19/dl_19.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Recurrent Neural Networks: Application"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import random\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import tensorflow as tf"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "with open(\"dinos.txt\", \"r\") as file:\n",
    "    data = file.readlines()\n",
    "    data = '<eos>'.join(data).lower()\n",
    "    data = data.split('<eos>')\n",
    "    random.shuffle(data)\n",
    "\n",
    "chars = ''.join(data)\n",
    "vocab = sorted(list(set(chars)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['serendipaceratops\\n',\n",
       " 'peishansaurus\\n',\n",
       " 'pneumatoarthrus\\n',\n",
       " 'vulcanodon\\n',\n",
       " 'dimodosaurus\\n',\n",
       " 'coronosaurus\\n',\n",
       " 'shuangbaisaurus\\n',\n",
       " 'eugongbusaurus\\n',\n",
       " 'canardia\\n',\n",
       " 'megalosaurus\\n',\n",
       " 'kittysaurus\\n',\n",
       " 'saichania\\n',\n",
       " 'arcusaurus\\n',\n",
       " 'dandakosaurus\\n',\n",
       " 'siamodracon\\n',\n",
       " 'leptospondylus\\n',\n",
       " 'cheneosaurus\\n',\n",
       " 'opisthocoelicaudia\\n',\n",
       " 'gongpoquansaurus\\n',\n",
       " 'megadontosaurus\\n',\n",
       " 'adamantisaurus\\n',\n",
       " 'quilmesaurus\\n',\n",
       " 'notohypsilophodon\\n',\n",
       " 'diabloceratops\\n',\n",
       " 'pyroraptor\\n',\n",
       " 'loncosaurus\\n',\n",
       " 'aragosaurus\\n',\n",
       " 'mantellodon\\n',\n",
       " 'chuxiongosaurus\\n',\n",
       " 'brasileosaurus\\n',\n",
       " 'serikornis\\n',\n",
       " 'zhenyuanlong\\n',\n",
       " 'shidaisaurus\\n',\n",
       " 'mantellisaurus\\n',\n",
       " 'auroraceratops\\n',\n",
       " 'archaeornis\\n',\n",
       " 'yibinosaurus\\n',\n",
       " 'zupaysaurus\\n',\n",
       " 'vectensia\\n',\n",
       " 'adeopapposaurus\\n',\n",
       " 'velociraptor\\n',\n",
       " 'syntarsus\\n',\n",
       " 'pachyrhinosaurus\\n',\n",
       " 'byronosaurus\\n',\n",
       " 'avaceratops\\n',\n",
       " 'geminiraptor\\n',\n",
       " 'parrosaurus\\n',\n",
       " 'hierosaurus\\n',\n",
       " 'protorosaurus\\n',\n",
       " 'edmontonia\\n',\n",
       " 'trimucrodon\\n',\n",
       " 'xenoposeidon\\n',\n",
       " 'gobivenator\\n',\n",
       " 'achelousaurus\\n',\n",
       " 'palaeoctonus\\n',\n",
       " 'leonerasaurus\\n',\n",
       " 'barilium\\n',\n",
       " 'euhelopus\\n',\n",
       " 'byranjaffia\\n',\n",
       " 'variraptor\\n',\n",
       " 'acanthopholis\\n',\n",
       " 'graciliraptor\\n',\n",
       " 'velocipes\\n',\n",
       " 'chialingosaurus\\n',\n",
       " 'dysganus\\n',\n",
       " 'rhabdodon\\n',\n",
       " 'spinosaurus\\n',\n",
       " 'muyelensaurus\\n',\n",
       " 'asiatosaurus\\n',\n",
       " 'ouranosaurus\\n',\n",
       " 'gojirasaurus\\n',\n",
       " 'animantarx\\n',\n",
       " 'delapparentia\\n',\n",
       " 'olorotitan\\n',\n",
       " 'lusotitan\\n',\n",
       " 'abydosaurus\\n',\n",
       " 'malawisaurus\\n',\n",
       " 'morelladon\\n',\n",
       " 'breviceratops\\n',\n",
       " 'nteregosaurus\\n',\n",
       " 'proceratosaurus\\n',\n",
       " 'cerasinops\\n',\n",
       " 'stenotholus\\n',\n",
       " 'dromaeosauroides\\n',\n",
       " 'ankistrodon\\n',\n",
       " 'hungarosaurus\\n',\n",
       " 'losillasaurus\\n',\n",
       " 'gavinosaurus\\n',\n",
       " 'mandschurosaurus\\n',\n",
       " 'borealosaurus\\n',\n",
       " 'struthiomimus\\n',\n",
       " 'erlikosaurus\\n',\n",
       " 'lophorhothon\\n',\n",
       " 'wyomingraptor\\n',\n",
       " 'elrhazosaurus\\n',\n",
       " 'teleocrater\\n',\n",
       " 'shunosaurus\\n',\n",
       " 'soriatitan\\n',\n",
       " 'basutodon\\n',\n",
       " 'rahiolisaurus\\n',\n",
       " 'tenontosaurus\\n',\n",
       " 'euoplocephalus\\n',\n",
       " 'penelopognathus\\n',\n",
       " 'sirindhorna\\n',\n",
       " 'squalodon\\n',\n",
       " 'willinakaqe\\n',\n",
       " 'linhevenator\\n',\n",
       " 'sonorasaurus\\n',\n",
       " 'fukuititan\\n',\n",
       " 'gallimimus\\n',\n",
       " 'deinocheirus\\n',\n",
       " 'orthogoniosaurus\\n',\n",
       " 'bellusaurus\\n',\n",
       " 'brohisaurus\\n',\n",
       " 'microcephale\\n',\n",
       " 'megaraptor\\n',\n",
       " 'muttaburrasaurus\\n',\n",
       " 'judiceratops\\n',\n",
       " 'taurovenator\\n',\n",
       " 'vahiny\\n',\n",
       " 'argentinosaurus\\n',\n",
       " 'qingxiusaurus\\n',\n",
       " 'raptorex\\n',\n",
       " 'teratophoneus\\n',\n",
       " 'osmakasaurus\\n',\n",
       " 'coelophysis\\n',\n",
       " 'pukyongosaurus\\n',\n",
       " 'rhoetosaurus\\n',\n",
       " 'magnamanus\\n',\n",
       " 'microcoelus\\n',\n",
       " 'regnosaurus\\n',\n",
       " 'bradycneme\\n',\n",
       " 'maleevus\\n',\n",
       " 'procheneosaurus\\n',\n",
       " 'sellosaurus\\n',\n",
       " 'rachitrema\\n',\n",
       " 'huaxiaosaurus\\n',\n",
       " 'crichtonpelta\\n',\n",
       " 'tomodon\\n',\n",
       " 'zapalasaurus\\n',\n",
       " 'wellnhoferia\\n',\n",
       " 'astrodon\\n',\n",
       " 'szechuanosaurus\\n',\n",
       " 'sanpasaurus\\n',\n",
       " 'zhuchengosaurus\\n',\n",
       " 'prenoceratops\\n',\n",
       " 'erketu\\n',\n",
       " 'narambuenatitan\\n',\n",
       " 'crichtonsaurus\\n',\n",
       " 'calamosaurus\\n',\n",
       " 'kuszholia\\n',\n",
       " 'zhuchengtitan\\n',\n",
       " 'yamaceratops\\n',\n",
       " 'limnornis\\n',\n",
       " 'dynamosaurus\\n',\n",
       " 'brachytaenius\\n',\n",
       " 'ultrasauros\\n',\n",
       " 'ningyuansaurus\\n',\n",
       " 'macelognathus\\n',\n",
       " 'anatosaurus\\n',\n",
       " 'velocisaurus\\n',\n",
       " 'calamospondylus\\n',\n",
       " 'klamelisauruskol\\n',\n",
       " 'griphornis\\n',\n",
       " 'illustration\\n',\n",
       " 'blikanasaurus\\n',\n",
       " 'titanosaurus\\n',\n",
       " 'rioarribasaurus\\n',\n",
       " 'kentrosaurus\\n',\n",
       " 'linheraptor\\n',\n",
       " 'malarguesaurus\\n',\n",
       " 'macrophalangia\\n',\n",
       " 'sphenospondylus\\n',\n",
       " 'proiguanodon\\n',\n",
       " 'galtonia\\n',\n",
       " 'qantassaurus\\n',\n",
       " 'theiophytalia\\n',\n",
       " 'haplocanthus\\n',\n",
       " 'amargasaurus\\n',\n",
       " 'xinjiangovenator\\n',\n",
       " 'lamplughsaura\\n',\n",
       " 'therosaurus\\n',\n",
       " 'hoplosaurus\\n',\n",
       " 'dryptosauroides\\n',\n",
       " 'kotasaurus\\n',\n",
       " 'martharaptor\\n',\n",
       " 'sanjuansaurus\\n',\n",
       " 'astrophocaudia\\n',\n",
       " 'levnesovia\\n',\n",
       " 'dysalotosaurus\\n',\n",
       " 'protarchaeopteryx\\n',\n",
       " 'chinshakiangosaurus\\n',\n",
       " 'protoceratops\\n',\n",
       " 'epanterias\\n',\n",
       " 'beipiaosaurus\\n',\n",
       " 'jingshanosaurus\\n',\n",
       " 'honghesaurus\\n',\n",
       " 'mussaurus\\n',\n",
       " 'angloposeidon\\n',\n",
       " 'lophostropheus\\n',\n",
       " 'lohuecotitan\\n',\n",
       " 'sauraechinodon\\n',\n",
       " 'haplocheirus\\n',\n",
       " 'lagerpeton\\n',\n",
       " 'metriacanthosaurus\\n',\n",
       " 'amtosaurus\\n',\n",
       " 'suchomimus\\n',\n",
       " 'cionodon\\n',\n",
       " 'apatosaurus\\n',\n",
       " 'shanyangosaurus\\n',\n",
       " 'uberabatitan\\n',\n",
       " 'tatankaceratops\\n',\n",
       " 'algoasaurus\\n',\n",
       " 'polyodontosaurus\\n',\n",
       " 'corythosaurus\\n',\n",
       " 'gravitholus\\n',\n",
       " 'platyceratops\\n',\n",
       " 'agathaumas\\n',\n",
       " 'tsaagan\\n',\n",
       " 'dasygnathoides\\n',\n",
       " 'huanghetitan\\n',\n",
       " 'sauropelta\\n',\n",
       " 'leyesaurus\\n',\n",
       " 'turiasaurus\\n',\n",
       " 'arrhinoceratops\\n',\n",
       " 'luanpingosaurus\\n',\n",
       " 'venenosaurus\\n',\n",
       " 'tsintaosaurus\\n',\n",
       " 'titanoceratops\\n',\n",
       " 'huanansaurus\\n',\n",
       " 'tetragonosaurus\\n',\n",
       " 'marasuchus\\n',\n",
       " 'ferganasaurus\\n',\n",
       " 'avipes\\n',\n",
       " 'maleevosaurus\\n',\n",
       " 'kileskus\\n',\n",
       " 'bihariosaurus\\n',\n",
       " 'acristavus\\n',\n",
       " 'brachyceratops\\n',\n",
       " 'hoplitosaurus\\n',\n",
       " 'mamenchisaurus\\n',\n",
       " 'sonidosaurus\\n',\n",
       " 'nyasasaurus\\n',\n",
       " 'cumnoria\\n',\n",
       " 'huehuecanauhtlus\\n',\n",
       " 'banji\\n',\n",
       " 'microceratops\\n',\n",
       " 'palaeopteryx\\n',\n",
       " 'paralititan\\n',\n",
       " 'pitekunsaurus\\n',\n",
       " 'europelta\\n',\n",
       " 'postosuchus\\n',\n",
       " 'stokesosaurus\\n',\n",
       " 'mojoceratops\\n',\n",
       " 'microraptor\\n',\n",
       " 'murusraptor\\n',\n",
       " 'torosaurus\\n',\n",
       " 'palaeosaurus\\n',\n",
       " 'othnielosaurus\\n',\n",
       " 'dongyangosaurus\\n',\n",
       " 'afrovenator\\n',\n",
       " 'picrodon\\n',\n",
       " 'gorgosaurus\\n',\n",
       " 'aardonyx\\n',\n",
       " 'manospondylus\\n',\n",
       " 'tonouchisaurus\\n',\n",
       " 'eotyrannus\\n',\n",
       " 'epichirostenotes\\n',\n",
       " 'anatotitan\\n',\n",
       " 'trialestes\\n',\n",
       " 'yunxianosaurus\\n',\n",
       " 'monoclonius\\n',\n",
       " 'archaeornithoides\\n',\n",
       " 'limaysaurus\\n',\n",
       " 'apatoraptor\\n',\n",
       " 'tichosteus\\n',\n",
       " 'agujaceratops\\n',\n",
       " 'clasmodosaurus\\n',\n",
       " 'crataeomus\\n',\n",
       " 'proa\\n',\n",
       " 'talenkauen\\n',\n",
       " 'morinosaurus\\n',\n",
       " 'hanwulosaurus\\n',\n",
       " 'bakesaurus\\n',\n",
       " 'spinophorosaurus\\n',\n",
       " 'atacamatitan\\n',\n",
       " 'nanningosaurus\\n',\n",
       " 'nemegtomaia\\n',\n",
       " 'rahonavis\\n',\n",
       " 'stygimoloch\\n',\n",
       " 'jintasaurus\\n',\n",
       " 'aucasaurus\\n',\n",
       " 'probrachylophosaurus\\n',\n",
       " 'coelurosauravus\\n',\n",
       " 'syngonosaurus\\n',\n",
       " 'lepidocheirosaurus\\n',\n",
       " 'prosaurolophus\\n',\n",
       " 'echinodon\\n',\n",
       " 'sinocalliopteryx\\n',\n",
       " 'shamosaurus\\n',\n",
       " 'pinacosaurus\\n',\n",
       " 'zaraapelta\\n',\n",
       " 'hadrosaurus\\n',\n",
       " 'jiangxisaurus\\n',\n",
       " 'betasuchus\\n',\n",
       " 'pandoravenator\\n',\n",
       " 'xixiasaurus\\n',\n",
       " 'beelemodon\\n',\n",
       " 'eurolimnornis\\n',\n",
       " 'goyocephale\\n',\n",
       " 'astrodontaurus\\n',\n",
       " 'iguanacolossus\\n',\n",
       " 'dakosaurus\\n',\n",
       " 'clepsysaurus\\n',\n",
       " 'jensenosaurus\\n',\n",
       " 'acheroraptor\\n',\n",
       " 'cruxicheiros\\n',\n",
       " 'riojasuchus\\n',\n",
       " 'boreonykus\\n',\n",
       " 'diplotomodon\\n',\n",
       " 'yunganglong\\n',\n",
       " 'propanoplosaurus\\n',\n",
       " 'lamaceratops\\n',\n",
       " 'deinodon\\n',\n",
       " 'ratchasimasaurus\\n',\n",
       " 'ornithopsis\\n',\n",
       " 'ischyrosaurus\\n',\n",
       " 'xianshanosaurus\\n',\n",
       " 'chuandongocoelurus\\n',\n",
       " 'syrmosaurus\\n',\n",
       " 'nodocephalosaurus\\n',\n",
       " 'ammosaurus\\n',\n",
       " 'massospondylus\\n',\n",
       " 'technosaurus\\n',\n",
       " 'elaltitan\\n',\n",
       " 'luanchuanraptor\\n',\n",
       " 'coloradia\\n',\n",
       " 'citipati\\n',\n",
       " 'frenguellisaurus\\n',\n",
       " 'morosaurus\\n',\n",
       " 'liaoningosaurus\\n',\n",
       " 'smilodon\\n',\n",
       " 'tornieria\\n',\n",
       " 'sugiyamasaurus\\n',\n",
       " 'laquintasaura\\n',\n",
       " 'rebbachisaurus\\n',\n",
       " 'conchoraptor\\n',\n",
       " 'emausaurus\\n',\n",
       " 'aviatyrannis\\n',\n",
       " 'cathetosaurus\\n',\n",
       " 'gracilisuchus\\n',\n",
       " 'medusaceratops\\n',\n",
       " 'gigantoscelus\\n',\n",
       " 'abdallahsaurus\\n',\n",
       " 'draconyx\\n',\n",
       " 'daemonosaurus\\n',\n",
       " 'pachysaurops\\n',\n",
       " 'ohmdenosaurus\\n',\n",
       " 'isaberrysaura\\n',\n",
       " 'orcomimus\\n',\n",
       " 'savannasaurus\\n',\n",
       " 'sarcosaurus\\n',\n",
       " 'lexovisaurus\\n',\n",
       " 'zhuchengceratops\\n',\n",
       " 'tarbosaurus\\n',\n",
       " 'indosaurus\\n',\n",
       " 'harpymimus\\n',\n",
       " 'zhanghenglong\\n',\n",
       " 'rhopalodon\\n',\n",
       " 'alaskacephale\\n',\n",
       " 'diplodocus\\n',\n",
       " 'caudipteryx\\n',\n",
       " 'urbacodon\\n',\n",
       " 'monkonosaurus\\n',\n",
       " 'eucoelophysis\\n',\n",
       " 'tsagantegia\\n',\n",
       " 'shuvuuia\\n',\n",
       " 'glacialisaurus\\n',\n",
       " 'shenzhousaurus\\n',\n",
       " 'daurosaurus\\n',\n",
       " 'jiutaisaurus\\n',\n",
       " 'panamericansaurus\\n',\n",
       " 'changdusaurus\\n',\n",
       " 'sauroniops\\n',\n",
       " 'utahraptor\\n',\n",
       " 'epidexipteryx\\n',\n",
       " 'nyororosaurus\\n',\n",
       " 'procerosaurus\\n',\n",
       " 'camptosaurus\\n',\n",
       " 'ekrixinatosaurus\\n',\n",
       " 'sahaliyania\\n',\n",
       " 'archaeoceratops\\n',\n",
       " 'pteropelyx\\n',\n",
       " 'vagaceratops\\n',\n",
       " 'camelotia\\n',\n",
       " 'sinopelta\\n',\n",
       " 'nothronychus\\n',\n",
       " 'tyrannotitan\\n',\n",
       " 'janenschia\\n',\n",
       " 'otogosaurus\\n',\n",
       " 'angulomastacator\\n',\n",
       " 'herrerasaurus\\n',\n",
       " 'araucanoraptor\\n',\n",
       " 'skorpiovenator\\n',\n",
       " 'tatankacephalus\\n',\n",
       " 'tapuiasaurus\\n',\n",
       " 'eucnemesaurus\\n',\n",
       " 'chilesaurus\\n',\n",
       " 'prenocephale\\n',\n",
       " 'tianyulong\\n',\n",
       " 'gastonia\\n',\n",
       " 'apatodon\\n',\n",
       " 'gwyneddosaurus\\n',\n",
       " 'ornithoides\\n',\n",
       " 'astrodonius\\n',\n",
       " 'uteodon\\n',\n",
       " 'zizhongosaurus\\n',\n",
       " 'pachyspondylus\\n',\n",
       " 'dromiceiomimus\\n',\n",
       " 'thescelosaurus\\n',\n",
       " 'jubbulpuria\\n',\n",
       " 'yezosaurus\\n',\n",
       " 'liaoceratops\\n',\n",
       " 'regaliceratops\\n',\n",
       " 'anoplosaurus\\n',\n",
       " 'tongtianlong\\n',\n",
       " 'nanuqsaurus\\n',\n",
       " 'chiayusaurus\\n',\n",
       " 'stegopelta\\n',\n",
       " 'pneumatoraptor\\n',\n",
       " 'jurapteryx\\n',\n",
       " 'teihivenator\\n',\n",
       " 'hippodraco\\n',\n",
       " 'oshanosaurus\\n',\n",
       " 'carnotaurus\\n',\n",
       " 'sidormimus\\n',\n",
       " 'magyarosaurus\\n',\n",
       " 'proplanicoxa\\n',\n",
       " 'pedopenna\\n',\n",
       " 'protiguanodon\\n',\n",
       " 'protognathosaurus\\n',\n",
       " 'fukuivenator\\n',\n",
       " 'jiangjunosaurus\\n',\n",
       " 'megapnosaurus\\n',\n",
       " 'triunfosaurus\\n',\n",
       " 'zhuchengtyrannus\\n',\n",
       " 'minmi\\n',\n",
       " 'siats\\n',\n",
       " 'damalasaurus\\n',\n",
       " 'aristosaurus\\n',\n",
       " 'unaysaurus\\n',\n",
       " 'rinconsaurus\\n',\n",
       " 'machairoceratops\\n',\n",
       " 'mymoorapelta\\n',\n",
       " 'gannansaurus\\n',\n",
       " 'kakuru\\n',\n",
       " 'montanoceratops\\n',\n",
       " 'monolophosaurus\\n',\n",
       " 'helopus\\n',\n",
       " 'jobaria\\n',\n",
       " 'chirostenotes\\n',\n",
       " 'albalophosaurus\\n',\n",
       " 'ornithomimus\\n',\n",
       " 'fendusaurus\\n',\n",
       " 'tarchia\\n',\n",
       " 'horshamosaurus\\n',\n",
       " 'batyrosaurus\\n',\n",
       " 'ephoenosaurus\\n',\n",
       " 'brontoraptor\\n',\n",
       " 'calamospondylus\\n',\n",
       " 'atlascopcosaurus\\n',\n",
       " 'brontomerus\\n',\n",
       " 'oviraptor\\n',\n",
       " 'chindesaurus\\n',\n",
       " 'scleromochlus\\n',\n",
       " 'krzyzanowskisaurus\\n',\n",
       " 'pellegrinisaurus\\n',\n",
       " 'daxiatitan\\n',\n",
       " 'zapsalis\\n',\n",
       " 'avimimus\\n',\n",
       " 'lametasaurus\\n',\n",
       " 'hadrosauravus\\n',\n",
       " 'centemodon\\n',\n",
       " 'yangchuanosaurus\\n',\n",
