lesson6-rnn-Bidir-Demo.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 151,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-21T15:17:37.213036Z",
     "start_time": "2018-04-21T15:17:34.000707Z"
    }
   },
   "outputs": [],
   "source": [
    "%reload_ext autoreload\n",
    "%autoreload 2\n",
    "%matplotlib inline\n",
    "\n",
    "from fastai.io import *\n",
    "from fastai.conv_learner import *\n",
    "\n",
    "\n",
    "from fastai.column_data import *"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Setup"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We're going to download the collected works of Nietzsche to use as our data for this class."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 152,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-21T15:17:37.523860Z",
     "start_time": "2018-04-21T15:17:37.217509Z"
    }
   },
   "outputs": [],
   "source": [
    "PATH='data/nietzsche/'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 153,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-21T15:17:37.839203Z",
     "start_time": "2018-04-21T15:17:37.527845Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "corpus length: 600893\n"
     ]
    }
   ],
   "source": [
    "get_data(\"https://s3.amazonaws.com/text-datasets/nietzsche.txt\", f'{PATH}nietzsche.txt')\n",
    "text = open(f'{PATH}nietzsche.txt',encoding='utf-8').read()\n",
    "print('corpus length:', len(text))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 154,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-21T15:17:38.186616Z",
     "start_time": "2018-04-21T15:17:37.844676Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'PREFACE\\n\\n\\nSUPPOSING that Truth is a woman--what then? Is there not ground\\nfor suspecting that all philosophers, in so far as they have been\\ndogmatists, have failed to understand women--that the terrible\\nseriousness and clumsy importunity with which they have usually paid\\ntheir addresses to Truth, have been unskilled and unseemly methods for\\nwinning a woman? Certainly she has never allowed herself '"
      ]
     },
     "execution_count": 154,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "text[:400]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 155,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-21T15:17:38.532068Z",
     "start_time": "2018-04-21T15:17:38.192622Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "total chars: 85\n"
     ]
    }
   ],
   "source": [
    "chars = sorted(list(set(text)))\n",
    "vocab_size = len(chars)+1\n",
    "print('total chars:', vocab_size)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Sometimes it's useful to have a zero value in the dataset, e.g. for padding"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 156,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-21T15:17:39.087074Z",
     "start_time": "2018-04-21T15:17:38.808260Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'\\x00\\n !\"\\'(),-.0123456789:;=?ABCDEFGHIJKLMNOPQRSTUVWXYZ[]_abcdefghijklmnopqrstuvwxy'"
      ]
     },
     "execution_count": 156,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "chars.insert(0, \"\\0\")\n",
    "\n",
    "''.join(chars[0:-6])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Map from chars to indices and back again"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 157,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-21T15:17:39.399905Z",
     "start_time": "2018-04-21T15:17:39.095153Z"
    }
   },
   "outputs": [],
   "source": [
    "char_indices = {c: i for i, c in enumerate(chars)}\n",
    "indices_char = {i: c for i, c in enumerate(chars)}"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "*idx* will be the data we use from now on - it simply converts all the characters to their index (based on the mapping above)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 158,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-21T15:17:39.804519Z",
     "start_time": "2018-04-21T15:17:39.404438Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[40, 42, 29, 30, 25, 27, 29, 1, 1, 1]"
      ]
     },
     "execution_count": 158,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "idx = [char_indices[c] for c in text]\n",
    "\n",
    "idx[:10]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 159,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-21T15:17:40.110795Z",
     "start_time": "2018-04-21T15:17:39.811700Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "600893"
      ]
     },
     "execution_count": 159,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(idx)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 160,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-21T15:17:40.421625Z",
     "start_time": "2018-04-21T15:17:40.115307Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'PREFACE\\n\\n\\nSUPPOSING that Truth is a woman--what then? Is there not gro'"
