Poetry_Generator.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Poetry-Generator.ipynb",
      "provenance": [],
      "authorship_tag": "ABX9TyMiD0o93sAul1fuBe695/YC",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/Arshpreet-Singh-1/Poetry-Generator-Using-NLP/blob/main/Poetry_Generator.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Imports"
      ],
      "metadata": {
        "id": "7pqDHUCRx7JI"
      }
    },
    {
      "cell_type": "code",
      "execution_count": 1,
      "metadata": {
        "id": "Hlxa8Q1oi5aV"
      },
      "outputs": [],
      "source": [
        "import tensorflow as tf\n",
        "\n",
        "from tensorflow.keras.preprocessing.sequence import pad_sequences\n",
        "from tensorflow.keras.layers import Embedding, LSTM, Dense, Bidirectional\n",
        "from tensorflow.keras.preprocessing.text import Tokenizer\n",
        "from tensorflow.keras.models import Sequential\n",
        "from tensorflow.keras.optimizers import Adam\n",
        "import numpy as np"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "!wget --no-check-certificate \\\n",
        "    https://storage.googleapis.com/laurencemoroney-blog.appspot.com/irish-lyrics-eof.txt \\\n",
        "    -O /tmp/irish-lyrics-eof.txt"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "CbsMDKN_jHjC",
        "outputId": "148a558c-fdda-4c87-fe1c-2805813dcd85"
      },
      "execution_count": 2,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "--2022-03-14 04:54:35--  https://storage.googleapis.com/laurencemoroney-blog.appspot.com/irish-lyrics-eof.txt\n",
            "Resolving storage.googleapis.com (storage.googleapis.com)... 108.177.121.128, 108.177.120.128, 142.250.128.128, ...\n",
            "Connecting to storage.googleapis.com (storage.googleapis.com)|108.177.121.128|:443... connected.\n",
            "HTTP request sent, awaiting response... 200 OK\n",
            "Length: 68970 (67K) [text/plain]\n",
            "Saving to: ‘/tmp/irish-lyrics-eof.txt’\n",
            "\n",
            "\r          /tmp/iris   0%[                    ]       0  --.-KB/s               \r/tmp/irish-lyrics-e 100%[===================>]  67.35K  --.-KB/s    in 0.001s  \n",
            "\n",
            "2022-03-14 04:54:36 (87.6 MB/s) - ‘/tmp/irish-lyrics-eof.txt’ saved [68970/68970]\n",
            "\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "#Getting Started with Tokenization"
      ],
      "metadata": {
        "id": "1eb4Mmahmj6h"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "TensorFlow Keras contains a library called preprocessing that provides a number of\n",
        "extremely useful tools to prepare data for machine learning. One of these is a\n",
        "Tokenizer that will allow you to take words and turn them into tokens\n",
        "\n",
        "In this case, we create a Tokenizer object and Once we have a tokenizer, calling fit_on_texts will create the tokenized word index.\n",
        "Printing this out will show a set of key/value pairs for the words in the corpus, like\n",
        "this:\n"
      ],
      "metadata": {
        "id": "P6SfCEE2pjCn"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "tokenizer = Tokenizer()\n",
        "\n",
        "data = open('/tmp/irish-lyrics-eof.txt').read()\n",
        "\n",
        "corpus = data.lower().split(\"\\n\")\n",
        "\n",
        "tokenizer.fit_on_texts(corpus)\n",
        "total_words = len(tokenizer.word_index) + 1\n",
        "\n",
        "print(tokenizer.word_index)\n",
        "print(total_words)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "6h9M4442jOsk",
        "outputId": "84e5d75a-67e1-481f-84e0-a530bc38f2c5"
      },
      "execution_count": 3,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "{'the': 1, 'and': 2, 'i': 3, 'to': 4, 'a': 5, 'of': 6, 'my': 7, 'in': 8, 'me': 9, 'for': 10, 'you': 11, 'all': 12, 'was': 13, 'she': 14, 'that': 15, 'on': 16, 'with': 17, 'her': 18, 'but': 19, 'as': 20, 'when': 21, 'love': 22, 'is': 23, 'your': 24, 'it': 25, 'will': 26, 'from': 27, 'by': 28, 'they': 29, 'be': 30, 'are': 31, 'so': 32, 'he': 33, 'old': 34, 'no': 35, 'oh': 36, 'ill': 37, 'at': 38, 'one': 39, 'his': 40, 'there': 41, 'were': 42, 'heart': 43, 'down': 44, 'now': 45, 'we': 46, 'where': 47, 'young': 48, 'never': 49, 'go': 50, 'come': 51, 'then': 52, 'did': 53, 'not': 54, 'said': 55, 'away': 56, 'their': 57, 'sweet': 58, 'them': 59, 'green': 60, 'if': 61, 'take': 62, 'our': 63, 'like': 64, 'night': 65, 'day': 66, 'o': 67, 'out': 68, 'fair': 69, 'this': 70, 'town': 71, 'have': 72, 'can': 73, 'true': 74, 'its': 75, 'thou': 76, 'see': 77, 'dear': 78, 'more': 79, 'theres': 80, 'or': 81, 'had': 82, 'would': 83, 'over': 84, 'hear': 85, 'up': 86, 'ive': 87, 'through': 88, 'home': 89, 'again': 90, 'well': 91, 'oer': 92, 'land': 93, 'good': 94, 'im': 95, 'ye': 96, 'sea': 97, 'left': 98, 'still': 99, 'father': 100, 'long': 101, 'rose': 102, 'could': 103, 'morning': 104, 'wild': 105, 'who': 106, 'eyes': 107, 'came': 108, 'while': 109, 'too': 110, 'back': 111, 'little': 112, 'an': 113, 'took': 114, 'him': 115, 'bow': 116, 'first': 117, 'let': 118, 'man': 119, 'shall': 120, 'know': 121, 'get': 122, 'high': 123, 'gone': 124, 'say': 125, 'ever': 126, 'some': 127, 'mary': 128, 'hand': 129, 'till': 130, 'put': 131, 'own': 132, 'time': 133, 'heard': 134, 'dead': 135, 'may': 136, 'bright': 137, 'mountain': 138, 'early': 139, 'rosin': 140, 'gave': 141, 'thee': 142, 'only': 143, 'far': 144, 'maid': 145, 'must': 146, 'find': 147, 'girl': 148, 'sure': 149, 'round': 150, 'dublin': 151, 'once': 152, 'world': 153, 'delight': 154, 'last': 155, 'johnny': 156, 'seen': 157, 'has': 158, 'fine': 159, 'road': 160, 'mother': 161, 'tis': 162, 'what': 163, 'way': 164, 'moon': 165, 'soul': 166, 'neer': 167, 'id': 168, 'just': 169, 'thats': 170, 'days': 171, 'darling': 172, 'went': 173, 'white': 174, 'die': 175, 'than': 176, 'hair': 177, 'goes': 178, 'meet': 179, 'today': 180, 'do': 181, 'girls': 182, 'shes': 183, 'thyme': 184, 'thy': 185, 'sing': 186, 'pretty': 187, 'new': 188, 'poor': 189, 'into': 190, 'life': 191, 'irish': 192, 'give': 193, 'boy': 194, 'youre': 195, 'make': 196, 'passed': 197, 'lovely': 198, 'black': 199, 'youll': 200, 'died': 201, 'red': 202, 'smile': 203, 'keep': 204, 'loves': 205, 'free': 206, 'leave': 207, 'friends': 208, 'each': 209, 'saw': 210, 'behind': 211, 'song': 212, 'ra': 213, 'dont': 214, 'arms': 215, 'am': 216, 'sun': 217, 'saying': 218, 'made': 219, 'wish': 220, 'cold': 221, 'met': 222, 'before': 223, 'should': 224, 