char-rnn.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true,
    "deletable": true,
    "editable": true
   },
   "outputs": [],
   "source": [
    "import glob\n",
    "import unicodedata\n",
    "import string"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['./data/names/Arabic.txt', './data/names/English.txt', './data/names/German.txt', './data/names/Vietnamese.txt', './data/names/Italian.txt', './data/names/Czech.txt', './data/names/Scottish.txt', './data/names/Dutch.txt', './data/names/Korean.txt', './data/names/Chinese.txt', './data/names/Greek.txt', './data/names/Irish.txt', './data/names/French.txt', './data/names/Portuguese.txt', './data/names/Russian.txt', './data/names/Japanese.txt', './data/names/Polish.txt', './data/names/Spanish.txt']\n"
     ]
    }
   ],
   "source": [
    "def find_files(path): return glob.glob(path)\n",
    "print(find_files('./data/names/*.txt'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Slusarski\n"
     ]
    }
   ],
   "source": [
    "all_letters = string.ascii_letters + \" .,;'\"\n",
    "n_letters = len(all_letters)\n",
    "\n",
    "# Turn a Unicode string to plain ASCII, thanks to http://stackoverflow.com/a/518232/2809427\n",
    "def unicode2ascii(s):\n",
    "    return ''.join(\n",
    "        c for c in unicodedata.normalize('NFD', s)\n",
    "        if unicodedata.category(c) != 'Mn'\n",
    "        and c in all_letters\n",
    "    )\n",
    "print(unicode2ascii(\"Ślusàrski\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# Build the category_lines dictionary, a list of names per language\n",
    "category_lines = {}\n",
    "all_categories = []\n",
    "\n",
    "# Read a file and split into lines\n",
    "def read_lines(filename):\n",
    "    lines = open(filename).read().strip().split('\\n')\n",
    "    return [unicode2ascii(line) for line in lines]\n",
    "\n",
    "for filename in find_files('./data/names/*.txt'):\n",
    "    category = filename.split('/')[-1].split('.')[0]\n",
    "    all_categories.append(category)\n",
    "    lines = read_lines(filename)\n",
    "    category_lines[category] = lines\n",
    "\n",
    "n_categories = len(all_categories)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['Ahn', 'Baik', 'Bang', 'Byon', 'Cha']\n"
     ]
    }
   ],
   "source": [
    "print(category_lines['Korean'][:5])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import torch\n",
    "\n",
    "# Find letter index from all_letters, e.g. \"a\" = 0\n",
    "def letter2index(letter):\n",
    "    return all_letters.find(letter)\n",
    "\n",
    "# Just for demonstration, turn a letter into a <1 x n_letters> Tensor\n",
    "def letter2tensor(letter):\n",
    "    tensor = torch.zeros(1, n_letters)\n",
    "    tensor[0][letter2index(letter)] = 1\n",
    "    return tensor\n",
    "\n",
    "# Turn a line into a <line_length x 1 x n_letters>,\n",
    "# or an array of one-hot letter vectors\n",
    "def line2tensor(line):\n",
    "    tensor = torch.zeros(len(line), 1, n_letters)\n",
    "    for li, letter in enumerate(line):\n",
    "        tensor[li][0][letter2index(letter)] = 1\n",
    "    return tensor"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "\n",
      "Columns 0 to 12 \n",
      "    0     0     0     0     0     0     0     0     0     0     0     0     0\n",
      "\n",
      "Columns 13 to 25 \n",
      "    0     0     0     0     0     0     0     0     0     0     0     0     0\n",
      "\n",
      "Columns 26 to 38 \n",
      "    0     0     0     0     0     0     0     0     0     1     0     0     0\n",