       " 'megadactylus\\n',\n",
       " 'hypsibema\\n',\n",
       " 'eupodosaurus\\n',\n",
       " 'dianchungosaurus\\n',\n",
       " 'fukuiraptor\\n',\n",
       " 'nanotyrannus\\n',\n",
       " 'zuniceratops\\n',\n",
       " 'koparion\\n',\n",
       " 'australodocus\\n',\n",
       " 'lapampasaurus\\n',\n",
       " 'sinosauropteryx\\n',\n",
       " 'marmarospondylus\\n',\n",
       " 'yueosaurus\\n',\n",
       " 'koshisaurus\\n',\n",
       " 'paronychodon\\n',\n",
       " 'psittacosaurus\\n',\n",
       " 'polacanthoides\\n',\n",
       " 'wyleyia\\n',\n",
       " 'ischioceratops\\n',\n",
       " 'amargastegos\\n',\n",
       " 'tendaguria\\n',\n",
       " 'pelorosaurus\\n',\n",
       " 'pleurocoelus\\n',\n",
       " 'juravenator\\n',\n",
       " 'tangvayosaurus\\n',\n",
       " 'bicentenaria\\n',\n",
       " 'becklespinax\\n',\n",
       " 'uintasaurus\\n',\n",
       " 'avalonianus\\n',\n",
       " 'chondrosteosaurus\\n',\n",
       " 'agnosphitys\\n',\n",
       " 'eoraptor\\n',\n",
       " 'lewisuchus\\n',\n",
       " 'appalachiosaurus\\n',\n",
       " 'europatitan\\n',\n",
       " 'dongbeititan\\n',\n",
       " 'cryptoraptor\\n',\n",
       " 'kelmayisaurus\\n',\n",
       " 'sanchusaurus\\n',\n",
       " 'fusuisaurus\\n',\n",
       " 'mifunesaurus\\n',\n",
       " 'efraasia\\n',\n",
       " 'antarctopelta\\n',\n",
       " 'laplatasaurus\\n',\n",
       " 'arcovenator\\n',\n",
       " 'diluvicursor\\n',\n",
       " 'haplocanthosaurus\\n',\n",
       " 'valdosaurus\\n',\n",
       " 'gobititan\\n',\n",
       " 'struthiosaurus\\n',\n",
       " 'yizhousaurus\\n',\n",
       " 'pachysauriscus\\n',\n",
       " 'kunmingosaurus\\n',\n",
       " 'cladeiodon\\n',\n",
       " 'angolatitan\\n',\n",
       " 'omeisaurus\\n',\n",
       " 'nasutoceratops\\n',\n",
       " 'overosaurus\\n',\n",
       " 'streptospondylus\\n',\n",
       " 'roccosaurus\\n',\n",
       " 'saurophaganax\\n',\n",
       " 'eolambia\\n',\n",
       " 'doryphorosaurus\\n',\n",
       " 'parasaurolophus\\n',\n",
       " 'lufengosaurus\\n',\n",
       " 'agustinia\\n',\n",
       " 'gilmoreosaurus\\n',\n",
       " 'oligosaurus\\n',\n",
       " 'ovoraptor\\n',\n",
       " 'coelurus\\n',\n",
       " 'anzu\\n',\n",
       " 'denversaurus\\n',\n",
       " 'fusinasus\\n',\n",
       " 'silvisaurus\\n',\n",
       " 'gansutitan\\n',\n",
       " 'panoplosaurus\\n',\n",
       " 'allosaurus\\n',\n",
       " 'yurgovuchia\\n',\n",
       " 'mononychus\\n',\n",
       " 'triassolestes\\n',\n",
       " 'velafrons\\n',\n",
       " 'austrocheirus\\n',\n",
       " 'wuerhosaurus\\n',\n",
       " 'barsboldia\\n',\n",
       " 'suuwassea\\n',\n",
       " 'ornithotarsus\\n',\n",
       " 'ostromia\\n',\n",
       " 'ugrunaaluk\\n',\n",
       " 'supersaurus\\n',\n",
       " 'parksosaurus\\n',\n",
       " 'xixiposaurus\\n',\n",
       " 'ugrosaurus\\n',\n",
       " 'duriatitan\\n',\n",
       " 'iguanodon\\n',\n",
       " 'danubiosaurus\\n',\n",
       " 'spiclypeus\\n',\n",
       " 'genusaurus\\n',\n",
       " 'gspsaurus\\n',\n",
       " 'elachistosuchus\\n',\n",
       " 'orosaurus\\n',\n",
       " 'barosaurus\\n',\n",
       " 'alwalkeria\\n',\n",
       " 'camptonotus\\n',\n",
       " 'neuquenraptor\\n',\n",
       " 'tianchungosaurus\\n',\n",
       " 'ceratosaurus\\n',\n",
       " 'ornithosuchus\\n',\n",
       " 'domeykosaurus\\n',\n",
       " 'scutellosaurus\\n',\n",
       " 'belodon\\n',\n",
       " 'chenanisaurus\\n',\n",
       " 'parasaurolophus\\n',\n",
       " 'thespesius\\n',\n",
       " 'aggiosaurus\\n',\n",
       " 'sinocoelurus\\n',\n",
       " 'sauroplites\\n',\n",
       " 'leipsanosaurus\\n',\n",
       " 'ischisaurus\\n',\n",
       " 'chaoyangsaurus\\n',\n",
       " 'albertavenator\\n',\n",
       " 'asiaceratops\\n',\n",
       " 'gideonmantellia\\n',\n",
       " 'xuanhanosaurus\\n',\n",
       " 'glyptodontopelta\\n',\n",
       " 'yunnanosaurus\\n',\n",
       " 'acrotholus\\n',\n",
       " 'liliensternus\\n',\n",
       " 'probactrosaurus\\n',\n",
       " 'rugocaudia\\n',\n",
       " 'siluosaurus\\n',\n",
       " 'caseosaurus\\n',\n",
       " 'xenoceratops\\n',\n",
       " 'gyposaurus\\n',\n",
       " 'zhongyuansaurus\\n',\n",
       " 'neimongosaurus\\n',\n",
       " 'tanystropheus\\n',\n",
       " 'diracodon\\n',\n",
       " 'huxleysaurus\\n',\n",
       " 'kitadanisaurus\\n',\n",
       " 'likhoelesaurus\\n',\n",
       " 'avisaurus\\n',\n",
       " 'shanshanosaurus\\n',\n",
       " 'bienosaurus\\n',\n",
       " 'chingkankousaurus\\n',\n",
       " 'walkersaurus\\n',\n",
       " 'vitakrisaurus\\n',\n",
       " 'compsognathus\\n',\n",
       " 'europasaurus\\n',\n",
       " 'abrosaurus\\n',\n",
       " 'tylosteus\\n',\n",
       " 'albisaurus\\n',\n",
       " 'shanag\\n',\n",
       " 'procompsognathus\\n',\n",
       " 'archaeodontosaurus\\n',\n",
       " 'xiongguanlong\\n',\n",
       " 'fukuisaurus\\n',\n",
       " 'peltosaurus\\n',\n",
       " 'mochlodon\\n',\n",
       " 'taveirosaurus\\n',\n",
       " 'borogovia\\n',\n",
       " 'macrogryphosaurus\\n',\n",
       " 'kayentavenator\\n',\n",
       " 'gongxianosaurus\\n',\n",
       " 'yulong\\n',\n",
       " 'yingshanosaurus\\n',\n",
       " 'zunityrannus\\n',\n",
       " 'limusaurus\\n',\n",
       " 'labocania\\n',\n",
       " 'masiakasaurus\\n',\n",
       " 'nanosaurus\\n',\n",
       " 'jainosaurus\\n',\n",
       " 'buriolestes\\n',\n",
       " 'kritosaurus\\n',\n",
       " 'lusitanosaurus\\n',\n",
       " 'sulaimanisaurus\\n',\n",
       " 'koutalisaurus\\n',\n",
       " 'itemirus\\n',\n",
       " 'saltriosaurus\\n',\n",
       " 'charonosaurus\\n',\n",
       " 'daanosaurus\\n',\n",
       " 'phaedrolosaurus\\n',\n",
       " 'skeleton\\n',\n",
       " 'microceratus\\n',\n",
       " 'pakisaurus\\n',\n",
       " 'sphaerotholus\\n',\n",
       " 'veterupristisaurus\\n',\n",
       " 'aepyornithomimus\\n',\n",
       " 'utahceratops\\n',\n",
       " 'alnashetri\\n',\n",
       " 'saltopus\\n',\n",
       " 'beibeilong\\n',\n",
       " 'aoniraptor\\n',\n",
       " 'tachiraptor\\n',\n",
       " 'concavenator\\n',\n",
       " 'longosaurus\\n',\n",
       " 'siamodon\\n',\n",
       " 'aquilops\\n',\n",
       " 'luoyanggia\\n',\n",
       " 'maiasaura\\n',\n",
       " 'neosaurus\\n',\n",
       " 'archaeornithomimus\\n',\n",
       " 'ornitholestes\\n',\n",
       " 'stegosaurides\\n',\n",
       " 'coelosaurus\\n',\n",
       " 'datonglong\\n',\n",
       " 'tanycolagreus\\n',\n",
       " 'juratyrant\\n',\n",
       " 'atlantosaurus\\n',\n",
       " 'xuanhuaceratops\\n',\n",
       " 'gigantoraptor\\n',\n",
       " 'tonganosaurus\\n',\n",
       " 'cedrorestes\\n',\n",
       " 'ojoceratops\\n',\n",
       " 'santanaraptor\\n',\n",
       " 'revueltosaurus\\n',\n",
       " 'yinlong\\n',\n",
       " 'macrodontophion\\n',\n",
       " 'bahariasaurus\\n',\n",
       " 'protecovasaurus\\n',\n",
       " 'amurosaurus\\n',\n",
       " 'saurornitholestes\\n',\n",
       " 'lanasaurus\\n',\n",
       " 'nopcsaspondylus\\n',\n",
       " 'elopteryx\\n',\n",
       " 'succinodon\\n',\n",
       " 'drinker\\n',\n",
       " 'ojoraptorsaurus\\n',\n",
       " 'ngexisaurus\\n',\n",
       " 'notoceratops\\n',\n",
       " 'camposaurus\\n',\n",
       " 'plateosauravus\\n',\n",
       " 'tylocephale\\n',\n",
       " 'brasilotitan\\n',\n",
       " 'craterosaurus\\n',\n",
       " 'scansoriopteryx\\n',\n",
       " 'tataouinea\\n',\n",
       " 'eustreptospondylus\\n',\n",
       " 'baotianmansaurus\\n',\n",
       " 'trinisaura\\n',\n",
       " 'guaibasaurus\\n',\n",
       " 'dongyangopelta\\n',\n",
       " 'burianosaurus\\n',\n",
       " 'huangshanlong\\n',\n",
       " 'priodontognathus\\n',\n",
       " 'orthomerus\\n',\n",
       " 'griphosaurus\\n',\n",
       " 'tribelesodon\\n',\n",
       " 'jaklapallisaurus\\n',\n",
       " 'chungkingosaurus\\n',\n",
       " 'nankangia\\n',\n",
       " 'aletopelta\\n',\n",
       " 'gryponyx\\n',\n",
       " 'manidens\\n',\n",
       " 'eucamerotus\\n',\n",
       " 'deltadromeus\\n',\n",
       " 'jeyawati\\n',\n",
       " 'genyodectes\\n',\n",
       " 'koreaceratops\\n',\n",
       " 'ponerosteus\\n',\n",
       " 'wiehenvenator\\n',\n",
       " 'zuul',\n",
       " 'ziapelta\\n',\n",
       " 'caenagnathus\\n',\n",
       " 'bonitasaura\\n',\n",
       " 'gigantosaurus\\n',\n",
       " 'mtotosaurus\\n',\n",
       " 'claosaurus\\n',\n",
       " 'lirainosaurus\\n',\n",
       " 'chasmosaurus\\n',\n",
       " 'thecospondylus\\n',\n",
       " 'ilokelesia\\n',\n",
       " 'sinusonasus\\n',\n",
       " 'anserimimus\\n',\n",
       " 'yehuecauhceratops\\n',\n",
       " 'azendohsaurus\\n',\n",
       " 'kulindadromeus\\n',\n",
       " 'tawa\\n',\n",
       " 'wadhurstia\\n',\n",
       " 'proyandusaurus\\n',\n",
       " 'eocursor\\n',\n",
       " 'jenghizkhan\\n',\n",
       " 'ampelosaurus\\n',\n",
       " 'pentaceratops\\n',\n",
       " 'amphisaurus\\n',\n",
       " 'edmarka\\n',\n",
       " 'heterosaurus\\n',\n",
       " 'majungatholus\\n',\n",
       " 'antrodemus\\n',\n",
       " 'rapetosaurus\\n',\n",
       " 'shaochilong\\n',\n",
       " 'albinykus\\n',\n",
       " 'shuvosaurus\\n',\n",
       " 'philovenator\\n',\n",
       " 'sigilmassasaurus\\n',\n",
       " 'fabrosaurus\\n',\n",
       " 'caulodon\\n',\n",
       " 'erectopus\\n',\n",
       " 'austroposeidon\\n',\n",
       " 'claorhynchus\\n',\n",
       " 'triceratops\\n',\n",
       " 'kazaklambia\\n',\n",
       " 'therizinosaurus\\n',\n",
       " 'kukufeldia\\n',\n",
       " 'centrosaurus\\n',\n",
       " 'dakotaraptor\\n',\n",
       " 'achillesaurus\\n',\n",
       " 'hortalotarsus\\n',\n",
       " 'amygdalodon\\n',\n",
       " 'tyrannosaurus\\n',\n",
       " 'ultrasaurus\\n',\n",
       " 'sinotyrannus\\n',\n",
       " 'orinosaurusorkoraptor\\n',\n",
       " 'titanosaurus\\n',\n",
       " 'viavenator\\n',\n",
       " 'lourinhasaurus\\n',\n",
       " 'rayososaurus\\n',\n",
       " 'dromicosaurus\\n',\n",
       " 'rileya\\n',\n",
       " 'sacisaurus\\n',\n",
       " 'choconsaurus\\n',\n",
       " 'moabosaurus\\n',\n",
       " 'rocasaurus\\n',\n",
       " 'campylodon\\n',\n",
       " 'leptorhynchos\\n',\n",
       " 'plesiohadros\\n',\n",
       " 'amphicoelias\\n',\n",
       " 'lesothosaurus\\n',\n",
       " 'koreanosaurus\\n',\n",
       " 'walgettosuchus\\n',\n",
       " 'heptasteornis\\n',\n",
       " 'hypselorhachis\\n',\n",
       " 'tethyshadros\\n',\n",
       " 'dromaeosaurus\\n',\n",
       " 'palaeoscincus\\n',\n",
       " 'yongjinglong\\n',\n",
       " 'dollodon\\n',\n",
       " 'rajasaurus\\n',\n",
       " 'saurolophus\\n',\n",
       " 'giganotosaurus\\n',\n",
       " 'datanglong\\n',\n",
       " 'stegoceras\\n',\n",
       " 'gigantspinosaurus\\n',\n",
       " 'qinlingosaurus\\n',\n",
       " 'hylaeosaurus\\n',\n",
       " 'mendozasaurus\\n',\n",
       " 'sphenosaurus\\n',\n",
       " 'kosmoceratops\\n',\n",
       " 'mei\\n',\n",
       " 'kunbarrasaurus\\n',\n",
       " 'leptoceratops\\n',\n",
       " 'nqwebasaurus\\n',\n",
       " 'tazoudasaurus\\n',\n",
       " 'nurosaurus\\n',\n",
       " 'yaverlandia\\n',\n",
       " 'bissektipelta\\n',\n",
       " 'galvesaurus\\n',\n",
       " 'lancangosaurus\\n',\n",
       " 'leinkupal\\n',\n",
       " 'laiyangosaurus\\n',\n",
       " 'priconodon\\n',\n",
       " 'sterrholophus\\n',\n",
       " 'xixianykus\\n',\n",
       " 'traukutitan\\n',\n",
       " 'gryphoceratops\\n',\n",
       " 'bolong\\n',\n",
       " 'zigongosaurus\\n',\n",
       " 'yimenosaurus\\n',\n",
       " 'isisaurus\\n',\n",
       " 'archaeoraptor\\n',\n",
       " 'razanandrongobe\\n',\n",
       " 'halticosaurus\\n',\n",
       " 'poposaurus\\n',\n",
       " 'laosaurus\\n',\n",
       " 'palaeocursornis\\n',\n",
       " 'sibirotitan\\n',\n",
       " 'unicerosaurus\\n',\n",
       " 'gobisaurus\\n',\n",
       " 'clarencea\\n',\n",
       " 'daliansaurus\\n',\n",
       " 'xenotarsosaurus\\n',\n",
       " 'yunmenglong\\n',\n",
       " 'eoceratops\\n',\n",
       " 'panphagia\\n',\n",
       " 'aachenosaurus\\n',\n",
       " 'valdoraptor\\n',\n",
       " 'niobrarasaurus\\n',\n",
       " 'sinopeltosaurus\\n',\n",
       " 'naashoibitosaurus\\n',\n",
       " 'podokesaurus\\n',\n",
       " 'futabasaurus\\n',\n",
       " 'embasaurus\\n',\n",
       " 'nectosaurus\\n',\n",
       " 'magnapaulia\\n',\n",
       " 'hanssuesia\\n',\n",
       " 'shanxia\\n',\n",
       " 'polyonax\\n',\n",
       " 'stenonychosaurus\\n',\n",
       " 'asylosaurus\\n',\n",
       " 'issasaurus\\n',\n",
       " 'lambeosaurus\\n',\n",
       " 'ligabuesaurus\\n',\n",
       " 'pectinodon\\n',\n",
       " 'chubutisaurus\\n',\n",
       " 'nambalia\\n',\n",
       " 'acrocanthosaurus\\n',\n",
       " 'leshansaurus\\n',\n",
       " 'hagryphus\\n',\n",
       " 'tenantosaurus\\n',\n",
       " 'aerosteon\\n',\n",
       " 'magnirostris\\n',\n",
       " 'aralosaurus\\n',\n",
       " 'mahakala\\n',\n",
       " 'proceratops\\n',\n",
       " 'zanclodon\\n',\n",
       " 'dinodocus\\n',\n",
       " 'tenchisaurus\\n',\n",
       " 'shantungosaurus\\n',\n",
       " 'linhenykus\\n',\n",
       " 'argyrosaurus\\n',\n",
       " 'wulagasaurus\\n',\n",
       " 'dystylosaurus\\n',\n",
       " 'ferganocephale\\n',\n",
       " 'daspletosaurus\\n',\n",
       " 'teratosaurus\\n',\n",
       " 'balaur\\n',\n",
       " 'microvenator\\n',\n",
       " 'umarsaurus\\n',\n",
       " 'mirischia\\n',\n",
       " 'poekilopleuron\\n',\n",
       " 'talos\\n',\n",
       " 'katsuyamasaurus\\n',\n",
       " 'styracosaurus\\n',\n",
       " 'paleosaurus\\n',\n",
       " 'palaeolimnornis\\n',\n",
       " 'herbstosaurus\\n',\n",
       " 'coeluroides\\n',\n",
       " 'changyuraptor\\n',\n",
       " 'ruyangosaurus\\n',\n",
       " 'godzillasaurus\\n',\n",
       " 'aurornis\\n',\n",
       " 'chilantaisaurus\\n',\n",
       " 'volkheimeria\\n',\n",
       " 'bonatitan\\n',\n",
       " 'cedarosaurus\\n',\n",
       " 'megacervixosaurus\\n',\n",
       " 'camarillasaurus\\n',\n",
       " 'panguraptor\\n',\n",
       " 'habodcraniosaurus\\n',\n",
       " 'agrosaurus\\n',\n",
       " 'rubeosaurus\\n',\n",
       " 'haya\\n',\n",
       " 'patagotitan\\n',\n",
       " 'gigantosaurus\\n',\n",
       " 'beishanlong\\n',\n",
       " 'zhejiangosaurus\\n',\n",
       " 'gryposaurus\\n',\n",
       " 'vouivria\\n',\n",
       " 'didanodondilong\\n',\n",
       " 'campylodoniscus\\n',\n",
       " 'chebsaurus\\n',\n",
       " 'chuanqilong\\n',\n",
       " 'jiangjunmiaosaurus\\n',\n",
       " 'symphyrophus\\n',\n",
       " 'tototlmimus\\n',\n",
       " 'dryosaurus\\n',\n",
       " 'sciurumimus\\n',\n",
       " 'hongshanosaurus\\n',\n",
       " 'elmisaurus\\n',\n",
       " 'arenysaurus\\n',\n",
       " 'bothriospondylus\\n',\n",
       " 'pulanesaura\\n',\n",
       " 'neuquensaurus\\n',\n",
       " 'hypselosaurus\\n',\n",
       " 'zuoyunlong\\n',\n",
       " 'yandusaurus\\n',\n",
       " 'corythoraptor\\n',\n",
       " 'udanoceratops\\n',\n",
       " 'venaticosuchus\\n',\n",
       " 'dracopelta\\n',\n",
       " 'yuanmousaurus\\n',\n",
       " 'nemegtosaurus\\n',\n",
       " 'piatnitzkysaurus\\n',\n",
       " 'oohkotokia\\n',\n",
       " 'dinheirosaurus\\n',\n",
       " 'ornithomimoides\\n',\n",
       " 'callovosaurus\\n',\n",
       " 'notocolossus\\n',\n",
       " 'microdontosaurus\\n',\n",
       " 'eohadrosaurus\\n',\n",
       " 'andesaurus\\n',\n",
       " 'koreanosaurus\\n',\n",
       " 'tehuelchesaurus\\n',\n",
       " 'jianchangosaurus\\n',\n",
       " 'sarcolestes\\n',\n",
       " 'velociraptor\\n',\n",
       " 'longisquama\\n',\n",
       " 'thecocoelurus\\n',\n",
       " 'angaturama\\n',\n",
       " 'nigersaurus\\n',\n",
       " 'latirhinus\\n',\n",
       " 'ceratops\\n',\n",
       " 'aorun\\n',\n",
       " 'ichthyovenator\\n',\n",
       " 'cathartesaura\\n',\n",
       " 'wannanosaurus\\n',\n",
       " 'ankylosaurus\\n',\n",