'rocky': 225, 'light': 226, 'wid': 227, 'boys': 228, 'best': 229, 'fields': 230, 'since': 231, 'ball': 232, 'water': 233, 'casey': 234, 'mind': 235, 'along': 236, 'loved': 237, 'place': 238, 'ireland': 239, 'next': 240, 'three': 241, 'many': 242, 'years': 243, 'door': 244, 'us': 245, 'drink': 246, 'got': 247, 'might': 248, 'live': 249, 'roses': 250, 'play': 251, 'soon': 252, 'ground': 253, 'times': 254, 'spent': 255, 'going': 256, 'tree': 257, 'barley': 258, 'grass': 259, 'kind': 260, 'twas': 261, 'bridge': 262, 'around': 263, 'blue': 264, 'tell': 265, 'row': 266, 'how': 267, 'money': 268, 'merry': 269, 'stepped': 270, 'corporal': 271, 'always': 272, 'though': 273, 'near': 274, 'taken': 275, 'ones': 276, 'daughter': 277, 'forever': 278, 'loo': 279, 'shining': 280, 'plenty': 281, 'hes': 282, 'ship': 283, 'banks': 284, 'think': 285, 'very': 286, 'stand': 287, 'heres': 288, 'snow': 289, 'mountains': 290, 'molly': 291, 'wheel': 292, 'street': 293, 'erin': 294, 'side': 295, 'feet': 296, 'star': 297, 'look': 298, 'brave': 299, 'woman': 300, 'sons': 301, 'two': 302, 'says': 303, 'asked': 304, 'lanigans': 305, 'singing': 306, 'men': 307, 'toome': 308, 'stole': 309, 'god': 310, 'hill': 311, 'lonely': 312, 'lover': 313, 'tears': 314, 'fathers': 315, 'low': 316, 'voice': 317, 'quite': 318, 'able': 319, 'nice': 320, 'laid': 321, 'comrades': 322, 'wind': 323, 'another': 324, 'sit': 325, 'face': 326, 'band': 327, 'call': 328, 'colleen': 329, 'until': 330, 'hills': 331, 'mine': 332, 'above': 333, 'upon': 334, 'eer': 335, 'youve': 336, 'fly': 337, 'been': 338, 'late': 339, 'alive': 340, 'ballyjamesduff': 341, 'looked': 342, 'great': 343, 'why': 344, 'every': 345, 'proud': 346, 'found': 347, 'bragh': 348, 'such': 349, 'birds': 350, 'wedding': 351, 'welcome': 352, 'dancing': 353, 'da': 354, 'fell': 355, 'thinking': 356, 'roddy': 357, 'mccorley': 358, 'smiling': 359, 'mallow': 360, 'blooming': 361, 'thought': 362, 'peace': 363, 'soft': 364, 'pure': 365, 'harp': 366, 'dream': 367, 'alas': 368, 'yet': 369, 'clear': 370, 'art': 371, 'off': 372, 'hope': 373, 'fought': 374, 'mothers': 375, 'shore': 376, 'ago': 377, 'fol': 378, 'de': 379, 'house': 380, 'married': 381, 'bound': 382, 'danced': 383, 'devil': 384, 'dawning': 385, 'makes': 386, 'same': 387, 'sat': 388, 'any': 389, 'glass': 390, 'gay': 391, 'relations': 392, 'evening': 393, 'watched': 394, 'right': 395, 'fellows': 396, 'whiskey': 397, 'bonnie': 398, 'grows': 399, 'women': 400, 'flowers': 401, 'beauty': 402, 'cannot': 403, 'handsome': 404, 'happy': 405, 'gold': 406, 'rover': 407, 'none': 408, 'doneen': 409, 'summers': 410, 'people': 411, 'set': 412, 'paddy': 413, 'morn': 414, 'most': 415, 'easy': 416, 'struck': 417, 'beautiful': 418, 'those': 419, 'golden': 420, 'run': 421, 'pipes': 422, 'glen': 423, 'dying': 424, 'here': 425, 'wall': 426, 'across': 427, 'fire': 428, 'eileen': 429, 'longer': 430, 'cheeks': 431, 'valley': 432, 'both': 433, 'dew': 434, 'care': 435, 'bride': 436, 'nothing': 437, 'wont': 438, 'theyre': 439, 'colonel': 440, 'maiden': 441, 'shed': 442, 'til': 443, 'brown': 444, 'breast': 445, 'corn': 446, 'sinking': 447, 'began': 448, 'name': 449, 'cruel': 450, 'sound': 451, 'spancil': 452, 'county': 453, 'lies': 454, 'color': 455, 'thing': 456, 'decay': 457, 'sleep': 458, 'hours': 459, 'loving': 460, 'weary': 461, 'ringing': 462, 'please': 463, 'forget': 464, 'lie': 465, 'ran': 466, 'tore': 467, 'country': 468, 'fear': 469, 'fortune': 470, 'kissed': 471, 'alone': 472, 'ould': 473, 'cry': 474, 'dreams': 475, 'used': 476, 'horse': 477, 'break': 478, 'bells': 479, 'didnt': 480, 'weeks': 481, 'without': 482, 'raw': 483, 'nor': 484, 'twenty': 485, 'tune': 486, 'hed': 487, 'roving': 488, 'leaves': 489, 'cant': 490, 'death': 491, 'ten': 492, 'prison': 493, 'judge': 494, 'against': 495, 'lads': 496, 'shell': 497, 'fill': 498, 'valleys': 499, 'other': 500, 'pale': 501, 'joy': 502, 'wide': 503, 'bring': 504, 'ah': 505, 'cliffs': 506, 'city': 507, 'end': 508, 'turn': 509, 'sky': 510, 'born': 511, 'knew': 512, 'smiled': 513, 'rosie': 514, 'comes': 515, 'sayin': 516, 'lord': 517, 'dungannon': 518, 'blood': 519, 'air': 520, 'danny': 521, 'calling': 522, 'sunshine': 523, 'spring': 524, 'bid': 525, 'grow': 526, 'truth': 527, 'tear': 528, 'rings': 529, 'guns': 530, 'bay': 531, 'oflynn': 532, 'och': 533, 'stick': 534, 'rest': 535, 'four': 536, 'jewel': 537, 'tried': 538, 'grief': 539, 'answer': 540, 'kathleen': 541, 'fond': 542, 'eye': 543, 'goin': 544, 'pistols': 545, 'musha': 546, 'whack': 547, 'creole': 548, 'together': 549, 'room': 550, 'fall': 551, 'swore': 552, 'being': 553, 'step': 554, 'lark': 555, 'cailín': 556, 'deas': 557, 'crúite': 558, 'na': 559, 'mbó': 560, 'sir': 561, 'isle': 562, 'waiting': 563, 'magic': 564, 'skibbereen': 565, 'loud': 566, 'raise': 567, 'bent': 568, 'aged': 569, 'summer': 570, 'jenny': 571, 'excise': 572, 'rigadoo': 573, 'auld': 574, 'hearts': 575, 'nay': 576, 'stool': 577, 'farrell': 578, 'garden': 579, 'precious': 580, 'child': 581, 'slumber': 582, 'sleeping': 583, 'watch': 584, 'gently': 585, 'minstrel': 586, 'praise': 587, 'bell': 588, 'shaken': 589, 'immortal': 590, 'pray': 591, 'stay': 592, 'spoke': 593, 'cross': 594, 'brothers': 595, 'much': 596, 'past': 597, 'killarney': 598, 'sang': 599, 'tones': 600, 'ral': 601, 'wander': 602, 'cot': 603, 'feel': 604, 'yore': 605, 'answered': 606, 'divil': 607, 'middle': 608, 'bit': 609, 'led': 610, 'soldiers': 611, 'lily': 612, 'bed': 613, 'lassie': 614, 'clothes': 615, 'return': 616, 'broken': 617, 'derry': 618, 'sighed': 619, 'english': 620, 'tomorrow': 621, 'souls': 622, 'van': 623, 'diemans': 624, 'law': 625, 'neither': 626, 'winds': 627, 'rather': 628, 'doesnt': 629, 'rosy': 630, 'neatest': 631, 'hands': 632, 'whereon': 633, 'stands': 634, 'write': 635, 'thousand': 636, 'fare': 637, 'youd': 638, 'velvet': 639, 'neat': 640, 'landed': 641, 'health': 642, 'kellswater': 643, 'quiet': 644, 'stars': 645, 'beside': 646, 'warm': 647, 'sunday': 648, 'grey': 649, 'ocean': 650, 'sad': 651, 'spend': 652, 'kilkenny': 653, 'silver': 654, 'view': 655, 'west': 656, 'plain': 657, 'barrow': 658, 'broad': 659, 'narrow': 660, 'crying': 661, 'wonder': 662, 'save': 663, 'stop': 664, 'tender': 665, 'told': 666, 'lip': 667, 'dance': 668, 'foot': 669, 'kilrain': 670, 'saint': 671, 'visit': 672, 'mossy': 673, 'wexford': 674, 'irishmen': 675, 'shadow': 676, 'tho': 677, 'salley': 678, 'gardens': 679, 'foolish': 680, 'youth': 681, 'fade': 682, 'war': 683, 'believe': 684, 'which': 685, 'change': 686, 'entwine': 687, 'turns': 688, 'turned': 689, 