      ]
     },
     "execution_count": 160,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "''.join(indices_char[i] for i in idx[:70])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 161,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-21T15:18:24.636494Z",
     "start_time": "2018-04-21T15:18:24.378325Z"
    }
   },
   "outputs": [],
   "source": [
    "def get_next(inp):\n",
    "    idxs = T(np.array([char_indices[c] for c in inp]))\n",
    "    p = m(*VV(idxs))\n",
    "    i = np.argmax(to_np(p))\n",
    "    return chars[i]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Create inputs"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This is the size of our unrolled RNN."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 162,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-21T15:18:26.482677Z",
     "start_time": "2018-04-21T15:18:26.115675Z"
    }
   },
   "outputs": [],
   "source": [
    "cs=8"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "For each of 0 through 7, create a list of every 8th character with that starting point. These will be the 8 inputs to our model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 163,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-21T15:18:26.900934Z",
     "start_time": "2018-04-21T15:18:26.486689Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "600885"
      ]
     },
     "execution_count": 163,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(idx)-cs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 209,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-21T15:18:31.302604Z",
     "start_time": "2018-04-21T15:18:26.905953Z"
    }
   },
   "outputs": [],
   "source": [
    "c_in_dat = [[idx[i+j] for i in range(cs)] for j in range(len(idx)-cs)]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Then create a list of the next character in each of these series. This will be the labels for our model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 210,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-21T15:18:31.896768Z",
     "start_time": "2018-04-21T15:18:31.307629Z"
    }
   },
   "outputs": [],
   "source": [
    "c_out_dat = [idx[j+cs] for j in range(len(idx)-cs)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 166,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-21T15:18:39.947205Z",
     "start_time": "2018-04-21T15:18:31.903975Z"
    }
   },
   "outputs": [],
   "source": [
    "xs = np.stack(c_in_dat, axis=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 167,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-21T15:18:40.317559Z",
     "start_time": "2018-04-21T15:18:39.951254Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[40, 42, 29, ..., 27, 29,  1],\n",
       "       [42, 29, 30, ..., 29,  1,  1],\n",
       "       [29, 30, 25, ...,  1,  1,  1],\n",
       "       ...,\n",
       "       [72, 62, 67, ..., 65, 67, 58],\n",
       "       [62, 67, 59, ..., 67, 58, 72],\n",
       "       [67, 59, 74, ..., 58, 72, 72]])"
      ]
     },
     "execution_count": 167,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "xs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 168,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-21T15:18:40.941019Z",
     "start_time": "2018-04-21T15:18:40.323799Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(600885, 8)"
      ]
     },
     "execution_count": 168,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "xs.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 169,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-21T15:18:47.206622Z",
     "start_time": "2018-04-21T15:18:40.945332Z"
    }
   },
   "outputs": [],
   "source": [
    "y = np.stack(c_out_dat)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 170,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-21T15:18:47.515364Z",
     "start_time": "2018-04-21T15:18:47.209128Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([ 1,  1, 43, ..., 72, 72, 10])"
      ]
     },
     "execution_count": 170,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "So each column below is one series of 8 characters from the text."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 171,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-21T15:18:47.878361Z",
     "start_time": "2018-04-21T15:18:47.518497Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[40, 42, 29, 30, 25, 27, 29,  1],\n",
       "       [42, 29, 30, 25, 27, 29,  1,  1],\n",
       "       [29, 30, 25, 27, 29,  1,  1,  1],\n",
       "       [30, 25, 27, 29,  1,  1,  1, 43],\n",
       "       [25, 27, 29,  1,  1,  1, 43, 45],\n",
       "       [27, 29,  1,  1,  1, 43, 45, 40],\n",
       "       [29,  1,  1,  1, 43, 45, 40, 40],\n",
       "       [ 1,  1,  1, 43, 45, 40, 40, 39]])"
      ]
     },