       " 'geranosaurus\\n',\n",
       " 'pycnonemosaurus\\n',\n",
       " 'seismosaurus\\n',\n",
       " 'gondwanatitan\\n',\n",
       " 'bagaraatan\\n',\n",
       " 'meroktenos\\n',\n",
       " 'gargoyleosaurus\\n',\n",
       " 'dinotyrannus\\n',\n",
       " 'dakotadon\\n',\n",
       " 'incisivosaurus\\n',\n",
       " ...]"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['\\n',\n",
       " 'a',\n",
       " 'b',\n",
       " 'c',\n",
       " 'd',\n",
       " 'e',\n",
       " 'f',\n",
       " 'g',\n",
       " 'h',\n",
       " 'i',\n",
       " 'j',\n",
       " 'k',\n",
       " 'l',\n",
       " 'm',\n",
       " 'n',\n",
       " 'o',\n",
       " 'p',\n",
       " 'q',\n",
       " 'r',\n",
       " 's',\n",
       " 't',\n",
       " 'u',\n",
       " 'v',\n",
       " 'w',\n",
       " 'x',\n",
       " 'y',\n",
       " 'z']"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "vocab"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "There are 1536 names and 27 unique tokens in your data.\n"
     ]
    }
   ],
   "source": [
    "data_size = len(data)\n",
    "vocab_size = len(vocab)\n",
    "\n",
    "print('There are {} names and {} unique tokens in your data.'.format(data_size, vocab_size))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "char_to_idx = {ch:i for i,ch in enumerate(vocab)}\n",
    "idx_to_char = {i:ch for i,ch in enumerate(vocab)}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'\\n': 0,\n",
       " 'a': 1,\n",
       " 'b': 2,\n",
       " 'c': 3,\n",
       " 'd': 4,\n",
       " 'e': 5,\n",
       " 'f': 6,\n",
       " 'g': 7,\n",
       " 'h': 8,\n",
       " 'i': 9,\n",
       " 'j': 10,\n",
       " 'k': 11,\n",
       " 'l': 12,\n",
       " 'm': 13,\n",
       " 'n': 14,\n",
       " 'o': 15,\n",
       " 'p': 16,\n",
       " 'q': 17,\n",
       " 'r': 18,\n",
       " 's': 19,\n",
       " 't': 20,\n",
       " 'u': 21,\n",
       " 'v': 22,\n",
       " 'w': 23,\n",
       " 'x': 24,\n",
       " 'y': 25,\n",
       " 'z': 26}"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "char_to_idx"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{0: '\\n',\n",
       " 1: 'a',\n",
       " 2: 'b',\n",
       " 3: 'c',\n",
       " 4: 'd',\n",
       " 5: 'e',\n",
       " 6: 'f',\n",
       " 7: 'g',\n",
       " 8: 'h',\n",
       " 9: 'i',\n",
       " 10: 'j',\n",
       " 11: 'k',\n",
       " 12: 'l',\n",
       " 13: 'm',\n",
       " 14: 'n',\n",
       " 15: 'o',\n",
       " 16: 'p',\n",
       " 17: 'q',\n",
       " 18: 'r',\n",
       " 19: 's',\n",
       " 20: 't',\n",
       " 21: 'u',\n",
       " 22: 'v',\n",
       " 23: 'w',\n",
       " 24: 'x',\n",
       " 25: 'y',\n",
       " 26: 'z'}"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "idx_to_char"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "max_length_name = max([len(d) for d in data])+5\n",
    "eos_id = char_to_idx['\\n']\n",
    "\n",
    "X_train = np.zeros(shape=(data_size, max_length_name, vocab_size), dtype='float32')\n",
    "for i, name in enumerate(data):\n",
    "    for j, character in enumerate(name):\n",
    "        index = char_to_idx[character]\n",
    "        X_train[i,j,index] = 1.0\n",
    "    X_train[i,len(name):,eos_id] = 1.0\n",
    "\n",
    "Y_train = np.zeros(shape=(data_size, max_length_name, vocab_size), dtype='float32')\n",
    "for i in range(data_size):\n",
    "    for j in range(max_length_name-1):\n",
    "        Y_train[i,j,:] = X_train[i,j+1,:]\n",
    "    Y_train[i,max_length_name-1,eos_id] = 1.0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "def Model(params):\n",
    "    dims = params['dims']\n",
    "    X_train = params['X_train']\n",
    "    Y_train = params['Y_train']\n",
    "    epochs = params['epochs']\n",
    "    learning_rate = params['learning_rate']\n",
    "    batch_size = params['batch_size']\n",
    "    f1 = tf.keras.layers.SimpleRNN(units=50,\n",
    "                                   activation='tanh',\n",
    "                                   kernel_initializer='glorot_uniform',\n",
    "                                   bias_initializer='zeros',\n",
    "                                   recurrent_initializer='zeros',\n",
    "                                   return_sequences=True, \n",
    "                                   name='rnn')\n",
    "    f2 = tf.keras.layers.Dense(units=dims[1][1], \n",
    "                               activation='softmax', \n",
    "                               kernel_initializer='glorot_uniform', \n",
    "                               bias_initializer='zeros',\n",
    "                               name='dense')\n",
    "    x = tf.keras.Input(shape=dims[0])\n",
    "    a = f1(x)\n",
    "    y = f2(a)\n",
    "    model = tf.keras.Model(x, y)    \n",
    "    optimizer = tf.keras.optimizers.Adam(learning_rate=learning_rate, \n",
    "                                         beta_1=0.9, \n",
    "                                         beta_2=0.999, \n",
    "                                         epsilon=1e-07)\n",
    "    model.compile(loss='categorical_crossentropy', metrics=['categorical_accuracy'], optimizer=optimizer)\n",
    "    model.summary()\n",
    "    callback1 = tf.keras.callbacks.ReduceLROnPlateau(monitor='loss', \n",
    "                                                     patience=200,\n",
    "                                                     min_delta=0.001,\n",
    "                                                     factor=0.1, \n",
    "                                                     min_lr=0.0001)\n",
    "    callback2 = tf.keras.callbacks.EarlyStopping(monitor='loss',\n",
    "                                                 patience=500,\n",
    "                                                 min_delta=0.001)\n",
    "    model.fit(X_train, Y_train, epochs=epochs, batch_size=batch_size, callbacks=[callback1, callback2])\n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_1 (InputLayer)         [(None, 32, 27)]          0         \n",
      "_________________________________________________________________\n",
      "rnn (SimpleRNN)              (None, 32, 50)            3900      \n",
      "_________________________________________________________________\n",
      "dense (Dense)                (None, 32, 27)            1377      \n",
      "=================================================================\n",
      "Total params: 5,277\n",
      "Trainable params: 5,277\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "Train on 1536 samples\n",
      "Epoch 1/1000\n",
      "1536/1536 [==============================] - 1s 542us/sample - loss: 1.3958 - categorical_accuracy: 0.6705\n",
      "Epoch 2/1000\n",
      "1536/1536 [==============================] - 0s 114us/sample - loss: 0.8216 - categorical_accuracy: 0.7672\n",
      "Epoch 3/1000\n",
      "1536/1536 [==============================] - 0s 117us/sample - loss: 0.7208 - categorical_accuracy: 0.7895\n",
      "Epoch 4/1000\n",
      "1536/1536 [==============================] - 0s 110us/sample - loss: 0.6896 - categorical_accuracy: 0.7958\n",
      "Epoch 5/1000\n",
      "1536/1536 [==============================] - 0s 109us/sample - loss: 0.6733 - categorical_accuracy: 0.8002\n",
      "Epoch 6/1000\n",
      "1536/1536 [==============================] - 0s 110us/sample - loss: 0.6572 - categorical_accuracy: 0.8035\n",
      "Epoch 7/1000\n",
      "1536/1536 [==============================] - 0s 110us/sample - loss: 0.6497 - categorical_accuracy: 0.8060\n",
      "Epoch 8/1000\n",
      "1536/1536 [==============================] - 0s 109us/sample - loss: 0.6404 - categorical_accuracy: 0.8083\n",
      "Epoch 9/1000\n",
      "1536/1536 [==============================] - 0s 110us/sample - loss: 0.6331 - categorical_accuracy: 0.8102\n",
      "Epoch 10/1000\n",
      "1536/1536 [==============================] - 0s 113us/sample - loss: 0.6272 - categorical_accuracy: 0.8128\n",
      "Epoch 11/1000\n",
      "1536/1536 [==============================] - 0s 114us/sample - loss: 0.6182 - categorical_accuracy: 0.8148\n",
      "Epoch 12/1000\n",
      "1536/1536 [==============================] - 0s 114us/sample - loss: 0.6126 - categorical_accuracy: 0.8159\n",
      "Epoch 13/1000\n",
      "1536/1536 [==============================] - 0s 118us/sample - loss: 0.6055 - categorical_accuracy: 0.8174\n",
      "Epoch 14/1000\n",
      "1536/1536 [==============================] - 0s 133us/sample - loss: 0.5994 - categorical_accuracy: 0.8196\n",
      "Epoch 15/1000\n",
      "1536/1536 [==============================] - 0s 112us/sample - loss: 0.5973 - categorical_accuracy: 0.8200\n",
      "Epoch 16/1000\n",
      "1536/1536 [==============================] - 0s 111us/sample - loss: 0.5931 - categorical_accuracy: 0.8203\n",
      "Epoch 17/1000\n",
      "1536/1536 [==============================] - 0s 119us/sample - loss: 0.5875 - categorical_accuracy: 0.8232\n",
      "Epoch 18/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.5832 - categorical_accuracy: 0.8233\n",
      "Epoch 19/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.5794 - categorical_accuracy: 0.8243\n",
      "Epoch 20/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.5749 - categorical_accuracy: 0.8264\n",
      "Epoch 21/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.5724 - categorical_accuracy: 0.8264\n",
      "Epoch 22/1000\n",
      "1536/1536 [==============================] - 0s 117us/sample - loss: 0.5687 - categorical_accuracy: 0.8283\n",
      "Epoch 23/1000\n",
      "1536/1536 [==============================] - 0s 133us/sample - loss: 0.5645 - categorical_accuracy: 0.8287\n",
      "Epoch 24/1000\n",
      "1536/1536 [==============================] - 0s 132us/sample - loss: 0.5606 - categorical_accuracy: 0.8302\n",
      "Epoch 25/1000\n",
      "1536/1536 [==============================] - 0s 132us/sample - loss: 0.5580 - categorical_accuracy: 0.8313\n",
      "Epoch 26/1000\n",
      "1536/1536 [==============================] - 0s 128us/sample - loss: 0.5550 - categorical_accuracy: 0.8317\n",
      "Epoch 27/1000\n",
      "1536/1536 [==============================] - 0s 133us/sample - loss: 0.5630 - categorical_accuracy: 0.8286\n",
      "Epoch 28/1000\n",
      "1536/1536 [==============================] - 0s 120us/sample - loss: 0.5510 - categorical_accuracy: 0.8337\n",
      "Epoch 29/1000\n",
      "1536/1536 [==============================] - 0s 118us/sample - loss: 0.5462 - categorical_accuracy: 0.8347\n",
      "Epoch 30/1000\n",
      "1536/1536 [==============================] - 0s 121us/sample - loss: 0.5429 - categorical_accuracy: 0.8350\n",
      "Epoch 31/1000\n",
      "1536/1536 [==============================] - 0s 136us/sample - loss: 0.5389 - categorical_accuracy: 0.8364\n",
      "Epoch 32/1000\n",
      "1536/1536 [==============================] - 0s 139us/sample - loss: 0.5362 - categorical_accuracy: 0.8372\n",
      "Epoch 33/1000\n",
      "1536/1536 [==============================] - 0s 137us/sample - loss: 0.5337 - categorical_accuracy: 0.8378\n",
      "Epoch 34/1000\n",
      "1536/1536 [==============================] - 0s 138us/sample - loss: 0.5323 - categorical_accuracy: 0.8380\n",
      "Epoch 35/1000\n",
      "1536/1536 [==============================] - 0s 132us/sample - loss: 0.5303 - categorical_accuracy: 0.8385\n",
      "Epoch 36/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.5277 - categorical_accuracy: 0.8392\n",
      "Epoch 37/1000\n",
      "1536/1536 [==============================] - 0s 147us/sample - loss: 0.5240 - categorical_accuracy: 0.8404\n",
      "Epoch 38/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.5205 - categorical_accuracy: 0.8406\n",
      "Epoch 39/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.5198 - categorical_accuracy: 0.8413\n",
      "Epoch 40/1000\n",
      "1536/1536 [==============================] - 0s 133us/sample - loss: 0.5158 - categorical_accuracy: 0.8430\n",
      "Epoch 41/1000\n",
      "1536/1536 [==============================] - 0s 130us/sample - loss: 0.5149 - categorical_accuracy: 0.8426\n",
      "Epoch 42/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.5133 - categorical_accuracy: 0.8444\n",
      "Epoch 43/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.5128 - categorical_accuracy: 0.8443\n",
      "Epoch 44/1000\n",
      "1536/1536 [==============================] - 0s 141us/sample - loss: 0.5093 - categorical_accuracy: 0.8452\n",
      "Epoch 45/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.5096 - categorical_accuracy: 0.8449\n",
      "Epoch 46/1000\n",
      "1536/1536 [==============================] - 0s 132us/sample - loss: 0.6478 - categorical_accuracy: 0.8136\n",
      "Epoch 47/1000\n",
      "1536/1536 [==============================] - 0s 138us/sample - loss: 0.6902 - categorical_accuracy: 0.8002\n",
      "Epoch 48/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.6070 - categorical_accuracy: 0.8175\n",
      "Epoch 49/1000\n",
      "1536/1536 [==============================] - 0s 145us/sample - loss: 0.5786 - categorical_accuracy: 0.8249\n",
      "Epoch 50/1000\n",
      "1536/1536 [==============================] - 0s 143us/sample - loss: 0.5654 - categorical_accuracy: 0.8297\n",
      "Epoch 51/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.5551 - categorical_accuracy: 0.8318\n",
      "Epoch 52/1000\n",
      "1536/1536 [==============================] - 0s 137us/sample - loss: 0.5457 - categorical_accuracy: 0.8348\n",
      "Epoch 53/1000\n",
      "1536/1536 [==============================] - 0s 143us/sample - loss: 0.5404 - categorical_accuracy: 0.8363\n",
      "Epoch 54/1000\n",
      "1536/1536 [==============================] - 0s 128us/sample - loss: 0.5379 - categorical_accuracy: 0.8368\n",
      "Epoch 55/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.5334 - categorical_accuracy: 0.8384\n",
      "Epoch 56/1000\n",
      "1536/1536 [==============================] - 0s 130us/sample - loss: 0.5305 - categorical_accuracy: 0.8387\n",
      "Epoch 57/1000\n",
      "1536/1536 [==============================] - 0s 135us/sample - loss: 0.5263 - categorical_accuracy: 0.8392\n",
      "Epoch 58/1000\n",
      "1536/1536 [==============================] - 0s 133us/sample - loss: 0.5241 - categorical_accuracy: 0.8405\n",
      "Epoch 59/1000\n",
      "1536/1536 [==============================] - 0s 138us/sample - loss: 0.5238 - categorical_accuracy: 0.8398\n",
      "Epoch 60/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.5202 - categorical_accuracy: 0.8424\n",
      "Epoch 61/1000\n",
      "1536/1536 [==============================] - 0s 135us/sample - loss: 0.5164 - categorical_accuracy: 0.8421\n",
      "Epoch 62/1000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1536/1536 [==============================] - 0s 143us/sample - loss: 0.5148 - categorical_accuracy: 0.8437\n",
      "Epoch 63/1000\n",
      "1536/1536 [==============================] - 0s 128us/sample - loss: 0.5115 - categorical_accuracy: 0.8442\n",
      "Epoch 64/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.5103 - categorical_accuracy: 0.8447\n",
      "Epoch 65/1000\n",
      "1536/1536 [==============================] - 0s 136us/sample - loss: 0.5085 - categorical_accuracy: 0.8453\n",
      "Epoch 66/1000\n",
      "1536/1536 [==============================] - 0s 133us/sample - loss: 0.5061 - categorical_accuracy: 0.8457\n",
      "Epoch 67/1000\n",
      "1536/1536 [==============================] - 0s 142us/sample - loss: 0.5045 - categorical_accuracy: 0.8454\n",
      "Epoch 68/1000\n",
      "1536/1536 [==============================] - 0s 149us/sample - loss: 0.5042 - categorical_accuracy: 0.8468\n",
      "Epoch 69/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.5027 - categorical_accuracy: 0.8465\n",
      "Epoch 70/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.5018 - categorical_accuracy: 0.8471\n",