'crown': 690, 'played': 691, 'captain': 692, 'blow': 693, 'children': 694, 'slainte': 695, 'gentle': 696, 'heavens': 697, 'bloom': 698, 'grand': 699, 'bush': 700, 'nest': 701, 'rich': 702, 'parting': 703, 'better': 704, 'window': 705, 'haste': 706, 'fresh': 707, 'stream': 708, 'rays': 709, 'ma': 710, 'ring': 711, 'lad': 712, 'athy': 713, 'drop': 714, 'hardly': 715, 'done': 716, 'arm': 717, 'leg': 718, 'beg': 719, 'drew': 720, 'bold': 721, 'drawn': 722, 'jail': 723, 'writin': 724, 'farewell': 725, 'tired': 726, 'lake': 727, 'want': 728, 'ringlets': 729, 'myself': 730, 'songs': 731, 'reel': 732, 'steps': 733, 'hearty': 734, 'fainted': 735, 'called': 736, 'under': 737, 'toe': 738, 'mairi': 739, 'fairest': 740, 'darlin': 741, 'bird': 742, 'memory': 743, 'lips': 744, 'sweetly': 745, 'morrow': 746, 'consent': 747, 'else': 748, 'sold': 749, 'stout': 750, 'pair': 751, 'drinking': 752, 'meself': 753, 'fray': 754, 'pike': 755, 'coat': 756, 'beneath': 757, 'rent': 758, 'part': 759, 'half': 760, 'head': 761, 'friend': 762, 'standing': 763, 'floor': 764, 'bare': 765, 'wed': 766, 'son': 767, 'pride': 768, 'vision': 769, 'sword': 770, 'after': 771, 'won': 772, 'farmers': 773, 'flower': 774, 'nut': 775, 'surely': 776, 'stood': 777, 'wandered': 778, 'athenry': 779, 'rising': 780, 'beating': 781, 'form': 782, 'dhu': 783, 'buy': 784, 'laughter': 785, 'wear': 786, 'raking': 787, 'rakes': 788, 'claret': 789, 'shure': 790, 'tralee': 791, 'slower': 792, 'lower': 793, 'deep': 794, 'wearin': 795, 'duram': 796, 'takes': 797, 'beware': 798, 'steal': 799, 'brings': 800, 'things': 801, 'joys': 802, 'bunch': 803, 'sailor': 804, 'chanced': 805, 'pass': 806, 'angels': 807, 'send': 808, 'drowsy': 809, 'keeping': 810, 'spirit': 811, 'stealing': 812, 'feeling': 813, 'roam': 814, 'presence': 815, 'heavenward': 816, 'dust': 817, 'dim': 818, 'journey': 819, 'waves': 820, 'frightened': 821, 'leaving': 822, 'struggle': 823, 'parents': 824, 'courage': 825, 'weeping': 826, 'pain': 827, 'mist': 828, 'felt': 829, 'roared': 830, 'making': 831, 'fever': 832, 'moment': 833, 'distance': 834, 'wailing': 835, 'oft': 836, 'held': 837, 'fast': 838, 'cabin': 839, 'honey': 840, 'diddle': 841, 'clearly': 842, 'open': 843, 'opened': 844, 'table': 845, 'wine': 846, 'lay': 847, 'shells': 848, 'sailed': 849, 'drown': 850, 'fetters': 851, 'chains': 852, 'wives': 853, 'sorrow': 854, 'thoughts': 855, 'cursed': 856, 'hell': 857, 'five': 858, 'buried': 859, 'lost': 860, 'endless': 861, 'slavery': 862, 'gun': 863, 'rain': 864, 'cares': 865, 'ghosts': 866, 'runaway': 867, 'twill': 868, 'month': 869, 'meadows': 870, 'prettiest': 871, 'winters': 872, 'satisfied': 873, 'few': 874, 'short': 875, 'lines': 876, 'shone': 877, 'shoulder': 878, 'belfast': 879, 'trade': 880, 'bad': 881, 'caused': 882, 'stray': 883, 'meaning': 884, 'damsel': 885, 'appear': 886, 'seven': 887, 'sentence': 888, 'jolly': 889, 'whenever': 890, 'wee': 891, 'wife': 892, 'lives': 893, 'martha': 894, 'courted': 895, 'bridgit': 896, 'omalley': 897, 'desolation': 898, 'thorn': 899, 'gaze': 900, 'stone': 901, 'approaching': 902, 'sets': 903, 'carrigfergus': 904, 'nights': 905, 'swim': 906, 'wings': 907, 'sober': 908, 'travel': 909, 'native': 910, 'places': 911, 'slopes': 912, 'hares': 913, 'lofty': 914, 'malone': 915, 'wheeled': 916, 'streets': 917, 'enough': 918, 'reilly': 919, 'tough': 920, 'whispers': 921, 'phil': 922, 'threw': 923, 'straight': 924, 'belles': 925, 'moor': 926, 'brand': 927, 'shapes': 928, 'work': 929, 'vow': 930, 'blarney': 931, 'paid': 932, 'bower': 933, 'remain': 934, 'charming': 935, 'storied': 936, 'chieftains': 937, 'slaughter': 938, 'bann': 939, 'boyne': 940, 'liffey': 941, 'gallant': 942, 'awake': 943, 'greet': 944, 'meadow': 945, 'sweeter': 946, 'dirty': 947, 'cats': 948, 'crossed': 949, 'field': 950, 'river': 951, 'full': 952, 'aroon': 953, 'sends': 954, 'woe': 955, 'chain': 956, 'main': 957, 'charms': 958, 'fondly': 959, 'fleet': 960, 'fairy': 961, 'thine': 962, 'known': 963, 'truly': 964, 'close': 965, 'story': 966, 'flag': 967, 'sweetest': 968, 'honor': 969, 'playing': 970, 'mauser': 971, 'music': 972, 'tom': 973, 'hurrah': 974, 'big': 975, 'lead': 976, 'south': 977, 'generation': 978, 'freedom': 979, 'agin': 980, 'creature': 981, 'dad': 982, 'venture': 983, 'word': 984, 'wonderful': 985, 'crazy': 986, 'lazy': 987, 'grave': 988, 'jest': 989, 'remark': 990, 'strangers': 991, 'strong': 992, 'shook': 993, 'walk': 994, 'north': 995, 'ours': 996, 'cease': 997, 'strife': 998, 'whats': 999, 'lilacs': 1000, 'prove': 1001, 'sweetheart': 1002, 'letters': 1003, 'sent': 1004, 'speak': 1005, 'brow': 1006, 'albert': 1007, 'mooney': 1008, 'fighting': 1009, 'fingers': 1010, 'toes': 1011, 'john': 1012, 'hurroo': 1013, 'drums': 1014, 'beguiled': 1015, 'carry': 1016, 'bone': 1017, 'havent': 1018, 'walkin': 1019, 'kilgary': 1020, 'pepper': 1021, 'countin': 1022, 'forth': 1023, 'deliver': 1024, 'daddy': 1025, 'em': 1026, 'deceive': 1027, 'between': 1028, 'even': 1029, 'prisoner': 1030, 'fists': 1031, 'knocked': 1032, 'carriages': 1033, 'rollin': 1034, 'juice': 1035, 'courtin': 1036, 'ponchartrain': 1037, 'does': 1038, 'stranger': 1039, 'marry': 1040, 'adieu': 1041, 'ask': 1042, 'tipped': 1043, 'arrived': 1044, 'ladies': 1045, 'potatoes': 1046, 'courting': 1047, 'miss': 1048, 'small': 1049, 'ned': 1050, 'ribbons': 1051, 'heel': 1052, 'bonny': 1053, 'pipe': 1054, 'thrush': 1055, 'sweethearts': 1056, 'unto': 1057, 'rise': 1058, 'softly': 1059, 'milking': 1060, 'rare': 1061, 'pity': 1062, 'treasure': 1063, 'noon': 1064, 'sailing': 1065, 'banish': 1066, 'riches': 1067, 'comfort': 1068, 'yonder': 1069, 'flows': 1070, 'fairer': 1071, 'lass': 1072, 'woods': 1073, 'strayed': 1074, 'locks': 1075, 'breaking': 1076, 'june': 1077, 'started': 1078, 'hearted': 1079, 'beer': 1080, 'daylight': 1081, 'among': 1082, 'bundle': 1083, 'connaught': 1084, 'quay': 1085, 'erins': 1086, 'galway': 1087, 'fearless': 1088, 'bravely': 1089, 'marches': 1090, 'fate': 1091, 'neck': 1092, 'trod': 1093, 'marched': 1094, 'antrim': 1095, 'sash': 1096, 'flashed': 1097, 'hath': 1098, 'foemans': 1099, 'fight': 1100, 'heavy': 1101, 'bore': 1102, 'mans': 1103, 'counter': 1104, 'dozen': 1105, 'gallon': 1106, 'bottles': 1107, 'diamond': 1108, 'resemble': 1109, 'tiny': 1110, 'friendly': 1111, 'weather': 1112, 'inside': 1113, 'remember': 1114, 'someone': 1115, 'hat': 1116, 'body': 1117, 'dancers': 1118, 'hanging': 1119, 'empty': 1120, 'shoes': 1121, 'broke': 1122, 'december': 1123, 'move': 1124, 'reason': 1125, 'roof': 1126, 'naught': 1127, 'tower': 1128, 'power': 