      "\n",
      "Columns 39 to 51 \n",
      "    0     0     0     0     0     0     0     0     0     0     0     0     0\n",
      "\n",
      "Columns 52 to 56 \n",
      "    0     0     0     0     0\n",
      "[torch.FloatTensor of size 1x57]\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(letter2tensor('J'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "torch.Size([4, 1, 57])\n"
     ]
    }
   ],
   "source": [
    "print(line2tensor('keon').size())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "n_letters: 57 n_hidden: 128 n_categories: 18\n",
      "RNN (\n",
      "  (i2h): Linear (185 -> 128)\n",
      "  (i2o): Linear (185 -> 18)\n",
      "  (softmax): LogSoftmax ()\n",
      ")\n"
     ]
    }
   ],
   "source": [
    "import torch.nn as nn\n",
    "from torch.autograd import Variable\n",
    "\n",
    "class RNN(nn.Module):\n",
    "    def __init__(self, input_size, hidden_size, output_size):\n",
    "        super(RNN, self).__init__()\n",
    "        \n",
    "        self.hidden_size = hidden_size\n",
    "        \n",
    "        self.i2h = nn.Linear(input_size + hidden_size, hidden_size)\n",
    "        self.i2o = nn.Linear(input_size + hidden_size, output_size)\n",
    "        self.softmax = nn.LogSoftmax()\n",
    "    \n",
    "    def forward(self, input, hidden):\n",
    "        combined = torch.cat((input, hidden), 1)\n",
    "        hidden = self.i2h(combined)\n",
    "        output = self.i2o(combined)\n",
    "        output = self.softmax(output)\n",
    "        return output, hidden\n",
    "\n",
    "    def init_hidden(self):\n",
    "        return Variable(torch.zeros(1, self.hidden_size))\n",
    "\n",
    "n_hidden = 128\n",
    "rnn = RNN(n_letters, n_hidden, n_categories)\n",
    "print(\"n_letters:\", n_letters, \"n_hidden:\", n_hidden, \"n_categories:\", n_categories)\n",
    "print(rnn)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Variable containing:\n",
      "\n",
      "Columns 0 to 9 \n",
      "-2.8618 -2.8753 -2.8834 -2.9676 -2.9437 -2.9598 -2.9653 -2.7539 -2.9182 -2.9475\n",
      "\n",
      "Columns 10 to 17 \n",
      "-2.9105 -2.7830 -2.9466 -2.8541 -2.9486 -2.8439 -2.8831 -2.8162\n",
      "[torch.FloatTensor of size 1x18]\n",
      "\n"
     ]
    }
   ],
   "source": [
    "input = Variable(line2tensor('Albert'))\n",
    "hidden = Variable(torch.zeros(1, n_hidden))\n",
    "\n",
    "output, next_hidden = rnn(input[0], hidden)\n",
    "print(output)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "('Dutch', 7)\n"
     ]
    }
   ],
   "source": [
    "def category_from_output(output):\n",
    "    top_n, top_i = output.data.topk(1) # Tensor out of Variable with .data\n",
    "    category_i = top_i[0][0]\n",
    "    return all_categories[category_i], category_i\n",
    "print(category_from_output(output))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "category : Polish  \tname: Rusnak\n",
      "category : Czech  \tname: Silhan\n",
      "category : Japanese  \tname: Ihara\n",
      "category : Japanese  \tname: Sassa\n",
      "category : Portuguese  \tname: Castro\n",
      "category : Irish  \tname: Maurice\n",
      "category : Scottish  \tname: Davidson\n",
      "category : Polish  \tname: Sokolsky\n",
      "category : Portuguese  \tname: Maria\n",
      "category : Spanish  \tname: Castillo\n"
     ]
    }
   ],
   "source": [
    "import random\n",
    "\n",
    "def choose(l):\n",
    "    return l[random.randint(0, len(l) - 1)]\n",
    "\n",
    "def training_pair():                                                                                                               \n",