      "Epoch 71/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.5027 - categorical_accuracy: 0.8461\n",
      "Epoch 72/1000\n",
      "1536/1536 [==============================] - 0s 133us/sample - loss: 0.4989 - categorical_accuracy: 0.8475\n",
      "Epoch 73/1000\n",
      "1536/1536 [==============================] - 0s 137us/sample - loss: 0.4987 - categorical_accuracy: 0.8470\n",
      "Epoch 74/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.4968 - categorical_accuracy: 0.8477\n",
      "Epoch 75/1000\n",
      "1536/1536 [==============================] - 0s 145us/sample - loss: 0.4949 - categorical_accuracy: 0.8484\n",
      "Epoch 76/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.4927 - categorical_accuracy: 0.8505\n",
      "Epoch 77/1000\n",
      "1536/1536 [==============================] - 0s 128us/sample - loss: 0.4902 - categorical_accuracy: 0.8497\n",
      "Epoch 78/1000\n",
      "1536/1536 [==============================] - 0s 148us/sample - loss: 0.4909 - categorical_accuracy: 0.8502\n",
      "Epoch 79/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.4885 - categorical_accuracy: 0.8509\n",
      "Epoch 80/1000\n",
      "1536/1536 [==============================] - 0s 149us/sample - loss: 0.4895 - categorical_accuracy: 0.8503\n",
      "Epoch 81/1000\n",
      "1536/1536 [==============================] - 0s 147us/sample - loss: 0.4910 - categorical_accuracy: 0.8502\n",
      "Epoch 82/1000\n",
      "1536/1536 [==============================] - 0s 151us/sample - loss: 0.4846 - categorical_accuracy: 0.8521\n",
      "Epoch 83/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.4841 - categorical_accuracy: 0.8519\n",
      "Epoch 84/1000\n",
      "1536/1536 [==============================] - 0s 150us/sample - loss: 0.4823 - categorical_accuracy: 0.8525\n",
      "Epoch 85/1000\n",
      "1536/1536 [==============================] - 0s 139us/sample - loss: 0.4816 - categorical_accuracy: 0.8534\n",
      "Epoch 86/1000\n",
      "1536/1536 [==============================] - 0s 128us/sample - loss: 0.4795 - categorical_accuracy: 0.8535\n",
      "Epoch 87/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.4800 - categorical_accuracy: 0.8544\n",
      "Epoch 88/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.4809 - categorical_accuracy: 0.8523\n",
      "Epoch 89/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.4789 - categorical_accuracy: 0.8536\n",
      "Epoch 90/1000\n",
      "1536/1536 [==============================] - 0s 130us/sample - loss: 0.4782 - categorical_accuracy: 0.8536\n",
      "Epoch 91/1000\n",
      "1536/1536 [==============================] - 0s 135us/sample - loss: 0.4784 - categorical_accuracy: 0.8536\n",
      "Epoch 92/1000\n",
      "1536/1536 [==============================] - 0s 128us/sample - loss: 0.4764 - categorical_accuracy: 0.8548\n",
      "Epoch 93/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.4761 - categorical_accuracy: 0.8543\n",
      "Epoch 94/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.4741 - categorical_accuracy: 0.8554\n",
      "Epoch 95/1000\n",
      "1536/1536 [==============================] - 0s 133us/sample - loss: 0.4754 - categorical_accuracy: 0.8541\n",
      "Epoch 96/1000\n",
      "1536/1536 [==============================] - 0s 143us/sample - loss: 0.4722 - categorical_accuracy: 0.8560\n",
      "Epoch 97/1000\n",
      "1536/1536 [==============================] - 0s 146us/sample - loss: 0.4714 - categorical_accuracy: 0.8556\n",
      "Epoch 98/1000\n",
      "1536/1536 [==============================] - 0s 128us/sample - loss: 0.4709 - categorical_accuracy: 0.8561\n",
      "Epoch 99/1000\n",
      "1536/1536 [==============================] - 0s 130us/sample - loss: 0.4688 - categorical_accuracy: 0.8566\n",
      "Epoch 100/1000\n",
      "1536/1536 [==============================] - 0s 132us/sample - loss: 0.4696 - categorical_accuracy: 0.8559\n",
      "Epoch 101/1000\n",
      "1536/1536 [==============================] - 0s 136us/sample - loss: 0.4681 - categorical_accuracy: 0.8561\n",
      "Epoch 102/1000\n",
      "1536/1536 [==============================] - 0s 135us/sample - loss: 0.4676 - categorical_accuracy: 0.8573\n",
      "Epoch 103/1000\n",
      "1536/1536 [==============================] - 0s 130us/sample - loss: 0.4655 - categorical_accuracy: 0.8567\n",
      "Epoch 104/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.4659 - categorical_accuracy: 0.8571\n",
      "Epoch 105/1000\n",
      "1536/1536 [==============================] - 0s 133us/sample - loss: 0.4642 - categorical_accuracy: 0.8580\n",
      "Epoch 106/1000\n",
      "1536/1536 [==============================] - 0s 130us/sample - loss: 0.4648 - categorical_accuracy: 0.8573\n",
      "Epoch 107/1000\n",
      "1536/1536 [==============================] - 0s 152us/sample - loss: 0.4633 - categorical_accuracy: 0.8577\n",
      "Epoch 108/1000\n",
      "1536/1536 [==============================] - 0s 167us/sample - loss: 0.4637 - categorical_accuracy: 0.8571\n",
      "Epoch 109/1000\n",
      "1536/1536 [==============================] - 0s 141us/sample - loss: 0.4649 - categorical_accuracy: 0.8569\n",
      "Epoch 110/1000\n",
      "1536/1536 [==============================] - 0s 147us/sample - loss: 0.4641 - categorical_accuracy: 0.8576\n",
      "Epoch 111/1000\n",
      "1536/1536 [==============================] - 0s 132us/sample - loss: 0.4620 - categorical_accuracy: 0.8590\n",
      "Epoch 112/1000\n",
      "1536/1536 [==============================] - 0s 142us/sample - loss: 0.4631 - categorical_accuracy: 0.8571\n",
      "Epoch 113/1000\n",
      "1536/1536 [==============================] - 0s 149us/sample - loss: 0.4611 - categorical_accuracy: 0.8578\n",
      "Epoch 114/1000\n",
      "1536/1536 [==============================] - 0s 137us/sample - loss: 0.4614 - categorical_accuracy: 0.8577\n",
      "Epoch 115/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.4590 - categorical_accuracy: 0.8591\n",
      "Epoch 116/1000\n",
      "1536/1536 [==============================] - 0s 132us/sample - loss: 0.4638 - categorical_accuracy: 0.8570\n",
      "Epoch 117/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.4602 - categorical_accuracy: 0.8588\n",
      "Epoch 118/1000\n",
      "1536/1536 [==============================] - 0s 138us/sample - loss: 0.4569 - categorical_accuracy: 0.8593\n",
      "Epoch 119/1000\n",
      "1536/1536 [==============================] - 0s 133us/sample - loss: 0.4575 - categorical_accuracy: 0.8597\n",
      "Epoch 120/1000\n",
      "1536/1536 [==============================] - 0s 140us/sample - loss: 0.4610 - categorical_accuracy: 0.8581\n",
      "Epoch 121/1000\n",
      "1536/1536 [==============================] - 0s 144us/sample - loss: 0.4569 - categorical_accuracy: 0.8587\n",
      "Epoch 122/1000\n",
      "1536/1536 [==============================] - 0s 146us/sample - loss: 0.4550 - categorical_accuracy: 0.8591\n",
      "Epoch 123/1000\n",
      "1536/1536 [==============================] - 0s 142us/sample - loss: 0.4577 - categorical_accuracy: 0.8599\n",
      "Epoch 124/1000\n",
      "1536/1536 [==============================] - 0s 143us/sample - loss: 0.4565 - categorical_accuracy: 0.8583\n",
      "Epoch 125/1000\n",
      "1536/1536 [==============================] - 0s 146us/sample - loss: 0.4552 - categorical_accuracy: 0.8589\n",
      "Epoch 126/1000\n",
      "1536/1536 [==============================] - 0s 156us/sample - loss: 0.4562 - categorical_accuracy: 0.8600\n",
      "Epoch 127/1000\n",
      "1536/1536 [==============================] - 0s 154us/sample - loss: 0.4571 - categorical_accuracy: 0.8593\n",
      "Epoch 128/1000\n",
      "1536/1536 [==============================] - 0s 147us/sample - loss: 0.4548 - categorical_accuracy: 0.8597\n",
      "Epoch 129/1000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1536/1536 [==============================] - 0s 153us/sample - loss: 0.4573 - categorical_accuracy: 0.8586\n",
      "Epoch 130/1000\n",
      "1536/1536 [==============================] - 0s 141us/sample - loss: 0.4552 - categorical_accuracy: 0.8602\n",
      "Epoch 131/1000\n",
      "1536/1536 [==============================] - 0s 140us/sample - loss: 0.4547 - categorical_accuracy: 0.8597\n",
      "Epoch 132/1000\n",
      "1536/1536 [==============================] - 0s 135us/sample - loss: 0.4540 - categorical_accuracy: 0.8600\n",
      "Epoch 133/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.4528 - categorical_accuracy: 0.8604\n",
      "Epoch 134/1000\n",
      "1536/1536 [==============================] - 0s 132us/sample - loss: 0.4543 - categorical_accuracy: 0.8586\n",
      "Epoch 135/1000\n",
      "1536/1536 [==============================] - 0s 130us/sample - loss: 0.4515 - categorical_accuracy: 0.8604\n",
      "Epoch 136/1000\n",
      "1536/1536 [==============================] - 0s 139us/sample - loss: 0.4540 - categorical_accuracy: 0.8602\n",
      "Epoch 137/1000\n",
      "1536/1536 [==============================] - 0s 149us/sample - loss: 0.4534 - categorical_accuracy: 0.8600\n",
      "Epoch 138/1000\n",
      "1536/1536 [==============================] - 0s 149us/sample - loss: 0.4534 - categorical_accuracy: 0.8595\n",
      "Epoch 139/1000\n",
      "1536/1536 [==============================] - 0s 139us/sample - loss: 0.4512 - categorical_accuracy: 0.8612\n",
      "Epoch 140/1000\n",
      "1536/1536 [==============================] - 0s 138us/sample - loss: 0.4523 - categorical_accuracy: 0.8612\n",
      "Epoch 141/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.4502 - categorical_accuracy: 0.8604\n",
      "Epoch 142/1000\n",
      "1536/1536 [==============================] - 0s 133us/sample - loss: 0.4498 - categorical_accuracy: 0.8604\n",
      "Epoch 143/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.4514 - categorical_accuracy: 0.8612\n",
      "Epoch 144/1000\n",
      "1536/1536 [==============================] - 0s 146us/sample - loss: 0.4513 - categorical_accuracy: 0.8611\n",
      "Epoch 145/1000\n",
      "1536/1536 [==============================] - 0s 143us/sample - loss: 0.4535 - categorical_accuracy: 0.8597\n",
      "Epoch 146/1000\n",
      "1536/1536 [==============================] - 0s 145us/sample - loss: 0.4496 - categorical_accuracy: 0.8609\n",
      "Epoch 147/1000\n",
      "1536/1536 [==============================] - 0s 141us/sample - loss: 0.4484 - categorical_accuracy: 0.8614\n",
      "Epoch 148/1000\n",
      "1536/1536 [==============================] - 0s 137us/sample - loss: 0.4475 - categorical_accuracy: 0.8610\n",
      "Epoch 149/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.4489 - categorical_accuracy: 0.8622\n",
      "Epoch 150/1000\n",
      "1536/1536 [==============================] - 0s 137us/sample - loss: 0.4516 - categorical_accuracy: 0.8597\n",
      "Epoch 151/1000\n",
      "1536/1536 [==============================] - 0s 136us/sample - loss: 0.4470 - categorical_accuracy: 0.8618\n",
      "Epoch 152/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.4485 - categorical_accuracy: 0.8613\n",
      "Epoch 153/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.4530 - categorical_accuracy: 0.8595\n",
      "Epoch 154/1000\n",
      "1536/1536 [==============================] - 0s 132us/sample - loss: 0.4516 - categorical_accuracy: 0.8605\n",
      "Epoch 155/1000\n",
      "1536/1536 [==============================] - 0s 133us/sample - loss: 0.4482 - categorical_accuracy: 0.8618\n",
      "Epoch 156/1000\n",
      "1536/1536 [==============================] - 0s 138us/sample - loss: 0.4487 - categorical_accuracy: 0.8622\n",
      "Epoch 157/1000\n",
      "1536/1536 [==============================] - 0s 133us/sample - loss: 0.4504 - categorical_accuracy: 0.8603\n",
      "Epoch 158/1000\n",
      "1536/1536 [==============================] - 0s 135us/sample - loss: 0.4464 - categorical_accuracy: 0.8629\n",
      "Epoch 159/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.4454 - categorical_accuracy: 0.8631\n",
      "Epoch 160/1000\n",
      "1536/1536 [==============================] - 0s 136us/sample - loss: 0.4445 - categorical_accuracy: 0.8624\n",
      "Epoch 161/1000\n",
      "1536/1536 [==============================] - 0s 138us/sample - loss: 0.4457 - categorical_accuracy: 0.8622\n",
      "Epoch 162/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.4456 - categorical_accuracy: 0.8618\n",
      "Epoch 163/1000\n",
      "1536/1536 [==============================] - 0s 136us/sample - loss: 0.4461 - categorical_accuracy: 0.8624\n",
      "Epoch 164/1000\n",
      "1536/1536 [==============================] - 0s 135us/sample - loss: 0.4421 - categorical_accuracy: 0.8633\n",
      "Epoch 165/1000\n",
      "1536/1536 [==============================] - 0s 139us/sample - loss: 0.4448 - categorical_accuracy: 0.8625\n",
      "Epoch 166/1000\n",
      "1536/1536 [==============================] - 0s 152us/sample - loss: 0.4474 - categorical_accuracy: 0.8613\n",
      "Epoch 167/1000\n",
      "1536/1536 [==============================] - 0s 136us/sample - loss: 0.4487 - categorical_accuracy: 0.8622\n",
      "Epoch 168/1000\n",
      "1536/1536 [==============================] - 0s 140us/sample - loss: 0.4463 - categorical_accuracy: 0.8625\n",
      "Epoch 169/1000\n",
      "1536/1536 [==============================] - 0s 150us/sample - loss: 0.4476 - categorical_accuracy: 0.8611\n",
      "Epoch 170/1000\n",
      "1536/1536 [==============================] - 0s 136us/sample - loss: 0.4481 - categorical_accuracy: 0.8615\n",
      "Epoch 171/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.4471 - categorical_accuracy: 0.8626\n",
      "Epoch 172/1000\n",
      "1536/1536 [==============================] - 0s 138us/sample - loss: 0.4434 - categorical_accuracy: 0.8637\n",
      "Epoch 173/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.4435 - categorical_accuracy: 0.8629\n",
      "Epoch 174/1000\n",
      "1536/1536 [==============================] - 0s 137us/sample - loss: 0.4460 - categorical_accuracy: 0.8625\n",
      "Epoch 175/1000\n",
      "1536/1536 [==============================] - 0s 139us/sample - loss: 0.4454 - categorical_accuracy: 0.8620\n",
      "Epoch 176/1000\n",
      "1536/1536 [==============================] - 0s 138us/sample - loss: 0.4424 - categorical_accuracy: 0.8632\n",
      "Epoch 177/1000\n",
      "1536/1536 [==============================] - 0s 137us/sample - loss: 0.4444 - categorical_accuracy: 0.8627\n",
      "Epoch 178/1000\n",
      "1536/1536 [==============================] - 0s 139us/sample - loss: 0.4416 - categorical_accuracy: 0.8642\n",
      "Epoch 179/1000\n",
      "1536/1536 [==============================] - 0s 137us/sample - loss: 0.4420 - categorical_accuracy: 0.8627\n",
      "Epoch 180/1000\n",
      "1536/1536 [==============================] - 0s 130us/sample - loss: 0.4406 - categorical_accuracy: 0.8645\n",
      "Epoch 181/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.4446 - categorical_accuracy: 0.8622\n",
      "Epoch 182/1000\n",
      "1536/1536 [==============================] - 0s 140us/sample - loss: 0.4461 - categorical_accuracy: 0.8631\n",
      "Epoch 183/1000\n",
      "1536/1536 [==============================] - 0s 138us/sample - loss: 0.4423 - categorical_accuracy: 0.8634\n",
      "Epoch 184/1000\n",
      "1536/1536 [==============================] - 0s 133us/sample - loss: 0.4414 - categorical_accuracy: 0.8637\n",
      "Epoch 185/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.4405 - categorical_accuracy: 0.8647\n",
      "Epoch 186/1000\n",
      "1536/1536 [==============================] - 0s 138us/sample - loss: 0.4405 - categorical_accuracy: 0.8643\n",
      "Epoch 187/1000\n",
      "1536/1536 [==============================] - 0s 133us/sample - loss: 0.4383 - categorical_accuracy: 0.8645\n",
      "Epoch 188/1000\n",
      "1536/1536 [==============================] - 0s 138us/sample - loss: 0.4423 - categorical_accuracy: 0.8632\n",