1129, 'king': 1130, 'dreaming': 1131, 'crew': 1132, 'whos': 1133, 'mccann': 1134, 'smoke': 1135, 'notes': 1136, 'yeoman': 1137, 'cavalry': 1138, 'guard': 1139, 'forced': 1140, 'brother': 1141, 'cousin': 1142, 'blame': 1143, 'croppy': 1144, 'dressed': 1145, 'trees': 1146, 'wore': 1147, 'words': 1148, 'swiftly': 1149, 'dawn': 1150, 'lovd': 1151, 'voices': 1152, 'moaning': 1153, 'dark': 1154, 'gather': 1155, 'tay': 1156, 'swinging': 1157, 'drinkin': 1158, 'sitting': 1159, 'stile': 1160, 'springing': 1161, 'yours': 1162, 'kept': 1163, 'aisey': 1164, 'rub': 1165, 'dub': 1166, 'dow': 1167, 'shelah': 1168, 'fairly': 1169, 'beggarman': 1170, 'begging': 1171, 'slept': 1172, 'holes': 1173, 'coming': 1174, 'thru': 1175, 'boo': 1176, 'lady': 1177, 'kerry': 1178, 'pipers': 1179, 'laugh': 1180, 'beaming': 1181, 'guineas': 1182, 'least': 1183, 'diggin': 1184, 'mourne': 1185, 'spending': 1186, 'mellow': 1187, 'plying': 1188, 'slowly': 1189, 'mooncoin': 1190, 'flow': 1191, 'sounds': 1192, 'shine': 1193, 'cool': 1194, 'crystal': 1195, 'fountain': 1196, 'moonlight': 1197, 'grandmother': 1198, 'crooning': 1199, 'merrily': 1200, 'spins': 1201, 'lightly': 1202, 'moving': 1203, 'lattice': 1204, 'grove': 1205, 'swings': 1206, 'finger': 1207, 'shamrock': 1208, 'pocket': 1209, 'springtime': 1210, 'gilgarra': 1211, 'rapier': 1212, 'ringum': 1213, 'mornin': 1214, 'heather': 1215, 'build': 1216, 'maidens': 1217, 'prime': 1218, 'nlyme': 1219, 'flavours': 1220, 'lusty': 1221, 'reminded': 1222, 'attend': 1223, 'guardian': 1224, 'creeping': 1225, 'dale': 1226, 'vigil': 1227, 'visions': 1228, 'revealing': 1229, 'breathes': 1230, 'holy': 1231, 'strains': 1232, 'hover': 1233, 'hark': 1234, 'solemn': 1235, 'winging': 1236, 'earthly': 1237, 'shalt': 1238, 'awaken': 1239, 'destiny': 1240, 'emigrants': 1241, 'amid': 1242, 'longing': 1243, 'parted': 1244, 'townland': 1245, 'vessel': 1246, 'crowded': 1247, 'disquieted': 1248, 'folk': 1249, 'escape': 1250, 'hardship': 1251, 'sustaining': 1252, 'glimpse': 1253, 'faded': 1254, 'strangely': 1255, 'seas': 1256, 'anger': 1257, 'desperate': 1258, 'plight': 1259, 'worsened': 1260, 'delirium': 1261, 'possessed': 1262, 'clouded': 1263, 'prayers': 1264, 'begged': 1265, 'forgiveness': 1266, 'seeking': 1267, 'distant': 1268, 'mither': 1269, 'simple': 1270, 'ditty': 1271, 'ld': 1272, 'li': 1273, 'hush': 1274, 'lullaby': 1275, 'huggin': 1276, 'hummin': 1277, 'rock': 1278, 'asleep': 1279, 'outside': 1280, 'modestly': 1281, 'ry': 1282, 'ay': 1283, 'di': 1284, 're': 1285, 'dai': 1286, 'rie': 1287, 'shc': 1288, 'bridle': 1289, 'stable': 1290, 'oats': 1291, 'eat': 1292, 'soldier': 1293, 'aisy': 1294, 'arose': 1295, 'christmas': 1296, '1803': 1297, 'australia': 1298, 'marks': 1299, 'carried': 1300, 'rusty': 1301, 'iron': 1302, 'wains': 1303, 'mainsails': 1304, 'unfurled': 1305, 'curses': 1306, 'hurled': 1307, 'swell': 1308, 'moth': 1309, 'firelights': 1310, 'horses': 1311, 'rode': 1312, 'taking': 1313, 'hades': 1314, 'twilight': 1315, 'forty': 1316, 'slime': 1317, 'climate': 1318, 'bravery': 1319, 'ended': 1320, 'bond': 1321, 'rebel': 1322, 'iii': 1323, 'violin': 1324, 'clay': 1325, 'sooner': 1326, 'sport': 1327, 'colour': 1328, 'knows': 1329, 'earth': 1330, 'serve': 1331, 'clyde': 1332, 'mourn': 1333, 'weep': 1334, 'suffer': 1335, 'diamonds': 1336, 'queen': 1337, 'hung': 1338, 'tied': 1339, 'apprenticed': 1340, 'happiness': 1341, 'misfortune': 1342, 'follow': 1343, 'strolling': 1344, 'selling': 1345, 'bar': 1346, 'customer': 1347, 'slipped': 1348, 'luck': 1349, 'jury': 1350, 'trial': 1351, 'case': 1352, 'warning': 1353, 'liquor': 1354, 'porter': 1355, 'pleasures': 1356, 'fishing': 1357, 'farming': 1358, 'glens': 1359, 'softest': 1360, 'dripping': 1361, 'snare': 1362, 'lose': 1363, 'court': 1364, 'primrose': 1365, 'bee': 1366, 'hopeless': 1367, 'wonders': 1368, 'admiration': 1369, 'haunt': 1370, 'wherever': 1371, 'sands': 1372, 'purer': 1373, 'within': 1374, 'grieve': 1375, 'drumslieve': 1376, 'ballygrant': 1377, 'deepest': 1378, 'boatsman': 1379, 'ferry': 1380, 'childhood': 1381, 'reflections': 1382, 'boyhood': 1383, 'melting': 1384, 'roaming': 1385, 'reported': 1386, 'marble': 1387, 'stones': 1388, 'ink': 1389, 'support': 1390, 'drunk': 1391, 'seldom': 1392, 'sick': 1393, 'numbered': 1394, 'foam': 1395, 'compare': 1396, 'sights': 1397, 'coast': 1398, 'clare': 1399, 'kilkee': 1400, 'kilrush': 1401, 'watching': 1402, 'pheasants': 1403, 'homes': 1404, 'streams': 1405, 'dublins': 1406, 'cockles': 1407, 'mussels': 1408, 'fish': 1409, 'monger': 1410, 'ghost': 1411, 'wheels': 1412, 'eden': 1413, 'vanished': 1414, 'finea': 1415, 'halfway': 1416, 'cootehill': 1417, 'gruff': 1418, 'whispering': 1419, 'crow': 1420, 'newborn': 1421, 'babies': 1422, 'huff': 1423, 'start': 1424, 'sorrowful': 1425, 'squall': 1426, 'babys': 1427, 'toil': 1428, 'worn': 1429, 'fore': 1430, 'flute': 1431, 'yer': 1432, 'boot': 1433, 'magee': 1434, 'scruff': 1435, 'slanderin': 1436, 'marchin': 1437, 'assisted': 1438, 'drain': 1439, 'dudeen': 1440, 'puff': 1441, 'whisperings': 1442, 'barrin': 1443, 'chocolate': 1444, 'feegee': 1445, 'sort': 1446, 'moonshiny': 1447, 'stuff': 1448, 'addle': 1449, 'brain': 1450, 'ringin': 1451, 'glamour': 1452, 'gas': 1453, 'guff': 1454, 'whisper': 1455, 'oil': 1456, 'remarkable': 1457, 'policeman': 1458, 'bluff': 1459, 'maintain': 1460, 'guril': 1461, 'sic': 1462, 'passage': 1463, 'rough': 1464, 'borne': 1465, 'breeze': 1466, 'boundless': 1467, 'stupendous': 1468, 'roll': 1469, 'thundering': 1470, 'motion': 1471, 'mermaids': 1472, 'fierce': 1473, 'tempest': 1474, 'gathers': 1475, 'oneill': 1476, 'odonnell': 1477, 'lucan': 1478, 'oconnell': 1479, 'brian': 1480, 'drove': 1481, 'danes': 1482, 'patrick': 1483, 'vermin': 1484, 'whose': 1485, 'benburb': 1486, 'blackwater': 1487, 'owen': 1488, 'roe': 1489, 'munroe': 1490, 'lambs': 1491, 'skip': 1492, 'views': 1493, 'enchanting': 1494, 'rostrevor': 1495, 'groves': 1496, 'lakes': 1497, 'ride': 1498, 'tide': 1499, 'majestic': 1500, 'shannon': 1501, 'sail': 1502, 'loch': 1503, 'neagh': 1504, 'ross': 1505, 'gorey': 1506, 'saxon': 1507, 'tory': 1508, 'soil': 1509, 'sanctified': 1510, 'enemies': 1511, 'links': 1512, 'encumbered': 1513, 'resound': 1514, 'hosannahs': 1515, 'bide': 1516, 'hushed': 1517, 'lying': 1518, 'kneel': 1519, 'ave': 1520, 'tread': 1521, 'fail': 1522, 'simply': 1523, 'gasworks': 1524, 'croft': 1525, 'dreamed': 1526, 'canal': 1527, 'factory': 1528, 'clouds': 1529, 'drifting': 1530, 'prowling': 1531, 'beat': 1532, 'springs': 1533, 'siren': 1534, 'docks': 1535, 'train': 1536, 'smelled': 1537, 