     "execution_count": 171,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "xs[:cs,:cs]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "...and this is the next character after each sequence."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 172,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-21T15:18:48.255880Z",
     "start_time": "2018-04-21T15:18:47.883365Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([ 1,  1, 43, 45, 40, 40, 39, 43])"
      ]
     },
     "execution_count": 172,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y[:cs]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Create and train model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 173,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-21T15:18:48.678477Z",
     "start_time": "2018-04-21T15:18:48.261870Z"
    }
   },
   "outputs": [],
   "source": [
    "a= list(range(10))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 174,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-21T15:18:48.998857Z",
     "start_time": "2018-04-21T15:18:48.682990Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[0, 1, 2, 3, 4, 5, 6, 7, 8, 9]"
      ]
     },
     "execution_count": 174,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "a"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 175,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-21T15:18:49.363113Z",
     "start_time": "2018-04-21T15:18:49.003371Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([8, 1])"
      ]
     },
     "execution_count": 175,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "get_cv_idxs(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 176,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-21T15:18:49.851472Z",
     "start_time": "2018-04-21T15:18:49.367146Z"
    }
   },
   "outputs": [],
   "source": [
    "val_idx = get_cv_idxs(len(idx)-cs-1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 177,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-21T15:18:50.259075Z",
     "start_time": "2018-04-21T15:18:49.854480Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(120176,)"
      ]
     },
     "execution_count": 177,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "val_idx.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 178,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-21T15:18:50.572457Z",
     "start_time": "2018-04-21T15:18:50.263589Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<function fastai.dataset.get_cv_idxs>"
      ]
     },
     "execution_count": 178,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "get_cv_idxs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 179,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-21T15:18:50.881288Z",
     "start_time": "2018-04-21T15:18:50.575965Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([480310, 419017, 232803, ..., 134355, 389158, 330599])"
      ]
     },
     "execution_count": 179,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "val_idx"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 180,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-21T15:18:51.511000Z",
     "start_time": "2018-04-21T15:18:50.887796Z"
    }
   },
   "outputs": [],
   "source": [
    "md = ColumnarModelData.from_arrays('.', val_idx, xs, y, bs=512)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Test model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 181,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-21T15:22:31.005589Z",
     "start_time": "2018-04-21T15:22:30.709264Z"
    }
   },
   "outputs": [],
   "source": [
    "def get_next(inp):\n",
    "    idxs = T(np.array([char_indices[c] for c in inp]))\n",
    "    p = m(*VV(idxs))\n",
    "    i = np.argmax(to_np(p))\n",
    "    return chars[i]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 182,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-21T15:23:23.826518Z",
     "start_time": "2018-04-21T15:23:23.680762Z"
    }
   },
   "outputs": [],
   "source": [
    "def get_next_n(inp, n):\n",
    "    res = inp\n",
    "    for i in range(n):\n",
    "        c = get_next(inp)\n",
    "        res += c\n",
    "        inp = inp[1:]+c\n",
    "    return res"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "## Multi-output model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Setup"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's take non-overlapping sets of characters this time"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 183,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-21T15:23:24.704853Z",
     "start_time": "2018-04-21T15:23:24.144475Z"
    }
   },
   "outputs": [],
   "source": [
    "c_in_dat = [[idx[i+j] for i in range(cs)] for j in range(0, len(idx)-cs-1, cs)]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Then create the exact same thing, offset by 1, as our labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 184,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-21T15:23:24.988250Z",