      "Epoch 189/1000\n",
      "1536/1536 [==============================] - 0s 140us/sample - loss: 0.4464 - categorical_accuracy: 0.8628\n",
      "Epoch 190/1000\n",
      "1536/1536 [==============================] - 0s 142us/sample - loss: 0.4442 - categorical_accuracy: 0.8623\n",
      "Epoch 191/1000\n",
      "1536/1536 [==============================] - 0s 136us/sample - loss: 0.4445 - categorical_accuracy: 0.8618\n",
      "Epoch 192/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.4438 - categorical_accuracy: 0.8627\n",
      "Epoch 193/1000\n",
      "1536/1536 [==============================] - 0s 137us/sample - loss: 0.4405 - categorical_accuracy: 0.8636\n",
      "Epoch 194/1000\n",
      "1536/1536 [==============================] - 0s 140us/sample - loss: 0.4417 - categorical_accuracy: 0.8638\n",
      "Epoch 195/1000\n",
      "1536/1536 [==============================] - 0s 137us/sample - loss: 0.4369 - categorical_accuracy: 0.8650\n",
      "Epoch 196/1000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1536/1536 [==============================] - 0s 137us/sample - loss: 0.4386 - categorical_accuracy: 0.8637\n",
      "Epoch 197/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.4376 - categorical_accuracy: 0.8656\n",
      "Epoch 198/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.4397 - categorical_accuracy: 0.8635\n",
      "Epoch 199/1000\n",
      "1536/1536 [==============================] - 0s 136us/sample - loss: 0.4390 - categorical_accuracy: 0.8636\n",
      "Epoch 200/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.4411 - categorical_accuracy: 0.8640\n",
      "Epoch 201/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.4436 - categorical_accuracy: 0.8623\n",
      "Epoch 202/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.4371 - categorical_accuracy: 0.8650\n",
      "Epoch 203/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.4386 - categorical_accuracy: 0.8644\n",
      "Epoch 204/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.4390 - categorical_accuracy: 0.8643\n",
      "Epoch 205/1000\n",
      "1536/1536 [==============================] - 0s 135us/sample - loss: 0.4380 - categorical_accuracy: 0.8639\n",
      "Epoch 206/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.4369 - categorical_accuracy: 0.8658\n",
      "Epoch 207/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.4378 - categorical_accuracy: 0.8645\n",
      "Epoch 208/1000\n",
      "1536/1536 [==============================] - 0s 135us/sample - loss: 0.4385 - categorical_accuracy: 0.8642\n",
      "Epoch 209/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.4390 - categorical_accuracy: 0.8641\n",
      "Epoch 210/1000\n",
      "1536/1536 [==============================] - 0s 133us/sample - loss: 0.4370 - categorical_accuracy: 0.8654\n",
      "Epoch 211/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.4418 - categorical_accuracy: 0.8628\n",
      "Epoch 212/1000\n",
      "1536/1536 [==============================] - 0s 136us/sample - loss: 0.4401 - categorical_accuracy: 0.8633\n",
      "Epoch 213/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.4359 - categorical_accuracy: 0.8646\n",
      "Epoch 214/1000\n",
      "1536/1536 [==============================] - 0s 173us/sample - loss: 0.4362 - categorical_accuracy: 0.8651\n",
      "Epoch 215/1000\n",
      "1536/1536 [==============================] - 0s 151us/sample - loss: 0.4339 - categorical_accuracy: 0.8653\n",
      "Epoch 216/1000\n",
      "1536/1536 [==============================] - 0s 143us/sample - loss: 0.4346 - categorical_accuracy: 0.8649\n",
      "Epoch 217/1000\n",
      "1536/1536 [==============================] - 0s 133us/sample - loss: 0.4374 - categorical_accuracy: 0.8644\n",
      "Epoch 218/1000\n",
      "1536/1536 [==============================] - 0s 132us/sample - loss: 0.4345 - categorical_accuracy: 0.8655\n",
      "Epoch 219/1000\n",
      "1536/1536 [==============================] - 0s 135us/sample - loss: 0.4352 - categorical_accuracy: 0.8645\n",
      "Epoch 220/1000\n",
      "1536/1536 [==============================] - 0s 130us/sample - loss: 0.4369 - categorical_accuracy: 0.8645\n",
      "Epoch 221/1000\n",
      "1536/1536 [==============================] - 0s 133us/sample - loss: 0.4382 - categorical_accuracy: 0.8637\n",
      "Epoch 222/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.4365 - categorical_accuracy: 0.8643\n",
      "Epoch 223/1000\n",
      "1536/1536 [==============================] - 0s 139us/sample - loss: 0.4394 - categorical_accuracy: 0.8636\n",
      "Epoch 224/1000\n",
      "1536/1536 [==============================] - 0s 136us/sample - loss: 0.4375 - categorical_accuracy: 0.8642\n",
      "Epoch 225/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.4325 - categorical_accuracy: 0.8665\n",
      "Epoch 226/1000\n",
      "1536/1536 [==============================] - 0s 137us/sample - loss: 0.4310 - categorical_accuracy: 0.8656\n",
      "Epoch 227/1000\n",
      "1536/1536 [==============================] - 0s 135us/sample - loss: 0.4361 - categorical_accuracy: 0.8646\n",
      "Epoch 228/1000\n",
      "1536/1536 [==============================] - 0s 132us/sample - loss: 0.4358 - categorical_accuracy: 0.8647\n",
      "Epoch 229/1000\n",
      "1536/1536 [==============================] - 0s 136us/sample - loss: 0.4327 - categorical_accuracy: 0.8658\n",
      "Epoch 230/1000\n",
      "1536/1536 [==============================] - 0s 132us/sample - loss: 0.4347 - categorical_accuracy: 0.8652\n",
      "Epoch 231/1000\n",
      "1536/1536 [==============================] - 0s 137us/sample - loss: 0.4338 - categorical_accuracy: 0.8640\n",
      "Epoch 232/1000\n",
      "1536/1536 [==============================] - 0s 135us/sample - loss: 0.4403 - categorical_accuracy: 0.8636\n",
      "Epoch 233/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.4408 - categorical_accuracy: 0.8636\n",
      "Epoch 234/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.4391 - categorical_accuracy: 0.8636\n",
      "Epoch 235/1000\n",
      "1536/1536 [==============================] - 0s 135us/sample - loss: 0.4407 - categorical_accuracy: 0.8625\n",
      "Epoch 236/1000\n",
      "1536/1536 [==============================] - 0s 136us/sample - loss: 0.4375 - categorical_accuracy: 0.8639\n",
      "Epoch 237/1000\n",
      "1536/1536 [==============================] - 0s 139us/sample - loss: 0.4347 - categorical_accuracy: 0.8645\n",
      "Epoch 238/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.4347 - categorical_accuracy: 0.8643\n",
      "Epoch 239/1000\n",
      "1536/1536 [==============================] - 0s 133us/sample - loss: 0.4355 - categorical_accuracy: 0.8637\n",
      "Epoch 240/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.4325 - categorical_accuracy: 0.8655\n",
      "Epoch 241/1000\n",
      "1536/1536 [==============================] - 0s 140us/sample - loss: 0.4370 - categorical_accuracy: 0.8652\n",
      "Epoch 242/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.4385 - categorical_accuracy: 0.8633\n",
      "Epoch 243/1000\n",
      "1536/1536 [==============================] - 0s 133us/sample - loss: 0.4443 - categorical_accuracy: 0.8624\n",
      "Epoch 244/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.4357 - categorical_accuracy: 0.8640\n",
      "Epoch 245/1000\n",
      "1536/1536 [==============================] - 0s 136us/sample - loss: 0.4307 - categorical_accuracy: 0.8672\n",
      "Epoch 246/1000\n",
      "1536/1536 [==============================] - 0s 135us/sample - loss: 0.4313 - categorical_accuracy: 0.8661\n",
      "Epoch 247/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.4329 - categorical_accuracy: 0.8643\n",
      "Epoch 248/1000\n",
      "1536/1536 [==============================] - 0s 137us/sample - loss: 0.4320 - categorical_accuracy: 0.8663\n",
      "Epoch 249/1000\n",
      "1536/1536 [==============================] - 0s 136us/sample - loss: 0.4363 - categorical_accuracy: 0.8644\n",
      "Epoch 250/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.4392 - categorical_accuracy: 0.8644\n",
      "Epoch 251/1000\n",
      "1536/1536 [==============================] - 0s 138us/sample - loss: 0.4341 - categorical_accuracy: 0.8657\n",
      "Epoch 252/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.4332 - categorical_accuracy: 0.8653\n",
      "Epoch 253/1000\n",
      "1536/1536 [==============================] - 0s 135us/sample - loss: 0.4329 - categorical_accuracy: 0.8654\n",
      "Epoch 254/1000\n",
      "1536/1536 [==============================] - 0s 138us/sample - loss: 0.4343 - categorical_accuracy: 0.8660\n",
      "Epoch 255/1000\n",
      "1536/1536 [==============================] - 0s 132us/sample - loss: 0.4367 - categorical_accuracy: 0.8644\n",
      "Epoch 256/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.4325 - categorical_accuracy: 0.8656\n",
      "Epoch 257/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.4303 - categorical_accuracy: 0.8679\n",
      "Epoch 258/1000\n",
      "1536/1536 [==============================] - 0s 135us/sample - loss: 0.4426 - categorical_accuracy: 0.8616\n",
      "Epoch 259/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.4403 - categorical_accuracy: 0.8641\n",
      "Epoch 260/1000\n",
      "1536/1536 [==============================] - 0s 137us/sample - loss: 0.4381 - categorical_accuracy: 0.8648\n",
      "Epoch 261/1000\n",
      "1536/1536 [==============================] - 0s 137us/sample - loss: 0.4348 - categorical_accuracy: 0.8653\n",
      "Epoch 262/1000\n",
      "1536/1536 [==============================] - 0s 136us/sample - loss: 0.4333 - categorical_accuracy: 0.8648\n",
      "Epoch 263/1000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1536/1536 [==============================] - 0s 136us/sample - loss: 0.4343 - categorical_accuracy: 0.8649\n",
      "Epoch 264/1000\n",
      "1536/1536 [==============================] - 0s 128us/sample - loss: 0.4301 - categorical_accuracy: 0.8664\n",
      "Epoch 265/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.4293 - categorical_accuracy: 0.8671\n",
      "Epoch 266/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.4389 - categorical_accuracy: 0.8647\n",
      "Epoch 267/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.4338 - categorical_accuracy: 0.8646\n",
      "Epoch 268/1000\n",
      "1536/1536 [==============================] - 0s 130us/sample - loss: 0.4303 - categorical_accuracy: 0.8667\n",
      "Epoch 269/1000\n",
      "1536/1536 [==============================] - 0s 130us/sample - loss: 0.4295 - categorical_accuracy: 0.8663\n",
      "Epoch 270/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.4325 - categorical_accuracy: 0.8652\n",
      "Epoch 271/1000\n",
      "1536/1536 [==============================] - 0s 132us/sample - loss: 0.4309 - categorical_accuracy: 0.8656\n",
      "Epoch 272/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.4317 - categorical_accuracy: 0.8659\n",
      "Epoch 273/1000\n",
      "1536/1536 [==============================] - 0s 136us/sample - loss: 0.4341 - categorical_accuracy: 0.8656\n",
      "Epoch 274/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.4310 - categorical_accuracy: 0.8672\n",
      "Epoch 275/1000\n",
      "1536/1536 [==============================] - 0s 133us/sample - loss: 0.4330 - categorical_accuracy: 0.8659\n",
      "Epoch 276/1000\n",
      "1536/1536 [==============================] - 0s 136us/sample - loss: 0.4331 - categorical_accuracy: 0.8656\n",
      "Epoch 277/1000\n",
      "1536/1536 [==============================] - 0s 133us/sample - loss: 0.4326 - categorical_accuracy: 0.8658\n",
      "Epoch 278/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.4323 - categorical_accuracy: 0.8656\n",
      "Epoch 279/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.4330 - categorical_accuracy: 0.8664\n",
      "Epoch 280/1000\n",
      "1536/1536 [==============================] - 0s 128us/sample - loss: 0.4336 - categorical_accuracy: 0.8658\n",
      "Epoch 281/1000\n",
      "1536/1536 [==============================] - 0s 133us/sample - loss: 0.4379 - categorical_accuracy: 0.8639\n",
      "Epoch 282/1000\n",
      "1536/1536 [==============================] - 0s 132us/sample - loss: 0.4335 - categorical_accuracy: 0.8648\n",
      "Epoch 283/1000\n",
      "1536/1536 [==============================] - 0s 132us/sample - loss: 0.4277 - categorical_accuracy: 0.8686\n",
      "Epoch 284/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.4290 - categorical_accuracy: 0.8667\n",
      "Epoch 285/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.4342 - categorical_accuracy: 0.8653\n",
      "Epoch 286/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.4382 - categorical_accuracy: 0.8642\n",
      "Epoch 287/1000\n",
      "1536/1536 [==============================] - 0s 137us/sample - loss: 0.4358 - categorical_accuracy: 0.8644\n",
      "Epoch 288/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.4362 - categorical_accuracy: 0.8645\n",
      "Epoch 289/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.4374 - categorical_accuracy: 0.8632\n",
      "Epoch 290/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.4382 - categorical_accuracy: 0.8637\n",
      "Epoch 291/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.4351 - categorical_accuracy: 0.8639\n",
      "Epoch 292/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.4375 - categorical_accuracy: 0.8644\n",
      "Epoch 293/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.4339 - categorical_accuracy: 0.8652\n",
      "Epoch 294/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.4315 - categorical_accuracy: 0.8661\n",
      "Epoch 295/1000\n",
      "1536/1536 [==============================] - 0s 133us/sample - loss: 0.4375 - categorical_accuracy: 0.8643\n",
      "Epoch 296/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.4307 - categorical_accuracy: 0.8666\n",
      "Epoch 297/1000\n",
      "1536/1536 [==============================] - 0s 138us/sample - loss: 0.4320 - categorical_accuracy: 0.8652\n",
      "Epoch 298/1000\n",
      "1536/1536 [==============================] - 0s 138us/sample - loss: 0.4311 - categorical_accuracy: 0.8669\n",
      "Epoch 299/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.4284 - categorical_accuracy: 0.8671\n",
      "Epoch 300/1000\n",
      "1536/1536 [==============================] - 0s 136us/sample - loss: 0.4277 - categorical_accuracy: 0.8670\n",
      "Epoch 301/1000\n",
      "1536/1536 [==============================] - 0s 132us/sample - loss: 0.4310 - categorical_accuracy: 0.8654\n",
      "Epoch 302/1000\n",
      "1536/1536 [==============================] - 0s 135us/sample - loss: 0.4340 - categorical_accuracy: 0.8652\n",
      "Epoch 303/1000\n",
      "1536/1536 [==============================] - 0s 145us/sample - loss: 0.4344 - categorical_accuracy: 0.8651\n",
      "Epoch 304/1000\n",
      "1536/1536 [==============================] - 0s 136us/sample - loss: 0.4376 - categorical_accuracy: 0.8628\n",
      "Epoch 305/1000\n",
      "1536/1536 [==============================] - 0s 135us/sample - loss: 0.4335 - categorical_accuracy: 0.8652\n",
      "Epoch 306/1000\n",
      "1536/1536 [==============================] - 0s 133us/sample - loss: 0.4416 - categorical_accuracy: 0.8627\n",
      "Epoch 307/1000\n",
      "1536/1536 [==============================] - 0s 145us/sample - loss: 0.4326 - categorical_accuracy: 0.8662\n",
      "Epoch 308/1000\n",
      "1536/1536 [==============================] - 0s 143us/sample - loss: 0.4310 - categorical_accuracy: 0.8669\n",
      "Epoch 309/1000\n",
      "1536/1536 [==============================] - 0s 151us/sample - loss: 0.4300 - categorical_accuracy: 0.8666\n",
      "Epoch 310/1000\n",
      "1536/1536 [==============================] - 0s 135us/sample - loss: 0.4404 - categorical_accuracy: 0.8628\n",