'smokey': 1538, 'sharp': 1539, 'axe': 1540, 'steel': 1541, 'tempered': 1542, 'chop': 1543, 't': 1544, 'agree': 1545, 'leaning': 1546, 'weirs': 1547, 'ray': 1548, 'glow': 1549, 'changeless': 1550, 'constant': 1551, 'bounding': 1552, 'castles': 1553, 'sacked': 1554, 'scattered': 1555, 'fixed': 1556, 'endearing': 1557, 'gifts': 1558, 'fading': 1559, 'wouldst': 1560, 'adored': 1561, 'loveliness': 1562, 'ruin': 1563, 'itself': 1564, 'verdantly': 1565, 'unprofaned': 1566, 'fervor': 1567, 'faith': 1568, 'forgets': 1569, 'sunflower': 1570, 'rag': 1571, 'games': 1572, 'hold': 1573, 'defend': 1574, 'veteran': 1575, 'volunteers': 1576, 'pat': 1577, 'pearse': 1578, 'clark': 1579, 'macdonagh': 1580, 'macdiarmada': 1581, 'mcbryde': 1582, 'james': 1583, 'connolly': 1584, 'placed': 1585, 'machine': 1586, 'ranting': 1587, 'hour': 1588, 'bullet': 1589, 'stuck': 1590, 'craw': 1591, 'poisoning': 1592, 'ceannt': 1593, 'lions': 1594, 'union': 1595, 'poured': 1596, 'dismay': 1597, 'horror': 1598, 'englishmen': 1599, 'khaki': 1600, 'renown': 1601, 'fame': 1602, 'forefathers': 1603, 'blaze': 1604, 'priests': 1605, 'offer': 1606, 'charmin': 1607, 'variety': 1608, 'renownd': 1609, 'learnin': 1610, 'piety': 1611, 'advance': 1612, 'widout': 1613, 'impropriety': 1614, 'flowr': 1615, 'cho': 1616, 'powrfulest': 1617, 'preacher': 1618, 'tenderest': 1619, 'teacher': 1620, 'kindliest': 1621, 'donegal': 1622, 'talk': 1623, 'provost': 1624, 'trinity': 1625, 'famous': 1626, 'greek': 1627, 'latinity': 1628, 'divils': 1629, 'divinity': 1630, 'd': 1631, 'likes': 1632, 'logic': 1633, 'mythology': 1634, 'thayology': 1635, 'conchology': 1636, 'sinners': 1637, 'wishful': 1638, 'childer': 1639, 'avick': 1640, 'gad': 1641, 'flock': 1642, 'grandest': 1643, 'control': 1644, 'checking': 1645, 'coaxin': 1646, 'onaisy': 1647, 'lifting': 1648, 'avoidin': 1649, 'frivolity': 1650, 'seasons': 1651, 'innocent': 1652, 'jollity': 1653, 'playboy': 1654, 'claim': 1655, 'equality': 1656, 'comicality': 1657, 'bishop': 1658, 'lave': 1659, 'gaiety': 1660, 'laity': 1661, 'clergy': 1662, 'jewels': 1663, 'plundering': 1664, 'pillage': 1665, 'starved': 1666, 'cries': 1667, 'thems': 1668, 'bondage': 1669, 'fourth': 1670, 'tabhair': 1671, 'dom': 1672, 'lámh': 1673, 'harmony': 1674, 'east': 1675, 'destroy': 1676, 'command': 1677, 'gesture': 1678, 'troubles': 1679, 'weak': 1680, 'peoples': 1681, 'creeds': 1682, 'lets': 1683, 'needs': 1684, 'passion': 1685, 'fashion': 1686, 'guide': 1687, 'share': 1688, 'sparkling': 1689, 'meeting': 1690, 'iull': 1691, 'contented': 1692, 'ache': 1693, 'painful': 1694, 'wrote': 1695, 'twisted': 1696, 'twined': 1697, 'cheek': 1698, 'bedim': 1699, 'holds': 1700, 'smiles': 1701, 'scarcely': 1702, 'darkning': 1703, 'beyond': 1704, 'yearn': 1705, 'laughs': 1706, 'humble': 1707, 'brightest': 1708, 'gleam': 1709, 'forgot': 1710, 'pulled': 1711, 'comb': 1712, 'counting': 1713, 'knock': 1714, 'murray': 1715, 'fellow': 1716, 'hail': 1717, 'tumblin': 1718, 'apple': 1719, 'pie': 1720, 'gets': 1721, 'doleful': 1722, 'enemy': 1723, 'nearly': 1724, 'slew': 1725, 'queer': 1726, 'mild': 1727, 'legs': 1728, 'indeed': 1729, 'island': 1730, 'sulloon': 1731, 'flesh': 1732, 'yere': 1733, 'armless': 1734, 'boneless': 1735, 'chickenless': 1736, 'egg': 1737, 'yell': 1738, 'bowl': 1739, 'rolling': 1740, 'swearing': 1741, 'rattled': 1742, 'saber': 1743, 'deceiver': 1744, 'rig': 1745, 'um': 1746, 'du': 1747, 'rum': 1748, 'jar': 1749, 'shinin': 1750, 'coins': 1751, 'promised': 1752, 'vowed': 1753, 'devils': 1754, 'awakened': 1755, 'six': 1756, 'guards': 1757, 'numbers': 1758, 'odd': 1759, 'flew': 1760, 'mistaken': 1761, 'mollys': 1762, 'robbing': 1763, 'sentry': 1764, 'sligo': 1765, 'fishin': 1766, 'bowlin': 1767, 'others': 1768, 'railroad': 1769, 'ties': 1770, 'crossings': 1771, 'swamps': 1772, 'elevations': 1773, 'resolved': 1774, 'sunset': 1775, 'higher': 1776, 'win': 1777, 'allegators': 1778, 'wood': 1779, 'treated': 1780, 'shoulders': 1781, 'paint': 1782, 'picture': 1783, 'vain': 1784, 'returned': 1785, 'cottage': 1786, 'sociable': 1787, 'foaming': 1788, 'n': 1789, 'jeremy': 1790, 'lanigan': 1791, 'battered': 1792, 'hadnt': 1793, 'pound': 1794, 'farm': 1795, 'acres': 1796, 'party': 1797, 'listen': 1798, 'glisten': 1799, 'rows': 1800, 'ructions': 1801, 'invitation': 1802, 'minute': 1803, 'bees': 1804, 'cask': 1805, 'judy': 1806, 'odaly': 1807, 'milliner': 1808, 'wink': 1809, 'peggy': 1810, 'mcgilligan': 1811, 'lashings': 1812, 'punch': 1813, 'cakes': 1814, 'bacon': 1815, 'tea': 1816, 'nolans': 1817, 'dolans': 1818, 'ogradys': 1819, 'sounded': 1820, 'taras': 1821, 'hall': 1822, 'nelly': 1823, 'gray': 1824, 'rat': 1825, 'catchers': 1826, 'doing': 1827, 'kinds': 1828, 'nonsensical': 1829, 'polkas': 1830, 'whirligig': 1831, 'julia': 1832, 'banished': 1833, 'nonsense': 1834, 'twist': 1835, 'jig': 1836, 'mavrone': 1837, 'mad': 1838, 'ceiling': 1839, 'brooks': 1840, 'academy': 1841, 'learning': 1842, 'learn': 1843, 'couples': 1844, 'groups': 1845, 'accident': 1846, 'happened': 1847, 'terrance': 1848, 'mccarthy': 1849, 'finnertys': 1850, 'hoops': 1851, 'cried': 1852, 'meelia': 1853, 'murther': 1854, 'gathered': 1855, 'carmody': 1856, 'further': 1857, 'satisfaction': 1858, 'midst': 1859, 'kerrigan': 1860, 'declared': 1861, 'painted': 1862, 'suppose': 1863, 'morgan': 1864, 'powerful': 1865, 'stretched': 1866, 'smashed': 1867, 'chaneys': 1868, 'runctions': 1869, 'lick': 1870, 'phelim': 1871, 'mchugh': 1872, 'replied': 1873, 'introduction': 1874, 'kicked': 1875, 'terrible': 1876, 'hullabaloo': 1877, 'piper': 1878, 'strangled': 1879, 'squeezed': 1880, 'bellows': 1881, 'chanters': 1882, 'entangled': 1883, 'gaily': 1884, 'mairis': 1885, 'hillways': 1886, 'myrtle': 1887, 'bracken': 1888, 'sheilings': 1889, 'sake': 1890, 'rowans': 1891, 'herring': 1892, 'meal': 1893, 'peat': 1894, 'creel': 1895, 'bairns': 1896, 'weel': 1897, 'toast': 1898, 'soar': 1899, 'blackbird': 1900, 'note': 1901, 'linnet': 1902, 'lure': 1903, 'cozy': 1904, 'catch': 1905, 'company': 1906, 'harm': 1907, 'wit': 1908, 'recall': 1909, 'leisure': 1910, 'awhile': 1911, 'sorely': 1912, 'ruby': 1913, 'enthralled': 1914, 'sorry': 1915, 'theyd': 1916, 'falls': 1917, 'lot': 1918, 'tuned': 1919, 'bough': 1920, 'cow': 1921, 'chanting': 1922, 'melodious': 1923, 'scarce': 1924, 'soothed': 1925, 'solace': 1926, 'courtesy': 1927, 'salute': 1928, 'amiable': 1929, 'captive': 1930, 'slave': 1931, 'future': 1932, 'banter': 1933, 'enamour': 1934, 'indies': 1935, 'afford': 1936, 'transparently': 1937, 'flame': 1938, 'add': 1939, 'fuel': 1940, 'grant': 1941, 