     "start_time": "2018-04-21T15:23:24.707358Z"
    }
   },
   "outputs": [],
   "source": [
    "c_out_dat = [[idx[i+j] for i in range(cs)] for j in range(1, len(idx)-cs, cs)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 185,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-21T15:23:25.507520Z",
     "start_time": "2018-04-21T15:23:24.991113Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(75111, 8)"
      ]
     },
     "execution_count": 185,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "xs = np.stack(c_in_dat)\n",
    "xs.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 186,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-21T15:23:26.010824Z",
     "start_time": "2018-04-21T15:23:25.509993Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(75111, 8)"
      ]
     },
     "execution_count": 186,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ys = np.stack(c_out_dat)\n",
    "ys.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 187,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-21T15:23:26.194311Z",
     "start_time": "2018-04-21T15:23:26.014334Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[40, 42, 29, 30, 25, 27, 29,  1],\n",
       "       [ 1,  1, 43, 45, 40, 40, 39, 43],\n",
       "       [33, 38, 31,  2, 73, 61, 54, 73],\n",
       "       [ 2, 44, 71, 74, 73, 61,  2, 62],\n",
       "       [72,  2, 54,  2, 76, 68, 66, 54],\n",
       "       [67,  9,  9, 76, 61, 54, 73,  2],\n",
       "       [73, 61, 58, 67, 24,  2, 33, 72],\n",
       "       [ 2, 73, 61, 58, 71, 58,  2, 67]])"
      ]
     },
     "execution_count": 187,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "xs[:cs,:cs]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 188,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-21T15:23:26.385319Z",
     "start_time": "2018-04-21T15:23:26.196818Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[42, 29, 30, 25, 27, 29,  1,  1],\n",
       "       [ 1, 43, 45, 40, 40, 39, 43, 33],\n",
       "       [38, 31,  2, 73, 61, 54, 73,  2],\n",
       "       [44, 71, 74, 73, 61,  2, 62, 72],\n",
       "       [ 2, 54,  2, 76, 68, 66, 54, 67],\n",
       "       [ 9,  9, 76, 61, 54, 73,  2, 73],\n",
       "       [61, 58, 67, 24,  2, 33, 72,  2],\n",
       "       [73, 61, 58, 71, 58,  2, 67, 68]])"
      ]
     },
     "execution_count": 188,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ys[:cs,:cs]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Create and train model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 189,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-21T15:51:11.689039Z",
     "start_time": "2018-04-21T15:51:11.546767Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([25679, 44278, 25725, ..., 49739, 71742, 16822])"
      ]
     },
     "execution_count": 189,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "get_cv_idxs(len(xs)-cs-1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 190,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-21T15:23:26.599389Z",
     "start_time": "2018-04-21T15:23:26.390835Z"
    }
   },
   "outputs": [],
   "source": [
    "val_idx = get_cv_idxs(len(xs)-cs-1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 191,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-21T15:23:26.862434Z",
     "start_time": "2018-04-21T15:23:26.604403Z"
    }
   },
   "outputs": [],
   "source": [
    "md = ColumnarModelData.from_arrays('.', val_idx, xs, ys, bs=512)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 192,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-21T15:23:27.646221Z",
     "start_time": "2018-04-21T15:23:27.412951Z"
    }
   },
   "outputs": [],
   "source": [
    "def nll_loss_seq(inp, targ):\n",
    "    sl,bs,nh = inp.size()\n",
    "    targ = targ.transpose(0,1).contiguous().view(-1)\n",
    "    return F.nll_loss(inp.view(-1,nh), targ)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Stateful model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Setup"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 193,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-21T15:23:44.173500Z",
     "start_time": "2018-04-21T15:23:43.666926Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "nietzsche.txt  \u001b[0m\u001b[01;34mtrn\u001b[0m/  \u001b[01;34mval\u001b[0m/\r\n"
     ]
    }
   ],
   "source": [
    "from torchtext import vocab, data\n",
    "\n",
    "from fastai.nlp import *\n",
    "from fastai.lm_rnn import *\n",
    "\n",
    "PATH=Path('data/nietzsche/')\n",
    "\n",
    "TRN_PATH = 'trn/'\n",
    "VAL_PATH = 'val/'\n",
    "TRN = f'{PATH}{TRN_PATH}'\n",
    "VAL = f'{PATH}{VAL_PATH}'\n",
    "\n",