    "    category = choose(all_categories)\n",
    "    line = choose(category_lines[category])\n",
    "    category_tensor = Variable(torch.LongTensor([all_categories.index(category)]))\n",
    "    line_tensor = Variable(line2tensor(line))\n",
    "    return category, line, category_tensor, line_tensor\n",
    "\n",
    "for i in range(10):\n",
    "    category, line, category_tensor, line_tensor = training_pair()\n",
    "    print('category :', category, ' \\tname:', line)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "criterion = nn.NLLLoss()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "learning_rate = 0.005\n",
    "def train (categroy_tensor, line_tensor):\n",
    "    hidden = rnn.init_hidden()\n",
    "    rnn.zero_grad()\n",
    "    \n",
    "    for i in range(line_tensor.size()[0]):\n",
    "        output, hidden = rnn(line_tensor[i], hidden)\n",
    "    loss = criterion(output, category_tensor)\n",
    "    loss.backward()\n",
    "    \n",
    "    for p in rnn.parameters():\n",
    "        p.data.add_(-learning_rate, p.grad.data)\n",
    "    return output, loss.data[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import time\n",
    "import math"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "5000 5% (0m 7s) 2.4473 Tholberg / Russian ✗ (Dutch)\n",
      "10000 10% (0m 14s) 2.9370 Higginson / Russian ✗ (English)\n",
      "15000 15% (0m 22s) 2.5217 Muir / Korean ✗ (Scottish)\n",
      "20000 20% (0m 29s) 1.0506 Reuter / German ✓\n",
      "25000 25% (0m 36s) 2.2771 Korycansky / Polish ✗ (Czech)\n",
      "30000 30% (0m 42s) 4.4783 Daal / Vietnamese ✗ (Dutch)\n",
      "35000 35% (0m 49s) 1.4944 Bang / Chinese ✗ (Korean)\n",
      "40000 40% (0m 57s) 1.8069 Nymmik / Polish ✗ (Russian)\n",
      "45000 45% (1m 4s) 1.4673 San / Chinese ✗ (Korean)\n",
      "50000 50% (1m 11s) 4.1062 Tovma / Spanish ✗ (Russian)\n",
      "55000 55% (1m 18s) 0.4954 Suk / Korean ✓\n",
      "60000 60% (1m 24s) 0.1622 O'Donoghue / Irish ✓\n",
      "65000 65% (1m 31s) 1.5637 Kattan / Arabic ✓\n",
      "70000 70% (1m 38s) 0.0953 Lupekhin / Russian ✓\n",
      "75000 75% (1m 45s) 1.2902 Artz / Spanish ✗ (German)\n",
      "80000 80% (1m 52s) 0.3286 Ankeren / Dutch ✓\n",
      "85000 85% (1m 58s) 3.0084 Leggett / French ✗ (English)\n",
      "90000 90% (2m 6s) 0.2628 Dam / Vietnamese ✓\n",
      "95000 95% (2m 12s) 1.3248 Vennen / English ✗ (Dutch)\n",
      "100000 100% (2m 19s) 1.2330 Ton / Chinese ✗ (Vietnamese)\n"
     ]
    }
   ],
   "source": [
    "n_epochs = 100000\n",
    "print_every = 5000\n",
    "plot_every = 1000\n",
    "\n",
    "rnn = RNN(n_letters, n_hidden, n_categories)\n",
    "\n",
    "# Keep track of losses for plotting\n",
    "current_loss = 0\n",
    "all_losses = []\n",
    "\n",
    "def time_since(since):\n",
    "    now = time.time()\n",
    "    s = now - since\n",
    "    m = math.floor(s / 60)\n",
    "    s -= m * 60\n",
    "    return '%dm %ds' % (m, s)\n",
    "\n",
    "start = time.time()\n",
    "\n",
    "for epoch in range(1, n_epochs + 1):\n",
    "    category, line, category_tensor, line_tensor = training_pair()\n",
    "    output, loss = train(category_tensor, line_tensor)\n",
    "    current_loss += loss\n",
    "\n",
    "    # Print epoch number, loss, name and guess\n",
    "    if epoch % print_every == 0:\n",
    "        guess, guess_i = category_from_output(output)\n",
    "        correct = '✓' if guess == category else '✗ (%s)' % category\n",