'desire': 1942, 'expire': 1943, 'wealth': 1944, 'damer': 1945, 'african': 1946, 'devonshire': 1947, 'lamp': 1948, 'alladin': 1949, 'genie': 1950, 'also': 1951, 'withdraw': 1952, 'tease': 1953, 'single': 1954, 'airy': 1955, 'embarrass': 1956, 'besides': 1957, 'almanack': 1958, 'useless': 1959, 'date': 1960, 'ware': 1961, 'rate': 1962, 'fragrance': 1963, 'loses': 1964, 'consumed': 1965, 'october': 1966, 'knowing': 1967, 'steer': 1968, 'blast': 1969, 'danger': 1970, 'farthing': 1971, 'affection': 1972, 'enjoy': 1973, 'choose': 1974, 'killarneys': 1975, 'sister': 1976, 'pains': 1977, 'loss': 1978, 'tuam': 1979, 'saluted': 1980, 'drank': 1981, 'pint': 1982, 'smother': 1983, 'reap': 1984, 'cut': 1985, 'goblins': 1986, 'bought': 1987, 'brogues': 1988, 'rattling': 1989, 'bogs': 1990, 'frightning': 1991, 'dogs': 1992, 'hunt': 1993, 'hare': 1994, 'follol': 1995, 'rah': 1996, 'mullingar': 1997, 'rested': 1998, 'limbs': 1999, 'blithe': 2000, 'heartfrom': 2001, 'paddys': 2002, 'cure': 2003, 'lassies': 2004, 'laughing': 2005, 'curious': 2006, 'style': 2007, 'twould': 2008, 'bubblin': 2009, 'hired': 2010, 'wages': 2011, 'required': 2012, 'almost': 2013, 'deprived': 2014, 'stroll': 2015, 'quality': 2016, 'locality': 2017, 'something': 2018, 'wobblin': 2019, 'enquiring': 2020, 'rogue': 2021, 'brogue': 2022, 'wasnt': 2023, 'vogue': 2024, 'spirits': 2025, 'falling': 2026, 'jumped': 2027, 'aboard': 2028, 'pigs': 2029, 'rigs': 2030, 'jigs': 2031, 'bubbling': 2032, 'holyhead': 2033, 'wished': 2034, 'instead': 2035, 'bouys': 2036, 'liverpool': 2037, 'safely': 2038, 'fool': 2039, 'boil': 2040, 'temper': 2041, 'losing': 2042, 'abusing': 2043, 'shillelagh': 2044, 'nigh': 2045, 'hobble': 2046, 'load': 2047, 'hurray': 2048, 'joined': 2049, 'affray': 2050, 'quitely': 2051, 'cleared': 2052, 'host': 2053, 'march': 2054, 'faces': 2055, 'farmstead': 2056, 'fishers': 2057, 'ban': 2058, 'vengeance': 2059, 'hapless': 2060, 'about': 2061, 'hemp': 2062, 'rope': 2063, 'clung': 2064, 'grim': 2065, 'array': 2066, 'earnest': 2067, 'stalwart': 2068, 'stainless': 2069, 'banner': 2070, 'marching': 2071, 'torn': 2072, 'furious': 2073, 'odds': 2074, 'keen': 2075, 'toomebridge': 2076, 'treads': 2077, 'upwards': 2078, 'traveled': 2079, 'quarters': 2080, 'below': 2081, 'hogshead': 2082, 'stack': 2083, 'stagger': 2084, 'dig': 2085, 'hole': 2086, 'couple': 2087, 'scratch': 2088, 'consolation': 2089, 'tyrant': 2090, 'remorseless': 2091, 'foe': 2092, 'lift': 2093, 'stranded': 2094, 'prince': 2095, 'edward': 2096, 'coffee': 2097, 'trace': 2098, 'fiddlin': 2099, 'dime': 2100, 'shy': 2101, 'hello': 2102, 'wintry': 2103, 'yellow': 2104, 'somewhere': 2105, 'written': 2106, 'begin': 2107, 'tap': 2108, 'caught': 2109, 'leap': 2110, 'clumsy': 2111, 'graceful': 2112, 'fiddlers': 2113, 'everywhere': 2114, 'boots': 2115, 'laughtcr': 2116, 'suits': 2117, 'easter': 2118, 'gowns': 2119, 'sailors': 2120, 'pianos': 2121, 'setting': 2122, 'someones': 2123, 'hats': 2124, 'rack': 2125, 'chair': 2126, 'wooden': 2127, 'feels': 2128, 'touch': 2129, 'awaitin': 2130, 'thc': 2131, 'fiddles': 2132, 'closet': 2133, 'strings': 2134, 'tbe': 2135, 'covers': 2136, 'buttoned': 2137, 'sometimes': 2138, 'melody': 2139, 'passes': 2140, 'slight': 2141, 'lack': 2142, 'moved': 2143, 'homeward': 2144, 'swan': 2145, 'moves': 2146, 'goods': 2147, 'gear': 2148, 'din': 2149, 'rude': 2150, 'wherein': 2151, 'dwell': 2152, 'abandon': 2153, 'energy': 2154, 'blight': 2155, 'praties': 2156, 'sheep': 2157, 'cattle': 2158, 'taxes': 2159, 'unpaid': 2160, 'redeem': 2161, 'bleak': 2162, 'landlord': 2163, 'sheriff': 2164, 'spleen': 2165, 'heaved': 2166, 'sigh': 2167, 'bade': 2168, 'goodbye': 2169, 'stony': 2170, 'anguish': 2171, 'seeing': 2172, 'feeble': 2173, 'frame': 2174, 'wrapped': 2175, 'c�ta': 2176, 'm�r': 2177, 'unseen': 2178, 'stern': 2179, 'rally': 2180, 'cheer': 2181, 'revenge': 2182, 'waking': 2183, 'wisdom': 2184, 'dwelling': 2185, 'battleshield': 2186, 'dignity': 2187, 'shelter': 2188, 'heed': 2189, 'inheritance': 2190, 'heavem': 2191, 'heaven': 2192, 'victory': 2193, 'reach': 2194, 'whatever': 2195, 'befall': 2196, 'ruler': 2197, 'pleasant': 2198, 'rambling': 2199, 'board': 2200, 'followed': 2201, 'shortly': 2202, 'anchor': 2203, '23rd': 2204, 'lrelands': 2205, 'daughters': 2206, 'crowds': 2207, 'assembled': 2208, 'fulfill': 2209, 'jovial': 2210, 'conversations': 2211, 'neighbors': 2212, 'turning': 2213, 'tailor': 2214, 'quigley': 2215, 'bould': 2216, 'britches': 2217, 'lived': 2218, 'flying': 2219, 'dove': 2220, 'hiii': 2221, 'dreamt': 2222, 'joking': 2223, 'manys': 2224, 'cock': 2225, 'shrill': 2226, 'awoke': 2227, 'california': 2228, 'miles': 2229, 'banbridge': 2230, 'july': 2231, 'boreen': 2232, 'sheen': 2233, 'coaxing': 2234, 'elf': 2235, 'shake': 2236, 'bantry': 2237, 'onward': 2238, 'sped': 2239, 'gazed': 2240, 'passerby': 2241, 'gem': 2242, 'irelands': 2243, 'travelled': 2244, 'hit': 2245, 'career': 2246, 'square': 2247, 'surrendered': 2248, 'tenant': 2249, 'shawl': 2250, 'gown': 2251, 'crossroads': 2252, 'dress': 2253, 'try': 2254, 'sheeps': 2255, 'deludhering': 2256, 'yoke': 2257, 'rust': 2258, 'plow': 2259, 'fireside': 2260, 'sits': 2261, 'whistle': 2262, 'changing': 2263, 'fright': 2264, 'downfall': 2265, 'cornwall': 2266, 'parlour': 2267, 'passing': 2268, 'william': 2269, 'betray': 2270, 'guinea': 2271, 'walking': 2272, 'mounted': 2273, 'platform': 2274, 'deny': 2275, 'walked': 2276, 'margin': 2277, 'lough': 2278, 'leane': 2279, 'bloomed': 2280, 'whom': 2281, 'cap': 2282, 'cloak': 2283, 'glossy': 2284, 'pail': 2285, 'palm': 2286, 'venus': 2287, 'bank': 2288, 'travelians': 2289, 'babes': 2290, 'freebirds': 2291, 'grew': 2292, 'matters': 2293, 'famine': 2294, 'rebelled': 2295, 'windswept': 2296, 'harbour': 2297, 'botany': 2298, 'whilst': 2299, 'wan': 2300, 'cloud': 2301, 'shannons': 2302, 'returnd': 2303, 'doubts': 2304, 'fears': 2305, 'aching': 2306, 'seemd': 2307, 'mingling': 2308, 'flood': 2309, 'path': 2310, 'wrath': 2311, 'lamenting': 2312, 'sudden': 2313, 'kissd': 2314, 'showrs': 2315, 'flowing': 2316, 'laughd': 2317, 'beam': 2318, 'soared': 2319, 'aloft': 2320, 'phantom': 2321, 'outspread': 2322, 'throbbing': 2323, 'hid': 2324, 'treasures': 2325, 'pots': 2326, 'tin': 2327, 'cans': 2328, 'mash': 2329, 'bran': 2330, 'barney': 2331, 'peeled': 2332, 'searching': 2333, 'connemara': 2334, 'butcher': 2335, 'quart': 2336, 'bottle': 2337, 'help': 2338, 'gate': 2339, 'glory': 2340, 'lane': 2341, 'village': 2342, 'church': 2343, 'spire': 2344, 'graveyard': 2345, 'baby': 2346, 'blessing': 2347, 'hoping': 2348, 