    "%ls {PATH}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 194,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-21T15:23:44.353735Z",
     "start_time": "2018-04-21T15:23:44.176007Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ls: cannot access 'data/nietzschetrn': No such file or directory\r\n"
     ]
    }
   ],
   "source": [
    "%ls {PATH}trn"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 195,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-21T15:23:47.001899Z",
     "start_time": "2018-04-21T15:23:44.355771Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1153, 55, 1, 590960)"
      ]
     },
     "execution_count": 195,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "TEXT = data.Field(lower=True, tokenize=list)\n",
    "bs=64; bptt=8; n_fac=42; n_hidden=256\n",
    "\n",
    "FILES = dict(train=TRN_PATH, validation=VAL_PATH, test=VAL_PATH)\n",
    "md = LanguageModelData.from_text_files(PATH, TEXT, **FILES, bs=bs, bptt=bptt, min_freq=3)\n",
    "\n",
    "len(md.trn_dl), md.nt, len(md.trn_ds), len(md.trn_ds[0].text)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### GRU"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 196,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_next(inp):\n",
    "    idxs = TEXT.numericalize(inp)\n",
    "    p = m(VV(idxs.transpose(0,1)))\n",
    "    r = torch.multinomial(p[-1].exp(), 1)\n",
    "    return TEXT.vocab.itos[to_np(r)[0]]\n",
    "\n",
    "def get_next_n(inp, n):\n",
    "    res = inp\n",
    "    for i in range(n):\n",
    "        c = get_next(inp)\n",
    "        res += c\n",
    "        inp = inp[1:]+c\n",
    "    return res"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Single Direction"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 201,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-21T15:24:42.660787Z",
     "start_time": "2018-04-21T15:24:42.450728Z"
    }
   },
   "outputs": [],
   "source": [
    "class CharSeqStatefulGRU(nn.Module):\n",
    "    def __init__(self, vocab_size, n_fac, bs):\n",
    "        super().__init__()\n",
    "        self.vocab_size = vocab_size\n",
    "        self.e = nn.Embedding(vocab_size, n_fac)\n",
    "        self.rnn = nn.GRU(n_fac, n_hidden)\n",
    "        self.l_out = nn.Linear(n_hidden, vocab_size)\n",
    "        self.init_hidden(bs)\n",
    "        \n",
    "    def forward(self, cs):\n",
    "        bs = cs[0].size(0)\n",
    "        if self.h.size(1) != bs: self.init_hidden(bs)\n",
    "        outp,h = self.rnn(self.e(cs), self.h)\n",
    "        self.h = repackage_var(h)\n",
    "        return F.log_softmax(self.l_out(outp), dim=-1).view(-1, self.vocab_size)\n",
    "    \n",
    "    def init_hidden(self, bs): self.h = V(torch.zeros(1, bs, n_hidden))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 202,
   "metadata": {},
   "outputs": [],
   "source": [
    "m = CharSeqStatefulGRU(md.nt, n_fac, 512).cuda()\n",
    "\n",
    "opt = optim.Adam(m.parameters(), 1e-3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 203,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-21T15:27:43.891611Z",
     "start_time": "2018-04-21T15:25:29.593040Z"
    }
   },
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "b223a18351f544c9bf768a1b1b50b758",
       "version_major": 2,
       "version_minor": 0
      },
      "text/html": [
       "<p>Failed to display Jupyter Widget of type <code>HBox</code>.</p>\n",
       "<p>\n",
       "  If you're reading this message in the Jupyter Notebook or JupyterLab Notebook, it may mean\n",
       "  that the widgets JavaScript is still loading. If this message persists, it\n",
       "  likely means that the widgets JavaScript library is either not installed or\n",
       "  not enabled. See the <a href=\"https://ipywidgets.readthedocs.io/en/stable/user_install.html\">Jupyter\n",
       "  Widgets Documentation</a> for setup instructions.\n",
       "</p>\n",
       "<p>\n",
       "  If you're reading this message in another frontend (for example, a static\n",
       "  rendering on GitHub or <a href=\"https://nbviewer.jupyter.org/\">NBViewer</a>),\n",
       "  it may mean that your frontend doesn't currently support widgets.\n",
       "</p>\n"
      ],
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='Epoch', max=3), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch      trn_loss   val_loss                                 \n",
      "    0      1.675948   1.668035  \n",
      "    1      1.515996   1.50374                                  \n",
      "    2      1.444927   1.435919                                 \n",
      "\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[array([1.43592])]"
      ]
     },