    "        print('%d %d%% (%s) %.4f %s / %s %s' % (epoch, epoch / n_epochs * 100, time_since(start), loss, line, guess, correct))\n",
    "\n",
    "    # Add current loss avg to list of losses\n",
    "    if epoch % plot_every == 0:\n",
    "        all_losses.append(current_loss / plot_every)\n",
    "        current_loss = 0\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x7f3f2f3a7a20>]"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAX4AAAD8CAYAAABw1c+bAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f3f326bac88>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import matplotlib.ticker as ticker\n",
    "%matplotlib inline\n",
    "\n",
    "plt.figure()\n",
    "plt.plot(all_losses)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAVwAAAEwCAYAAAD7IMkNAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f3f372a3ef0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Keep track of correct guesses in a confusion matrix\n",
    "confusion = torch.zeros(n_categories, n_categories)\n",
    "n_confusion = 10000\n",
    "\n",
    "# Just return an output given a line\n",
    "def evaluate(line_tensor):\n",
    "    hidden = rnn.init_hidden()\n",
    "    \n",
    "    for i in range(line_tensor.size()[0]):\n",
    "        output, hidden = rnn(line_tensor[i], hidden)\n",
    "    \n",
    "    return output\n",
    "\n",
    "# Go through a bunch of examples and record which are correctly guessed\n",
    "for i in range(n_confusion):\n",
    "    category, line, category_tensor, line_tensor = training_pair()\n",
    "    output = evaluate(line_tensor)\n",
    "    guess, guess_i = category_from_output(output)\n",
    "    category_i = all_categories.index(category)\n",
    "    confusion[category_i][guess_i] += 1\n",
    "\n",
    "# Normalize by dividing every row by its sum\n",
    "for i in range(n_categories):\n",
    "    confusion[i] = confusion[i] / confusion[i].sum()\n",
    "\n",
    "# Set up plot\n",
    "fig = plt.figure()\n",
    "ax = fig.add_subplot(111)\n",
    "cax = ax.matshow(confusion.numpy())\n",
    "fig.colorbar(cax)\n",
    "\n",
    "# Set up axes\n",
    "ax.set_xticklabels([''] + all_categories, rotation=90)\n",
    "ax.set_yticklabels([''] + all_categories)\n",
    "\n",
    "# Force label at every tick\n",
    "ax.xaxis.set_major_locator(ticker.MultipleLocator(1))\n",
    "ax.yaxis.set_major_locator(ticker.MultipleLocator(1))\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "> Dovesky\n",
      "(-1.03) Czech\n",
      "(-1.25) Russian\n",
      "(-1.39) Polish\n",
      "\n",
      "> Jackson\n",
      "(-0.21) Scottish\n",
      "(-2.41) English\n",
      "(-3.27) Russian\n",
      "\n",
      "> Satoshi\n",
      "(-1.16) Italian\n",
      "(-1.68) Arabic\n",
      "(-1.77) Japanese\n"
     ]
    }
   ],
   "source": [
    "\n",
    "def predict(input_line, n_predictions=3):\n",
    "    print('\\n> %s' % input_line)\n",
    "    output = evaluate(Variable(line2tensor(input_line)))\n",
    "\n",
    "    # Get top N categories\n",
    "    topv, topi = output.data.topk(n_predictions, 1, True)\n",
    "    predictions = []\n",
    "\n",
    "    for i in range(n_predictions):\n",
    "        value = topv[0][i]\n",
    "        category_index = topi[0][i]\n",
    "        print('(%.2f) %s' % (value, all_categories[category_index]))\n",
    "        predictions.append([value, all_categories[category_index]])\n",
    "\n",
    "predict('Dovesky')\n",
    "predict('Jackson')\n",
    "predict('Satoshi')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.5.2+"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}