'trust': 2349, 'strength': 2350, 'thank': 2351, 'bidding': 2352, 'bread': 2353, 'shines': 2354, 'fifty': 2355, 'often': 2356, 'shut': 2357, 'frisky': 2358, 'pig': 2359, 'whisky': 2360, 'uncle': 2361, 'enlisted': 2362, 'trudged': 2363, 'bosom': 2364, 'daisy': 2365, 'drubbing': 2366, 'shirts': 2367, 'battle': 2368, 'blows': 2369, 'pate': 2370, 'bothered': 2371, 'rarely': 2372, 'dropped': 2373, 'honest': 2374, 'thinks': 2375, 'eight': 2376, 'score': 2377, 'basin': 2378, 'zoo': 2379, 'everybody': 2380, 'calls': 2381, 'trades': 2382, 'dinner': 2383, 'slip': 2384, 'corner': 2385, 'barn': 2386, 'currabawn': 2387, 'shocking': 2388, 'wet': 2389, 'raindrops': 2390, 'rats': 2391, 'peek': 2392, 'waken': 2393, 'spotted': 2394, 'apron': 2395, 'calico': 2396, 'blouse': 2397, 'frighten': 2398, 'afraid': 2399, 'flaxen': 2400, 'haired': 2401, 'rags': 2402, 'tags': 2403, 'leggins': 2404, 'collar': 2405, 'tie': 2406, 'goggles': 2407, 'fashioned': 2408, 'bag': 2409, 'bulging': 2410, 'sack': 2411, 'peeping': 2412, 'skin': 2413, 'rink': 2414, 'doodle': 2415, 'getting': 2416, 'raked': 2417, 'gladness': 2418, 'tuning': 2419, 'fills': 2420, 'eily': 2421, 'prouder': 2422, 'thady': 2423, 'boldly': 2424, 'lasses': 2425, 'fled': 2426, 'silent': 2427, 'glad': 2428, 'echo': 2429, 'companions': 2430, 'soars': 2431, 'enchanted': 2432, 'granted': 2433, 'adoration': 2434, 'gives': 2435, 'joyous': 2436, 'elation': 2437, 'covered': 2438, 'winter': 2439, 'riding': 2440, 'cherry': 2441, 'coal': 2442, 'falter': 2443, 'bowed': 2444, 'bonnet': 2445, 'courteous': 2446, 'looks': 2447, 'engaging': 2448, 'sell': 2449, 'purse': 2450, 'yearly': 2451, 'need': 2452, 'market': 2453, 'gain': 2454, 'dearly': 2455, 'tarry': 2456, 'although': 2457, 'parlay': 2458, 'ranks': 2459, 'girded': 2460, 'slung': 2461, 'warrior': 2462, 'bard': 2463, 'betrays': 2464, 'rights': 2465, 'faithful': 2466, 'chords': 2467, 'asunder': 2468, 'sully': 2469, 'bravry': 2470, 'londons': 2471, 'sight': 2472, 'workin': 2473, 'sow': 2474, 'wheat': 2475, 'gangs': 2476, 'sweep': 2477, 'expressed': 2478, 'london': 2479, 'top': 2480, 'dresses': 2481, 'bath': 2482, 'startin': 2483, 'fashions': 2484, 'mccree': 2485, 'nature': 2486, 'designed': 2487, 'complexions': 2488, 'cream': 2489, 'regard': 2490, 'sip': 2491, 'colors': 2492, 'wait': 2493, 'waitin': 2494, 'sweeps': 2495, 'beauing': 2496, 'belling': 2497, 'windows': 2498, 'cursing': 2499, 'faster': 2500, 'waiters': 2501, 'bailiffs': 2502, 'duns': 2503, 'bacchus': 2504, 'begotten': 2505, 'politicians': 2506, 'funds': 2507, 'dadda': 2508, 'living': 2509, 'drives': 2510, 'having': 2511, 'racking': 2512, 'tenants': 2513, 'stewards': 2514, 'teasing': 2515, 'raising': 2516, 'wishing': 2517, 'sunny': 2518, 'doves': 2519, 'coo': 2520, 'neath': 2521, 'sunbeam': 2522, 'robin': 2523, 'waters': 2524, 'larks': 2525, 'join': 2526, 'breaks': 2527, 'oftimes': 2528, 'lilies': 2529, 'declining': 2530, 'vale': 2531, 'shades': 2532, 'mantle': 2533, 'spreading': 2534, 'listening': 2535, 'shedding': 2536, 'beginning': 2537, 'spinning': 2538, 'blind': 2539, 'drowsily': 2540, 'knitting': 2541, 'cheerily': 2542, 'noiselessly': 2543, 'whirring': 2544, 'foots': 2545, 'stirring': 2546, 'sprightly': 2547, 'chara': 2548, 'tapping': 2549, 'ivy': 2550, 'flapping': 2551, 'somebody': 2552, 'sighing': 2553, 'autumn': 2554, 'noise': 2555, 'chirping': 2556, 'holly': 2557, 'shoving': 2558, 'wrong': 2559, 'coolin': 2560, 'casement': 2561, 'rove': 2562, 'moons': 2563, 'brightly': 2564, 'shakes': 2565, 'lays': 2566, 'longs': 2567, 'lingers': 2568, 'glance': 2569, 'puts': 2570, 'lazily': 2571, 'easily': 2572, 'lowly': 2573, 'reels': 2574, 'noiseless': 2575, 'leaps': 2576, 'ere': 2577, 'lovers': 2578, 'roved': 2579, 'verdant': 2580, 'braes': 2581, 'skreen': 2582, 'countrie': 2583, 'foreign': 2584, 'strand': 2585, 'dewy': 2586, 'climb': 2587, 'rob': 2588, 'boat': 2589, 'sails': 2590, 'loaded': 2591, 'sink': 2592, 'leaned': 2593, 'oak': 2594, 'trusty': 2595, 'false': 2596, 'reached': 2597, 'pricked': 2598, 'waxes': 2599, 'fades': 2600, 'wholl': 2601, 'cockle': 2602, 'gloom': 2603, 'news': 2604, 'forbid': 2605, 'patricks': 2606, 'napper': 2607, 'tandy': 2608, 'hows': 2609, 'distressful': 2610, 'englands': 2611, 'remind': 2612, 'pull': 2613, 'throw': 2614, 'sod': 2615, 'root': 2616, 'underfoot': 2617, 'laws': 2618, 'blades': 2619, 'growin': 2620, 'dare': 2621, 'show': 2622, 'caubeen': 2623, 'year': 2624, 'returning': 2625, 'store': 2626, 'ale': 2627, 'frequent': 2628, 'landlady': 2629, 'credit': 2630, 'custom': 2631, 'sovereigns': 2632, 'landladys': 2633, 'wines': 2634, 'confess': 2635, 'pardon': 2636, 'prodigal': 2637, 'caress': 2638, 'forgive': 2639, 'ofttimes': 2640, 'wondering': 2641, 'powr': 2642, 'beguile': 2643, 'teardrop': 2644, 'lilting': 2645, 'laughters': 2646, 'twinkle': 2647, 'lilt': 2648, 'seems': 2649, 'linnets': 2650, 'real': 2651, 'regret': 2652, 'throughout': 2653, 'youths': 2654, 'chance': 2655, 'spied': 2656, 'receiver': 2657, 'counted': 2658, 'penny': 2659, 'bu': 2660, 'rungum': 2661, 'chamber': 2662, 'course': 2663, 'charges': 2664, 'filled': 2665, 'ready': 2666, 'footmen': 2667, 'likewise': 2668, 'draw': 2669, 'pistol': 2670, 'couldnt': 2671, 'shoot': 2672, 'robbin': 2673, 'jailer': 2674, 'tight': 2675, 'fisted': 2676, 'army': 2677, 'stationed': 2678, 'cork': 2679, 'roamin': 2680, 'swear': 2681, 'treat': 2682, 'sportin': 2683, 'hurley': 2684, 'bollin': 2685, 'maids': 2686, 'summertime': 2687, 'pluck': 2688, 'yon': 2689}\n",
            "2690\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Turning Sequences into Input Sequencess"
      ],
      "metadata": {
        "id": "3dbbCt-cqnnH"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Now that you’ve seen how to take words and tokenize them into numbers, the next\n",
        "step is to encode the sentences into sequences of numbers. The tokenizer has a\n",
        "method for this called text_to_sequences—all you have to do is pass it your list of\n",
        "sentences, and it will give you back a list of sequences.\n",
        "\n",
        "But When predicting text, you need to train a neural network with an input sequence\n",
        "(feature) that has an associated label. Matching sequences to labels is the key to pre‐\n",
        "dicting text.\n",
        "\n",
        "To do this you’ll need to go through each line in the corpus and turn it into a list of\n",
        "tokens using texts_to_sequences. Then you can split each list by looping through\n",
        "each token and making a list of all the tokens up to it."