      "Epoch 311/1000\n",
      "1536/1536 [==============================] - 0s 141us/sample - loss: 0.4356 - categorical_accuracy: 0.8644\n",
      "Epoch 312/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.4337 - categorical_accuracy: 0.8650\n",
      "Epoch 313/1000\n",
      "1536/1536 [==============================] - 0s 135us/sample - loss: 0.4314 - categorical_accuracy: 0.8651\n",
      "Epoch 314/1000\n",
      "1536/1536 [==============================] - 0s 137us/sample - loss: 0.4311 - categorical_accuracy: 0.8658\n",
      "Epoch 315/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.4279 - categorical_accuracy: 0.8669\n",
      "Epoch 316/1000\n",
      "1536/1536 [==============================] - 0s 135us/sample - loss: 0.4303 - categorical_accuracy: 0.8666\n",
      "Epoch 317/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.4304 - categorical_accuracy: 0.8667\n",
      "Epoch 318/1000\n",
      "1536/1536 [==============================] - 0s 137us/sample - loss: 0.4285 - categorical_accuracy: 0.8676\n",
      "Epoch 319/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.4263 - categorical_accuracy: 0.8679\n",
      "Epoch 320/1000\n",
      "1536/1536 [==============================] - 0s 136us/sample - loss: 0.4277 - categorical_accuracy: 0.8683\n",
      "Epoch 321/1000\n",
      "1536/1536 [==============================] - 0s 132us/sample - loss: 0.4315 - categorical_accuracy: 0.8662\n",
      "Epoch 322/1000\n",
      "1536/1536 [==============================] - 0s 136us/sample - loss: 0.4372 - categorical_accuracy: 0.8640\n",
      "Epoch 323/1000\n",
      "1536/1536 [==============================] - 0s 137us/sample - loss: 0.4327 - categorical_accuracy: 0.8654\n",
      "Epoch 324/1000\n",
      "1536/1536 [==============================] - 0s 150us/sample - loss: 0.4327 - categorical_accuracy: 0.8653\n",
      "Epoch 325/1000\n",
      "1536/1536 [==============================] - 0s 133us/sample - loss: 0.4318 - categorical_accuracy: 0.8659\n",
      "Epoch 326/1000\n",
      "1536/1536 [==============================] - 0s 132us/sample - loss: 0.4290 - categorical_accuracy: 0.8664\n",
      "Epoch 327/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.4286 - categorical_accuracy: 0.8676\n",
      "Epoch 328/1000\n",
      "1536/1536 [==============================] - 0s 133us/sample - loss: 0.4330 - categorical_accuracy: 0.8647\n",
      "Epoch 329/1000\n",
      "1536/1536 [==============================] - 0s 156us/sample - loss: 0.4297 - categorical_accuracy: 0.8666\n",
      "Epoch 330/1000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1536/1536 [==============================] - 0s 139us/sample - loss: 0.4298 - categorical_accuracy: 0.8662\n",
      "Epoch 331/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.4342 - categorical_accuracy: 0.8643\n",
      "Epoch 332/1000\n",
      "1536/1536 [==============================] - 0s 132us/sample - loss: 0.4354 - categorical_accuracy: 0.8641\n",
      "Epoch 333/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.4419 - categorical_accuracy: 0.8629\n",
      "Epoch 334/1000\n",
      "1536/1536 [==============================] - 0s 136us/sample - loss: 0.4418 - categorical_accuracy: 0.8639\n",
      "Epoch 335/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.4366 - categorical_accuracy: 0.8643\n",
      "Epoch 336/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.4368 - categorical_accuracy: 0.8633\n",
      "Epoch 337/1000\n",
      "1536/1536 [==============================] - 0s 133us/sample - loss: 0.4326 - categorical_accuracy: 0.8659\n",
      "Epoch 338/1000\n",
      "1536/1536 [==============================] - 0s 132us/sample - loss: 0.4281 - categorical_accuracy: 0.8664\n",
      "Epoch 339/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.4346 - categorical_accuracy: 0.8637\n",
      "Epoch 340/1000\n",
      "1536/1536 [==============================] - 0s 137us/sample - loss: 0.4365 - categorical_accuracy: 0.8644\n",
      "Epoch 341/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.4387 - categorical_accuracy: 0.8637\n",
      "Epoch 342/1000\n",
      "1536/1536 [==============================] - 0s 132us/sample - loss: 0.4327 - categorical_accuracy: 0.8655\n",
      "Epoch 343/1000\n",
      "1536/1536 [==============================] - 0s 132us/sample - loss: 0.4297 - categorical_accuracy: 0.8661\n",
      "Epoch 344/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.4312 - categorical_accuracy: 0.8666\n",
      "Epoch 345/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.4237 - categorical_accuracy: 0.8685\n",
      "Epoch 346/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.4302 - categorical_accuracy: 0.8675\n",
      "Epoch 347/1000\n",
      "1536/1536 [==============================] - 0s 138us/sample - loss: 0.4312 - categorical_accuracy: 0.8658\n",
      "Epoch 348/1000\n",
      "1536/1536 [==============================] - 0s 128us/sample - loss: 0.4296 - categorical_accuracy: 0.8658\n",
      "Epoch 349/1000\n",
      "1536/1536 [==============================] - 0s 133us/sample - loss: 0.4288 - categorical_accuracy: 0.8663\n",
      "Epoch 350/1000\n",
      "1536/1536 [==============================] - 0s 130us/sample - loss: 0.4269 - categorical_accuracy: 0.8671\n",
      "Epoch 351/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.4353 - categorical_accuracy: 0.8642\n",
      "Epoch 352/1000\n",
      "1536/1536 [==============================] - 0s 133us/sample - loss: 0.4315 - categorical_accuracy: 0.8659\n",
      "Epoch 353/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.4239 - categorical_accuracy: 0.8681\n",
      "Epoch 354/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.4254 - categorical_accuracy: 0.8674\n",
      "Epoch 355/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.4247 - categorical_accuracy: 0.8678\n",
      "Epoch 356/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.4268 - categorical_accuracy: 0.8674\n",
      "Epoch 357/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.4306 - categorical_accuracy: 0.8664\n",
      "Epoch 358/1000\n",
      "1536/1536 [==============================] - 0s 133us/sample - loss: 0.4349 - categorical_accuracy: 0.8646\n",
      "Epoch 359/1000\n",
      "1536/1536 [==============================] - 0s 132us/sample - loss: 0.4333 - categorical_accuracy: 0.8646\n",
      "Epoch 360/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.4295 - categorical_accuracy: 0.8667\n",
      "Epoch 361/1000\n",
      "1536/1536 [==============================] - 0s 132us/sample - loss: 0.4331 - categorical_accuracy: 0.8654\n",
      "Epoch 362/1000\n",
      "1536/1536 [==============================] - 0s 128us/sample - loss: 0.4300 - categorical_accuracy: 0.8649\n",
      "Epoch 363/1000\n",
      "1536/1536 [==============================] - 0s 136us/sample - loss: 0.4310 - categorical_accuracy: 0.8655\n",
      "Epoch 364/1000\n",
      "1536/1536 [==============================] - 0s 135us/sample - loss: 0.4297 - categorical_accuracy: 0.8666\n",
      "Epoch 365/1000\n",
      "1536/1536 [==============================] - 0s 133us/sample - loss: 0.4300 - categorical_accuracy: 0.8659\n",
      "Epoch 366/1000\n",
      "1536/1536 [==============================] - 0s 128us/sample - loss: 0.4289 - categorical_accuracy: 0.8664\n",
      "Epoch 367/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.4290 - categorical_accuracy: 0.8666\n",
      "Epoch 368/1000\n",
      "1536/1536 [==============================] - 0s 136us/sample - loss: 0.4280 - categorical_accuracy: 0.8667\n",
      "Epoch 369/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.4296 - categorical_accuracy: 0.8665\n",
      "Epoch 370/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.4312 - categorical_accuracy: 0.8669\n",
      "Epoch 371/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.4312 - categorical_accuracy: 0.8663\n",
      "Epoch 372/1000\n",
      "1536/1536 [==============================] - 0s 128us/sample - loss: 0.4348 - categorical_accuracy: 0.8653\n",
      "Epoch 373/1000\n",
      "1536/1536 [==============================] - 0s 136us/sample - loss: 0.4353 - categorical_accuracy: 0.8652\n",
      "Epoch 374/1000\n",
      "1536/1536 [==============================] - 0s 135us/sample - loss: 0.4373 - categorical_accuracy: 0.8637\n",
      "Epoch 375/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.4432 - categorical_accuracy: 0.8619\n",
      "Epoch 376/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.4375 - categorical_accuracy: 0.8652\n",
      "Epoch 377/1000\n",
      "1536/1536 [==============================] - 0s 128us/sample - loss: 0.4342 - categorical_accuracy: 0.8654\n",
      "Epoch 378/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.4346 - categorical_accuracy: 0.8650\n",
      "Epoch 379/1000\n",
      "1536/1536 [==============================] - 0s 130us/sample - loss: 0.4303 - categorical_accuracy: 0.8669\n",
      "Epoch 380/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.4288 - categorical_accuracy: 0.8659\n",
      "Epoch 381/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.4316 - categorical_accuracy: 0.8665\n",
      "Epoch 382/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.4356 - categorical_accuracy: 0.8653\n",
      "Epoch 383/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.4287 - categorical_accuracy: 0.8668\n",
      "Epoch 384/1000\n",
      "1536/1536 [==============================] - 0s 130us/sample - loss: 0.4264 - categorical_accuracy: 0.8673\n",
      "Epoch 385/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.4292 - categorical_accuracy: 0.8661\n",
      "Epoch 386/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.4255 - categorical_accuracy: 0.8676\n",
      "Epoch 387/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.4247 - categorical_accuracy: 0.8681\n",
      "Epoch 388/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.4259 - categorical_accuracy: 0.8688\n",
      "Epoch 389/1000\n",
      "1536/1536 [==============================] - 0s 130us/sample - loss: 0.4285 - categorical_accuracy: 0.8666\n",
      "Epoch 390/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.4244 - categorical_accuracy: 0.8679\n",
      "Epoch 391/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.4304 - categorical_accuracy: 0.8658\n",
      "Epoch 392/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.4325 - categorical_accuracy: 0.8659\n",
      "Epoch 393/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.4285 - categorical_accuracy: 0.8677\n",
      "Epoch 394/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.4303 - categorical_accuracy: 0.8663\n",
      "Epoch 395/1000\n",
      "1536/1536 [==============================] - 0s 137us/sample - loss: 0.4291 - categorical_accuracy: 0.8661\n",
      "Epoch 396/1000\n",
      "1536/1536 [==============================] - 0s 128us/sample - loss: 0.4509 - categorical_accuracy: 0.8610\n",
      "Epoch 397/1000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.4472 - categorical_accuracy: 0.8615\n",
      "Epoch 398/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.4387 - categorical_accuracy: 0.8639\n",
      "Epoch 399/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.4358 - categorical_accuracy: 0.8652\n",
      "Epoch 400/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.4279 - categorical_accuracy: 0.8674\n",
      "Epoch 401/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.4398 - categorical_accuracy: 0.8642\n",
      "Epoch 402/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.4375 - categorical_accuracy: 0.8644\n",
      "Epoch 403/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.4322 - categorical_accuracy: 0.8654\n",
      "Epoch 404/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.4274 - categorical_accuracy: 0.8668\n",
      "Epoch 405/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.4270 - categorical_accuracy: 0.8674\n",
      "Epoch 406/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.4277 - categorical_accuracy: 0.8671\n",
      "Epoch 407/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.4243 - categorical_accuracy: 0.8679\n",
      "Epoch 408/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.4286 - categorical_accuracy: 0.8667\n",
      "Epoch 409/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.4367 - categorical_accuracy: 0.8646\n",
      "Epoch 410/1000\n",
      "1536/1536 [==============================] - 0s 128us/sample - loss: 0.4382 - categorical_accuracy: 0.8645\n",
      "Epoch 411/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.4409 - categorical_accuracy: 0.8630\n",
      "Epoch 412/1000\n",
      "1536/1536 [==============================] - 0s 130us/sample - loss: 0.4510 - categorical_accuracy: 0.8599\n",
      "Epoch 413/1000\n",
      "1536/1536 [==============================] - 0s 128us/sample - loss: 0.4420 - categorical_accuracy: 0.8633\n",
      "Epoch 414/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.4347 - categorical_accuracy: 0.8645\n",
      "Epoch 415/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.4280 - categorical_accuracy: 0.8678\n",
      "Epoch 416/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.4251 - categorical_accuracy: 0.8682\n",
      "Epoch 417/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.4243 - categorical_accuracy: 0.8676\n",
      "Epoch 418/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.4266 - categorical_accuracy: 0.8673\n",
      "Epoch 419/1000\n",
      "1536/1536 [==============================] - 0s 128us/sample - loss: 0.4290 - categorical_accuracy: 0.8664\n",
      "Epoch 420/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.4319 - categorical_accuracy: 0.8656\n",
      "Epoch 421/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.4272 - categorical_accuracy: 0.8673\n",
      "Epoch 422/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.4247 - categorical_accuracy: 0.8687\n",
      "Epoch 423/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.4237 - categorical_accuracy: 0.8681\n",
      "Epoch 424/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.4248 - categorical_accuracy: 0.8672\n",
      "Epoch 425/1000\n",
      "1536/1536 [==============================] - 0s 132us/sample - loss: 0.4267 - categorical_accuracy: 0.8680\n",
      "Epoch 426/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.4264 - categorical_accuracy: 0.8673\n",
      "Epoch 427/1000\n",
      "1536/1536 [==============================] - 0s 128us/sample - loss: 0.4287 - categorical_accuracy: 0.8665\n",
      "Epoch 428/1000\n",
      "1536/1536 [==============================] - 0s 130us/sample - loss: 0.4265 - categorical_accuracy: 0.8675\n",
      "Epoch 429/1000\n",
      "1536/1536 [==============================] - 0s 128us/sample - loss: 0.4300 - categorical_accuracy: 0.8665\n",
      "Epoch 430/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.4348 - categorical_accuracy: 0.8648\n",
      "Epoch 431/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.4301 - categorical_accuracy: 0.8665\n",
      "Epoch 432/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.4324 - categorical_accuracy: 0.8661\n",
      "Epoch 433/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.4412 - categorical_accuracy: 0.8640\n",
      "Epoch 434/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.4342 - categorical_accuracy: 0.8655\n",
      "Epoch 435/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.4264 - categorical_accuracy: 0.8670\n",
      "Epoch 436/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.4261 - categorical_accuracy: 0.8677\n",
      "Epoch 437/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.4208 - categorical_accuracy: 0.8687\n",
      "Epoch 438/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.4222 - categorical_accuracy: 0.8686\n",
      "Epoch 439/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.4224 - categorical_accuracy: 0.8690\n",
