup": false,
      "wait_interval": 0
     }
    },
    "colab_type": "code",
    "collapsed": false,
    "id": "jjJXYA_XzV79"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "True\n"
     ]
    }
   ],
   "source": [
    "num_points = 100\n",
    "\n",
    "tsne = TSNE(perplexity=30, n_components=2, init='pca', n_iter=5000)\n",
    "two_d_embeddings = tsne.fit_transform(final_embeddings[1:num_points+1, :])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "cellView": "both",
    "colab": {
     "autoexec": {
      "startup": false,
      "wait_interval": 0
     },
     "output_extras": [
      {
       "item_id": 1
      }
     ]
    },
    "colab_type": "code",
    "collapsed": false,
    "executionInfo": {
     "elapsed": 4763,
     "status": "ok",
     "timestamp": 1445965465525,
     "user": {
      "color": "#1FA15D",
      "displayName": "Vincent Vanhoucke",
      "isAnonymous": false,
      "isMe": true,
      "permissionId": "05076109866853157986",
      "photoUrl": "//lh6.googleusercontent.com/-cCJa7dTDcgQ/AAAAAAAAAAI/AAAAAAAACgw/r2EZ_8oYer4/s50-c-k-no/photo.jpg",
      "sessionId": "2f1ffade4c9f20de",
      "userId": "102167687554210253930"
     },
     "user_tz": 420
    },
    "id": "o_e0D_UezcDe",
    "outputId": "df22e4a5-e8ec-4e5e-d384-c6cf37c68c34"
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA3wAAANmCAYAAAChWYnfAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nBwAA0CmBDwAAoFP/H2MO0dkulYkXAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x12321be90>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "def plot(embeddings, labels):\n",
    "  assert embeddings.shape[0] >= len(labels), 'More labels than embeddings'\n",
    "  pylab.figure(figsize=(15,15))  # in inches\n",
    "  for i, label in enumerate(labels):\n",
    "    x, y = embeddings[i,:]\n",
    "    pylab.scatter(x, y)\n",
    "    pylab.annotate(label, xy=(x, y), xytext=(5, 2), textcoords='offset points',\n",
    "                   ha='right', va='bottom')\n",
    "  pylab.show()\n",
    "\n",
    "words = [reverse_dictionary[i] for i in range(1, num_points+1)]\n",
    "plot(two_d_embeddings, words)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "colab": {
   "default_view": {},
   "name": "5_word2vec.ipynb",
   "provenance": [],
   "version": "0.3.2",
   "views": {}
  },
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.11"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}