      ],
      "metadata": {
        "id": "fd467wi5qsMk"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "input_sequences = []\n",
        "for line in corpus:\n",
        "\ttoken_list = tokenizer.texts_to_sequences([line])[0]\n",
        "\tfor i in range(1, len(token_list)):\n",
        "\t\tn_gram_sequence = token_list[:i+1]\n",
        "\t\tinput_sequences.append(n_gram_sequence)"
      ],
      "metadata": {
        "id": "SUzlMYv0mVLH"
      },
      "execution_count": 4,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "Once you have these input sequences, you can pad them into a regular shape. We’ll\n",
        "use prepadding.\n",
        "\n",
        "To do this, you’ll need to find the longest sentence in the input sequences, and pad\n",
        "everything to that length"
      ],
      "metadata": {
        "id": "GNkc8iswsMwF"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "max_sequence_len = max([len(x) for x in input_sequences])\n",
        "input_sequences = np.array(pad_sequences(input_sequences, maxlen=max_sequence_len, padding='pre'))"
      ],
      "metadata": {
        "id": "RsMjPCLmmYKm"
      },
      "execution_count": 5,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "Finally, once you have a set of padded input sequences, you can split these into features and labels, where the label is simply the last token in the input sequence"
      ],
      "metadata": {
        "id": "fT2MRgBAsjBq"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "xs, labels = input_sequences[:,:-1],input_sequences[:,-1]"
      ],
      "metadata": {
        "id": "9gUN0Xu1mads"
      },
      "execution_count": 6,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "Next, you need to encode the label. To encode your labels into a set of Ys that you can then use to train, you can use the\n",
        "to_categorical utility in tf.keras:"
      ],
      "metadata": {
        "id": "AgKlg373s0sR"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "ys = tf.keras.utils.to_categorical(labels, num_classes=total_words)"
      ],
      "metadata": {
        "id": "Gd0YsOiRmdHT"
      },
      "execution_count": 7,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Creating the Model\n",
        "\n",
        "Let’s now create a simple model that can be trained with this input data. It will consist of just an embedding layer, followed by two LSTM, followed by a dense layer.\n",
        "\n",
        "For the embedding you’ll need one vector per word, so the parameters will be the\n",
        "total number of words and the number of dimensions you want to embed on. \n",
        "\n",
        "You can make the LSTM bidirectional, and the number of steps can be the length of a sequence, which is our max length minus 1 (because we took one token off the end to make the label). We are using multiple\n",
        "stacked LSTMs. This is pretty straightforward—just ensure that you set\n",
        "return_sequences to True on the first of them.\n",
        "\n",
        "Finally, the output layer will be a dense layer with the total number of words as a parameter, activated by softmax. Each neuron in this layer will be the probability that the next word matches the word for that index value:\n",
        "\n",
        "Compile the model with a categorical loss function such as categorical cross entropy and an optimizer like Adam. You can also specify that you want to capture metrics."
      ],
      "metadata": {
        "id": "7JMI8LXPtDKr"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model = Sequential()\n",
        "model.add(Embedding(total_words, 8))\n",
        "model.add(Bidirectional(LSTM(max_sequence_len-1, return_sequences='True')))\n",
        "model.add(Bidirectional(LSTM(max_sequence_len-1)))\n",
        "model.add(Dense(total_words, activation='softmax'))\n",
        "model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])\n",
        "history = model.fit(xs, ys, epochs=10, verbose=1)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "NoiCUoMVjXs2",
        "outputId": "c1eac253-5ac8-42bf-b4b5-a893092ff812"
      },
      "execution_count": 8,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/10\n",
            "377/377 [==============================] - 21s 22ms/step - loss: 6.8550 - accuracy: 0.0652\n",
            "Epoch 2/10\n",
            "377/377 [==============================] - 8s 22ms/step - loss: 6.3941 - accuracy: 0.0657\n",
            "Epoch 3/10\n",
            "377/377 [==============================] - 9s 23ms/step - loss: 6.2822 - accuracy: 0.0656\n",
            "Epoch 4/10\n",
            "377/377 [==============================] - 8s 22ms/step - loss: 6.2019 - accuracy: 0.0674\n",
            "Epoch 5/10\n",
            "377/377 [==============================] - 8s 22ms/step - loss: 6.1262 - accuracy: 0.0712\n",
            "Epoch 6/10\n",
            "377/377 [==============================] - 8s 21ms/step - loss: 6.0520 - accuracy: 0.0741\n",
            "Epoch 7/10\n",
            "377/377 [==============================] - 8s 22ms/step - loss: 5.9876 - accuracy: 0.0763\n",
            "Epoch 8/10\n",
            "377/377 [==============================] - 8s 22ms/step - loss: 5.9310 - accuracy: 0.0796\n",
            "Epoch 9/10\n",
            "377/377 [==============================] - 8s 22ms/step - loss: 5.8814 - accuracy: 0.0814\n",
            "Epoch 10/10\n",
            "377/377 [==============================] - 9s 23ms/step - loss: 5.8352 - accuracy: 0.0848\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "import matplotlib.pyplot as plt\n",
        "\n",
        "\n",
        "def plot_graphs(history, string):\n",
        "  plt.plot(history.history[string])\n",
        "  plt.xlabel(\"Epochs\")\n",
        "  plt.ylabel(string)\n",
        "  plt.show()\n",
        "\n",
        "plot_graphs(history, 'accuracy')\n",
        "plot_graphs(history, 'loss')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 541
        },
        "id": "SuJmt8mLjjIF",
        "outputId": "e00a1b8e-63bb-4385-929a-348b5c6ad34f"
      },
      "execution_count": 9,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Generating Text\n",
        "Now that you’ve trained a network that can predict the next word in a sequence, the\n",
        "next step is to give it a sequence of text and have it predict the next word. Let’s take a\n",
        "look at how to do that"
      ],
      "metadata": {
        "id": "VhhvPCkEvg-W"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Predicting the Next Word\n",
        "You’ll start by creating a phrase called the seed text. This is the initial expression on which the network will base all the content it generates. It will do this by predicting the next word.\n",
        "\n",
        "Next you need to tokenize this using texts_to_sequences. This returns an array,\n",
        "even if there’s only one value, so take the first element in that array.\n",
        "\n",
        "Then you need to pad that sequence to get it into the same shape as the data used for training:\n",
        "\n",
        "Now you can predict the next word for this token list by calling model.predict on the token list. This will return the probabilities for each word in the corpus, so pass the results to np.argmax to get the most likely one.\n",
        "\n",
        "You can look it up in code by searching through the word index items until you find predicted and printing it out."
      ],
      "metadata": {
        "id": "pTvVN4ZzwFk5"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "seed_text = \"sweet jeremy saw dublin\"\n",
        "\n",
        "token_list = tokenizer.texts_to_sequences([seed_text])[0]\n",
        "token_list = pad_sequences([token_list], maxlen=max_sequence_len-1, padding='pre')\n",
        "#print(model.predict(token_list))  \n",
        "#predicted = model.predict_classes(token_list)\n",
        "predicted = np.argmax(model.predict(token_list), axis=-1)\n",
        "pred_classes=model.predict(token_list)\n",
        "print(pred_classes.reshape(-1)[predicted])\n",
        "print(predicted)\n",
        "for word, index in tokenizer.word_index.items():\n",
        "\tif index == predicted:\n",
        "\t\tprint(word)\n",
        "\t\tbreak"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "GrRs9Xi_jtXN",
        "outputId": "c69a9d7f-0b6b-4d3f-d46c-c8d3222d1a88"
      },
      "execution_count": 10,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[0.02533435]\n",
            "[22]\n",
            "love\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Compounding Predictions to Generate Text\n",
        "In the previous section you saw how to use the model to predict the next word given a seed text. To have the neural network now create new text, you simply repeat the prediction, adding new words each time"
      ],
      "metadata": {
        "id": "HU3GwJoSxYMB"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "seed_text = \"sweet jeremy saw dublin\"\n",
        "next_words = 10\n",
        "\n",
        "  \n",
        "for _ in range(next_words):\n",
        "\ttoken_list = tokenizer.texts_to_sequences([seed_text])[0]\n",
        "\ttoken_list = pad_sequences([token_list], maxlen=max_sequence_len-1, padding='pre')\n",
        "\t#predicted = model.predict_classes(token_list, verbose=0)\n",
        "\tpredicted = np.argmax(model.predict(token_list), axis=-1)\n",
        "\toutput_word = \"\"\n",
        "\tfor word, index in tokenizer.word_index.items():\n",
        "\t\tif index == predicted:\n",
        "\t\t\toutput_word = word\n",
        "\t\t\tbreak\n",
        "\tseed_text += \" \" + output_word\n",
        "print(seed_text)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "9ZDeh6Ckjt9N",
        "outputId": "1cd326ea-cf02-4243-e24a-da02af538567"
      },
      "execution_count": 11,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "sweet jeremy saw dublin love the heart of the land of the land of\n"
          ]
        }
      ]
    }
  ]
}