      "Epoch 440/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.4324 - categorical_accuracy: 0.8664\n",
      "Epoch 441/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.4326 - categorical_accuracy: 0.8651\n",
      "Epoch 442/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.4341 - categorical_accuracy: 0.8641\n",
      "Epoch 443/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.4339 - categorical_accuracy: 0.8647\n",
      "Epoch 444/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.4335 - categorical_accuracy: 0.8647\n",
      "Epoch 445/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.4303 - categorical_accuracy: 0.8651\n",
      "Epoch 446/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.4300 - categorical_accuracy: 0.8665\n",
      "Epoch 447/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.4277 - categorical_accuracy: 0.8677\n",
      "Epoch 448/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.4278 - categorical_accuracy: 0.8663\n",
      "Epoch 449/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.4267 - categorical_accuracy: 0.8669\n",
      "Epoch 450/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.4291 - categorical_accuracy: 0.8664\n",
      "Epoch 451/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.4370 - categorical_accuracy: 0.8647\n",
      "Epoch 452/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.4390 - categorical_accuracy: 0.8649\n",
      "Epoch 453/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.4345 - categorical_accuracy: 0.8651\n",
      "Epoch 454/1000\n",
      "1536/1536 [==============================] - 0s 132us/sample - loss: 0.4243 - categorical_accuracy: 0.8683\n",
      "Epoch 455/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.4318 - categorical_accuracy: 0.8654\n",
      "Epoch 456/1000\n",
      "1536/1536 [==============================] - 0s 128us/sample - loss: 0.4294 - categorical_accuracy: 0.8677\n",
      "Epoch 457/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.4266 - categorical_accuracy: 0.8682\n",
      "Epoch 458/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.4280 - categorical_accuracy: 0.8665\n",
      "Epoch 459/1000\n",
      "1536/1536 [==============================] - 0s 128us/sample - loss: 0.4266 - categorical_accuracy: 0.8678\n",
      "Epoch 460/1000\n",
      "1536/1536 [==============================] - 0s 133us/sample - loss: 0.4273 - categorical_accuracy: 0.8671\n",
      "Epoch 461/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.4286 - categorical_accuracy: 0.8664\n",
      "Epoch 462/1000\n",
      "1536/1536 [==============================] - 0s 120us/sample - loss: 0.4349 - categorical_accuracy: 0.8649\n",
      "Epoch 463/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.4502 - categorical_accuracy: 0.8602\n",
      "Epoch 464/1000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.4468 - categorical_accuracy: 0.8622\n",
      "Epoch 465/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.4429 - categorical_accuracy: 0.8630\n",
      "Epoch 466/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.4343 - categorical_accuracy: 0.8642\n",
      "Epoch 467/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.4292 - categorical_accuracy: 0.8664\n",
      "Epoch 468/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.4285 - categorical_accuracy: 0.8666\n",
      "Epoch 469/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.4258 - categorical_accuracy: 0.8681\n",
      "Epoch 470/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.4259 - categorical_accuracy: 0.8674\n",
      "Epoch 471/1000\n",
      "1536/1536 [==============================] - 0s 130us/sample - loss: 0.4250 - categorical_accuracy: 0.8673\n",
      "Epoch 472/1000\n",
      "1536/1536 [==============================] - 0s 128us/sample - loss: 0.4282 - categorical_accuracy: 0.8682\n",
      "Epoch 473/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.4292 - categorical_accuracy: 0.8667\n",
      "Epoch 474/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.4308 - categorical_accuracy: 0.8665\n",
      "Epoch 475/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.4262 - categorical_accuracy: 0.8676\n",
      "Epoch 476/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.4276 - categorical_accuracy: 0.8673\n",
      "Epoch 477/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.4351 - categorical_accuracy: 0.8655\n",
      "Epoch 478/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.4392 - categorical_accuracy: 0.8629\n",
      "Epoch 479/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.4358 - categorical_accuracy: 0.8646\n",
      "Epoch 480/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.4378 - categorical_accuracy: 0.8628\n",
      "Epoch 481/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.4379 - categorical_accuracy: 0.8646\n",
      "Epoch 482/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.4362 - categorical_accuracy: 0.8639\n",
      "Epoch 483/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.4340 - categorical_accuracy: 0.8657\n",
      "Epoch 484/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.4314 - categorical_accuracy: 0.8661\n",
      "Epoch 485/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.4281 - categorical_accuracy: 0.8664\n",
      "Epoch 486/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.4271 - categorical_accuracy: 0.8665\n",
      "Epoch 487/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.4277 - categorical_accuracy: 0.8657\n",
      "Epoch 488/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.4268 - categorical_accuracy: 0.8666\n",
      "Epoch 489/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.4354 - categorical_accuracy: 0.8652\n",
      "Epoch 490/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.4283 - categorical_accuracy: 0.8674\n",
      "Epoch 491/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.4262 - categorical_accuracy: 0.8671\n",
      "Epoch 492/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.4284 - categorical_accuracy: 0.8681\n",
      "Epoch 493/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.4275 - categorical_accuracy: 0.8677\n",
      "Epoch 494/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.4325 - categorical_accuracy: 0.8665\n",
      "Epoch 495/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.4313 - categorical_accuracy: 0.8661\n",
      "Epoch 496/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.4287 - categorical_accuracy: 0.8670\n",
      "Epoch 497/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.4292 - categorical_accuracy: 0.8678\n",
      "Epoch 498/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.4322 - categorical_accuracy: 0.8655\n",
      "Epoch 499/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.4368 - categorical_accuracy: 0.8642\n",
      "Epoch 500/1000\n",
      "1536/1536 [==============================] - 0s 146us/sample - loss: 0.4353 - categorical_accuracy: 0.8652\n",
      "Epoch 501/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.4312 - categorical_accuracy: 0.8663\n",
      "Epoch 502/1000\n",
      "1536/1536 [==============================] - 0s 130us/sample - loss: 0.4358 - categorical_accuracy: 0.8656\n",
      "Epoch 503/1000\n",
      "1536/1536 [==============================] - 0s 132us/sample - loss: 0.4402 - categorical_accuracy: 0.8625\n",
      "Epoch 504/1000\n",
      "1536/1536 [==============================] - 0s 130us/sample - loss: 0.4392 - categorical_accuracy: 0.8630\n",
      "Epoch 505/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.4321 - categorical_accuracy: 0.8653\n",
      "Epoch 506/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.4366 - categorical_accuracy: 0.8646\n",
      "Epoch 507/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.4381 - categorical_accuracy: 0.8649\n",
      "Epoch 508/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.4391 - categorical_accuracy: 0.8634\n",
      "Epoch 509/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.4332 - categorical_accuracy: 0.8654\n",
      "Epoch 510/1000\n",
      "1536/1536 [==============================] - 0s 139us/sample - loss: 0.4279 - categorical_accuracy: 0.8656\n",
      "Epoch 511/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.4310 - categorical_accuracy: 0.8651\n",
      "Epoch 512/1000\n",
      "1536/1536 [==============================] - 0s 132us/sample - loss: 0.4361 - categorical_accuracy: 0.8654\n",
      "Epoch 513/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.4515 - categorical_accuracy: 0.8611\n",
      "Epoch 514/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.4410 - categorical_accuracy: 0.8628\n",
      "Epoch 515/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.4296 - categorical_accuracy: 0.8666\n",
      "Epoch 516/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.4231 - categorical_accuracy: 0.8685\n",
      "Epoch 517/1000\n",
      "1536/1536 [==============================] - 0s 138us/sample - loss: 0.4240 - categorical_accuracy: 0.8674\n",
      "Epoch 518/1000\n",
      "1536/1536 [==============================] - 0s 130us/sample - loss: 0.4253 - categorical_accuracy: 0.8674\n",
      "Epoch 519/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.4272 - categorical_accuracy: 0.8675\n",
      "Epoch 520/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.4259 - categorical_accuracy: 0.8671\n",
      "Epoch 521/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.4238 - categorical_accuracy: 0.8674\n",
      "Epoch 522/1000\n",
      "1536/1536 [==============================] - 0s 137us/sample - loss: 0.4244 - categorical_accuracy: 0.8676\n",
      "Epoch 523/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.4256 - categorical_accuracy: 0.8676\n",
      "Epoch 524/1000\n",
      "1536/1536 [==============================] - 0s 130us/sample - loss: 0.4329 - categorical_accuracy: 0.8650\n",
      "Epoch 525/1000\n",
      "1536/1536 [==============================] - 0s 130us/sample - loss: 0.4481 - categorical_accuracy: 0.8608\n",
      "Epoch 526/1000\n",
      "1536/1536 [==============================] - 0s 132us/sample - loss: 0.4347 - categorical_accuracy: 0.8649\n",
      "Epoch 527/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.4325 - categorical_accuracy: 0.8649\n",
      "Epoch 528/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.4338 - categorical_accuracy: 0.8651\n",
      "Epoch 529/1000\n",
      "1536/1536 [==============================] - 0s 136us/sample - loss: 0.4347 - categorical_accuracy: 0.8648\n",
      "Epoch 530/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.4315 - categorical_accuracy: 0.8662\n",
      "Epoch 531/1000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1536/1536 [==============================] - 0s 138us/sample - loss: 0.4256 - categorical_accuracy: 0.8669\n",
      "Epoch 532/1000\n",
      "1536/1536 [==============================] - 0s 130us/sample - loss: 0.4265 - categorical_accuracy: 0.8678\n",
      "Epoch 533/1000\n",
      "1536/1536 [==============================] - 0s 132us/sample - loss: 0.4275 - categorical_accuracy: 0.8680\n",
      "Epoch 534/1000\n",
      "1536/1536 [==============================] - 0s 133us/sample - loss: 0.4320 - categorical_accuracy: 0.8666\n",
      "Epoch 535/1000\n",
      "1536/1536 [==============================] - 0s 130us/sample - loss: 0.4369 - categorical_accuracy: 0.8646\n",
      "Epoch 536/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.4410 - categorical_accuracy: 0.8626\n",
      "Epoch 537/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.4327 - categorical_accuracy: 0.8655\n",
      "Epoch 538/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.4337 - categorical_accuracy: 0.8647\n",
      "Epoch 539/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.4278 - categorical_accuracy: 0.8671\n",
      "Epoch 540/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.4342 - categorical_accuracy: 0.8648\n",
      "Epoch 541/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.4316 - categorical_accuracy: 0.8649\n",
      "Epoch 542/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.4332 - categorical_accuracy: 0.8652\n",
      "Epoch 543/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.4319 - categorical_accuracy: 0.8655\n",
      "Epoch 544/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.4262 - categorical_accuracy: 0.8670\n",
      "Epoch 545/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.4301 - categorical_accuracy: 0.8656\n",
      "Epoch 546/1000\n",
      "1536/1536 [==============================] - 0s 120us/sample - loss: 0.4337 - categorical_accuracy: 0.8644\n",
      "Epoch 547/1000\n",
      "1536/1536 [==============================] - 0s 121us/sample - loss: 0.4298 - categorical_accuracy: 0.8667\n",
      "Epoch 548/1000\n",
      "1536/1536 [==============================] - 0s 121us/sample - loss: 0.4284 - categorical_accuracy: 0.8671\n",
      "Epoch 549/1000\n",
      "1536/1536 [==============================] - 0s 121us/sample - loss: 0.4259 - categorical_accuracy: 0.8675\n",
      "Epoch 550/1000\n",
      "1536/1536 [==============================] - 0s 120us/sample - loss: 0.4247 - categorical_accuracy: 0.8686\n",
      "Epoch 551/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.4281 - categorical_accuracy: 0.8670\n",
      "Epoch 552/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.4306 - categorical_accuracy: 0.8659\n",
      "Epoch 553/1000\n",
      "1536/1536 [==============================] - 0s 128us/sample - loss: 0.4308 - categorical_accuracy: 0.8675\n",
      "Epoch 554/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.4259 - categorical_accuracy: 0.8668\n",
      "Epoch 555/1000\n",
      "1536/1536 [==============================] - 0s 121us/sample - loss: 0.4292 - categorical_accuracy: 0.8654\n",
      "Epoch 556/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.4287 - categorical_accuracy: 0.8662\n",
      "Epoch 557/1000\n",
      "1536/1536 [==============================] - 0s 120us/sample - loss: 0.4383 - categorical_accuracy: 0.8632\n",
      "Epoch 558/1000\n",
      "1536/1536 [==============================] - 0s 117us/sample - loss: 0.4470 - categorical_accuracy: 0.8613\n",
      "Epoch 559/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.4387 - categorical_accuracy: 0.8640\n",
      "Epoch 560/1000\n",
      "1536/1536 [==============================] - 0s 128us/sample - loss: 0.4331 - categorical_accuracy: 0.8646\n",
      "Epoch 561/1000\n",
      "1536/1536 [==============================] - 0s 130us/sample - loss: 0.4289 - categorical_accuracy: 0.8657\n",
      "Epoch 562/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.4287 - categorical_accuracy: 0.8665\n",
      "Epoch 563/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.4456 - categorical_accuracy: 0.8616\n",
      "Epoch 564/1000\n",
      "1536/1536 [==============================] - 0s 130us/sample - loss: 0.4386 - categorical_accuracy: 0.8634\n",
      "Epoch 565/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.4447 - categorical_accuracy: 0.8620\n",
      "Epoch 566/1000\n",
      "1536/1536 [==============================] - 0s 128us/sample - loss: 0.4360 - categorical_accuracy: 0.8643\n",
      "Epoch 567/1000\n",
      "1536/1536 [==============================] - 0s 120us/sample - loss: 0.4319 - categorical_accuracy: 0.8659\n",
      "Epoch 568/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.4282 - categorical_accuracy: 0.8661\n",
      "Epoch 569/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.4350 - categorical_accuracy: 0.8649\n",
      "Epoch 570/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.4326 - categorical_accuracy: 0.8662\n",
      "Epoch 571/1000\n",
      "1536/1536 [==============================] - 0s 130us/sample - loss: 0.4474 - categorical_accuracy: 0.8618\n",
      "Epoch 572/1000\n",