     "execution_count": 203,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "fit(m, md, 3, opt, F.nll_loss)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 204,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "for those should always were not dif-present morality.83=patancem,precisely intentimento,. something enhated, amongtime be was usefulness of which is vailcap, asoot of the warlingsivalitys. like soul just demands \"say finally pendery cast us, a beholder call that will to dangerousand shave all tracte of the somethings!\"\"inexperances, of somelong morate. at the waring and sometexpeldays aboduagable expenia\n"
     ]
    }
   ],
   "source": [
    "print(get_next_n('for thos', 400))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Bi-dircetional"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 205,
   "metadata": {},
   "outputs": [],
   "source": [
    "class CharSeqStatefulGRU_Bidir(nn.Module):\n",
    "    def __init__(self, vocab_size, n_fac, bs):\n",
    "        super().__init__()\n",
    "        self.vocab_size = vocab_size\n",
    "        self.e = nn.Embedding(vocab_size, n_fac)\n",
    "        self.rnn = nn.GRU(n_fac, n_hidden, bidirectional = True)\n",
    "        self.l_out = nn.Linear(n_hidden*2, vocab_size)\n",
    "        self.init_hidden(bs)\n",
    "        \n",
    "    def forward(self, cs):\n",
    "        bs = cs[0].size(0)\n",
    "        if self.h.size(1) != bs: self.init_hidden(bs)\n",
    "        outp,h = self.rnn(self.e(cs), self.h)\n",
    "        self.h = repackage_var(h)\n",
    "        return F.log_softmax(self.l_out(outp), dim=-1).view(-1, self.vocab_size)\n",
    "    \n",
    "    def init_hidden(self, bs): self.h = V(torch.zeros(2, bs, n_hidden))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 211,
   "metadata": {},
   "outputs": [],
   "source": [
    "m = CharSeqStatefulGRU_Bidir(md.nt, n_fac, 512).cuda()\n",
    "\n",
    "opt = optim.Adam(m.parameters(), 1e-3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 212,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "5ffbef948606434d89b609b7e6c0f8e5",
       "version_major": 2,
       "version_minor": 0
      },
      "text/html": [
       "<p>Failed to display Jupyter Widget of type <code>HBox</code>.</p>\n",
       "<p>\n",
       "  If you're reading this message in the Jupyter Notebook or JupyterLab Notebook, it may mean\n",
       "  that the widgets JavaScript is still loading. If this message persists, it\n",
       "  likely means that the widgets JavaScript library is either not installed or\n",
       "  not enabled. See the <a href=\"https://ipywidgets.readthedocs.io/en/stable/user_install.html\">Jupyter\n",
       "  Widgets Documentation</a> for setup instructions.\n",
       "</p>\n",
       "<p>\n",
       "  If you're reading this message in another frontend (for example, a static\n",
       "  rendering on GitHub or <a href=\"https://nbviewer.jupyter.org/\">NBViewer</a>),\n",
       "  it may mean that your frontend doesn't currently support widgets.\n",
       "</p>\n"
      ],
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='Epoch', max=3), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch      trn_loss   val_loss                                  \n",
      "    0      0.489046   0.429586  \n",
      "    1      0.464659   0.384862                                  \n",
      "    2      0.415929   0.362482                                  \n",
      "\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[array([0.36248])]"
      ]
     },
     "execution_count": 212,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "fit(m, md, 3, opt, F.nll_loss)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 213,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "for thoselor alouslooksbourly know,rly know, lonker chilaker chrisker chrisker who soar who holy who holy who holy doover, odorier, oforier, of rath, of rath, deerath, deen the dest the dest this muthoursomuchoursomechoursome has some has some has some has some has some hesesome cosessmence-themence-them jelomops jeposess jecosess jecosess jecosess jecosess jecosess jecosess jecosesmajectses_ajocts simjoc\n"
     ]
    }
   ],
   "source": [
    "print(get_next_n('for thos', 400))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "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": "3.6.4"
  },
  "nav_menu": {},
  "toc": {
   "colors": {
    "hover_highlight": "#DAA520",
    "navigate_num": "#000000",
    "navigate_text": "#333333",
    "running_highlight": "#FF0000",
    "selected_highlight": "#FFD700",
    "sidebar_border": "#EEEEEE",
    "wrapper_background": "#FFFFFF"
   },
   "moveMenuLeft": true,
   "nav_menu": {
    "height": "216px",
    "width": "252px"
   },
   "navigate_menu": true,
   "number_sections": true,
   "sideBar": true,
   "threshold": 4,
   "toc_cell": false,
   "toc_section_display": "block",
   "toc_window_display": false,
   "widenNotebook": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}