      "1536/1536 [==============================] - 0s 133us/sample - loss: 0.4392 - categorical_accuracy: 0.8649\n",
      "Epoch 573/1000\n",
      "1536/1536 [==============================] - 0s 130us/sample - loss: 0.4346 - categorical_accuracy: 0.8652\n",
      "Epoch 574/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.4283 - categorical_accuracy: 0.8672\n",
      "Epoch 575/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.4276 - categorical_accuracy: 0.8664\n",
      "Epoch 576/1000\n",
      "1536/1536 [==============================] - 0s 128us/sample - loss: 0.4306 - categorical_accuracy: 0.8664\n",
      "Epoch 577/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.4312 - categorical_accuracy: 0.8669\n",
      "Epoch 578/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.4272 - categorical_accuracy: 0.8675\n",
      "Epoch 579/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.4301 - categorical_accuracy: 0.8665\n",
      "Epoch 580/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.4323 - categorical_accuracy: 0.8660\n",
      "Epoch 581/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.4284 - categorical_accuracy: 0.8676\n",
      "Epoch 582/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.4260 - categorical_accuracy: 0.8671\n",
      "Epoch 583/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.4307 - categorical_accuracy: 0.8665\n",
      "Epoch 584/1000\n",
      "1536/1536 [==============================] - 0s 130us/sample - loss: 0.4237 - categorical_accuracy: 0.8681\n",
      "Epoch 585/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.4234 - categorical_accuracy: 0.8675\n",
      "Epoch 586/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.4257 - categorical_accuracy: 0.8680\n",
      "Epoch 587/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.4222 - categorical_accuracy: 0.8691\n",
      "Epoch 588/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.4274 - categorical_accuracy: 0.8661\n",
      "Epoch 589/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.4293 - categorical_accuracy: 0.8660\n",
      "Epoch 590/1000\n",
      "1536/1536 [==============================] - 0s 128us/sample - loss: 0.4262 - categorical_accuracy: 0.8674\n",
      "Epoch 591/1000\n",
      "1536/1536 [==============================] - 0s 128us/sample - loss: 0.4239 - categorical_accuracy: 0.8681\n",
      "Epoch 592/1000\n",
      "1536/1536 [==============================] - 0s 141us/sample - loss: 0.4214 - categorical_accuracy: 0.8683\n",
      "Epoch 593/1000\n",
      "1536/1536 [==============================] - 0s 132us/sample - loss: 0.4268 - categorical_accuracy: 0.8668\n",
      "Epoch 594/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.4293 - categorical_accuracy: 0.8672\n",
      "Epoch 595/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.4290 - categorical_accuracy: 0.8668\n",
      "Epoch 596/1000\n",
      "1536/1536 [==============================] - 0s 128us/sample - loss: 0.4345 - categorical_accuracy: 0.8656\n",
      "Epoch 597/1000\n",
      "1536/1536 [==============================] - 0s 128us/sample - loss: 0.4367 - categorical_accuracy: 0.8644\n",
      "Epoch 598/1000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.4274 - categorical_accuracy: 0.8681\n",
      "Epoch 599/1000\n",
      "1536/1536 [==============================] - 0s 128us/sample - loss: 0.4343 - categorical_accuracy: 0.8656\n",
      "Epoch 600/1000\n",
      "1536/1536 [==============================] - 0s 133us/sample - loss: 0.4350 - categorical_accuracy: 0.8634\n",
      "Epoch 601/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.4323 - categorical_accuracy: 0.8656\n",
      "Epoch 602/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.4281 - categorical_accuracy: 0.8673\n",
      "Epoch 603/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.4285 - categorical_accuracy: 0.8663\n",
      "Epoch 604/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.4356 - categorical_accuracy: 0.8634\n",
      "Epoch 605/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.4415 - categorical_accuracy: 0.8611\n",
      "Epoch 606/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.4380 - categorical_accuracy: 0.8649\n",
      "Epoch 607/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.4419 - categorical_accuracy: 0.8608\n",
      "Epoch 608/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.4495 - categorical_accuracy: 0.8608\n",
      "Epoch 609/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.4493 - categorical_accuracy: 0.8603\n",
      "Epoch 610/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.4369 - categorical_accuracy: 0.8650\n",
      "Epoch 611/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.4312 - categorical_accuracy: 0.8665\n",
      "Epoch 612/1000\n",
      "1536/1536 [==============================] - 0s 120us/sample - loss: 0.4284 - categorical_accuracy: 0.8661\n",
      "Epoch 613/1000\n",
      "1536/1536 [==============================] - 0s 121us/sample - loss: 0.4268 - categorical_accuracy: 0.8658\n",
      "Epoch 614/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.4246 - categorical_accuracy: 0.8679\n",
      "Epoch 615/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.4265 - categorical_accuracy: 0.8676\n",
      "Epoch 616/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.4302 - categorical_accuracy: 0.8667\n",
      "Epoch 617/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.4350 - categorical_accuracy: 0.8648\n",
      "Epoch 618/1000\n",
      "1536/1536 [==============================] - 0s 121us/sample - loss: 0.4336 - categorical_accuracy: 0.8648\n",
      "Epoch 619/1000\n",
      "1536/1536 [==============================] - 0s 120us/sample - loss: 0.4344 - categorical_accuracy: 0.8652\n",
      "Epoch 620/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.4338 - categorical_accuracy: 0.8656\n",
      "Epoch 621/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.4363 - categorical_accuracy: 0.8639\n",
      "Epoch 622/1000\n",
      "1536/1536 [==============================] - 0s 118us/sample - loss: 0.4295 - categorical_accuracy: 0.8666\n",
      "Epoch 623/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.4257 - categorical_accuracy: 0.8686\n",
      "Epoch 624/1000\n",
      "1536/1536 [==============================] - 0s 116us/sample - loss: 0.4199 - categorical_accuracy: 0.8691\n",
      "Epoch 625/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.4204 - categorical_accuracy: 0.8684\n",
      "Epoch 626/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.4278 - categorical_accuracy: 0.8672\n",
      "Epoch 627/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.4278 - categorical_accuracy: 0.8674\n",
      "Epoch 628/1000\n",
      "1536/1536 [==============================] - 0s 121us/sample - loss: 0.4292 - categorical_accuracy: 0.8656\n",
      "Epoch 629/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.4250 - categorical_accuracy: 0.8687\n",
      "Epoch 630/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.4257 - categorical_accuracy: 0.8661\n",
      "Epoch 631/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.4259 - categorical_accuracy: 0.8672\n",
      "Epoch 632/1000\n",
      "1536/1536 [==============================] - 0s 132us/sample - loss: 0.4289 - categorical_accuracy: 0.8671\n",
      "Epoch 633/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.4467 - categorical_accuracy: 0.8608\n",
      "Epoch 634/1000\n",
      "1536/1536 [==============================] - 0s 128us/sample - loss: 0.4362 - categorical_accuracy: 0.8648\n",
      "Epoch 635/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.4394 - categorical_accuracy: 0.8631\n",
      "Epoch 636/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.4379 - categorical_accuracy: 0.8633\n",
      "Epoch 637/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.4333 - categorical_accuracy: 0.8658\n",
      "Epoch 638/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.4163 - categorical_accuracy: 0.8707\n",
      "Epoch 639/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.4056 - categorical_accuracy: 0.8743\n",
      "Epoch 640/1000\n",
      "1536/1536 [==============================] - 0s 132us/sample - loss: 0.4007 - categorical_accuracy: 0.8759\n",
      "Epoch 641/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.3985 - categorical_accuracy: 0.8771\n",
      "Epoch 642/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.3967 - categorical_accuracy: 0.8777\n",
      "Epoch 643/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.3953 - categorical_accuracy: 0.8783\n",
      "Epoch 644/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.3942 - categorical_accuracy: 0.8789\n",
      "Epoch 645/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.3931 - categorical_accuracy: 0.8787\n",
      "Epoch 646/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.3926 - categorical_accuracy: 0.8788\n",
      "Epoch 647/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.3917 - categorical_accuracy: 0.8791\n",
      "Epoch 648/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.3910 - categorical_accuracy: 0.8791\n",
      "Epoch 649/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.3905 - categorical_accuracy: 0.8793\n",
      "Epoch 650/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.3899 - categorical_accuracy: 0.8795\n",
      "Epoch 651/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.3895 - categorical_accuracy: 0.8795\n",
      "Epoch 652/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.3891 - categorical_accuracy: 0.8797\n",
      "Epoch 653/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.3889 - categorical_accuracy: 0.8799\n",
      "Epoch 654/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.3883 - categorical_accuracy: 0.8799\n",
      "Epoch 655/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.3880 - categorical_accuracy: 0.8798\n",
      "Epoch 656/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.3877 - categorical_accuracy: 0.8800\n",
      "Epoch 657/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.3874 - categorical_accuracy: 0.8801\n",
      "Epoch 658/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.3869 - categorical_accuracy: 0.8800\n",
      "Epoch 659/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.3867 - categorical_accuracy: 0.8803\n",
      "Epoch 660/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.3864 - categorical_accuracy: 0.8804\n",
      "Epoch 661/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.3860 - categorical_accuracy: 0.8804\n",
      "Epoch 662/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.3859 - categorical_accuracy: 0.8806\n",
      "Epoch 663/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.3857 - categorical_accuracy: 0.8808\n",
      "Epoch 664/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.3853 - categorical_accuracy: 0.8809\n",
      "Epoch 665/1000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.3851 - categorical_accuracy: 0.8813\n",
      "Epoch 666/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.3850 - categorical_accuracy: 0.8811\n",
      "Epoch 667/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.3847 - categorical_accuracy: 0.8809\n",
      "Epoch 668/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.3859 - categorical_accuracy: 0.8807\n",
      "Epoch 669/1000\n",
      "1536/1536 [==============================] - 0s 130us/sample - loss: 0.3863 - categorical_accuracy: 0.8801\n",
      "Epoch 670/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.3852 - categorical_accuracy: 0.8806\n",
      "Epoch 671/1000\n",
      "1536/1536 [==============================] - 0s 128us/sample - loss: 0.3846 - categorical_accuracy: 0.8809\n",
      "Epoch 672/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.3842 - categorical_accuracy: 0.8811\n",
      "Epoch 673/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.3838 - categorical_accuracy: 0.8812\n",
      "Epoch 674/1000\n",
      "1536/1536 [==============================] - 0s 130us/sample - loss: 0.3839 - categorical_accuracy: 0.8814\n",
      "Epoch 675/1000\n",
      "1536/1536 [==============================] - 0s 130us/sample - loss: 0.3836 - categorical_accuracy: 0.8814\n",
      "Epoch 676/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.3832 - categorical_accuracy: 0.8812\n",
      "Epoch 677/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.3830 - categorical_accuracy: 0.8813\n",
      "Epoch 678/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.3830 - categorical_accuracy: 0.8818\n",
      "Epoch 679/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.3826 - categorical_accuracy: 0.8819\n",
      "Epoch 680/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.3842 - categorical_accuracy: 0.8813\n",
      "Epoch 681/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.3834 - categorical_accuracy: 0.8818\n",
      "Epoch 682/1000\n",
      "1536/1536 [==============================] - 0s 121us/sample - loss: 0.3827 - categorical_accuracy: 0.8821\n",
      "Epoch 683/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.3825 - categorical_accuracy: 0.8820\n",
      "Epoch 684/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.3820 - categorical_accuracy: 0.8818\n",
      "Epoch 685/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.3818 - categorical_accuracy: 0.8822\n",
      "Epoch 686/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.3817 - categorical_accuracy: 0.8820\n",
      "Epoch 687/1000\n",
      "1536/1536 [==============================] - 0s 121us/sample - loss: 0.3814 - categorical_accuracy: 0.8821\n",
      "Epoch 688/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.3813 - categorical_accuracy: 0.8825\n",
      "Epoch 689/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.3811 - categorical_accuracy: 0.8822\n",
      "Epoch 690/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.3811 - categorical_accuracy: 0.8828\n",
      "Epoch 691/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.3808 - categorical_accuracy: 0.8821\n",
      "Epoch 692/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.3809 - categorical_accuracy: 0.8823\n",
      "Epoch 693/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.3805 - categorical_accuracy: 0.8825\n",
      "Epoch 694/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.3804 - categorical_accuracy: 0.8825\n",
      "Epoch 695/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.3807 - categorical_accuracy: 0.8829\n",
      "Epoch 696/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.3804 - categorical_accuracy: 0.8823\n",
      "Epoch 697/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.3804 - categorical_accuracy: 0.8825\n",
      "Epoch 698/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.3802 - categorical_accuracy: 0.8825\n",
      "Epoch 699/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.3806 - categorical_accuracy: 0.8828\n",
      "Epoch 700/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.3804 - categorical_accuracy: 0.8824\n",
      "Epoch 701/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.3803 - categorical_accuracy: 0.8818\n",
      "Epoch 702/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.3803 - categorical_accuracy: 0.8823\n",
      "Epoch 703/1000\n",
      "1536/1536 [==============================] - 0s 121us/sample - loss: 0.3799 - categorical_accuracy: 0.8827\n",
      "Epoch 704/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.3798 - categorical_accuracy: 0.8830\n",
      "Epoch 705/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.3797 - categorical_accuracy: 0.8829\n",
      "Epoch 706/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.3793 - categorical_accuracy: 0.8829\n",
      "Epoch 707/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.3791 - categorical_accuracy: 0.8834\n",
      "Epoch 708/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.3788 - categorical_accuracy: 0.8832\n",
      "Epoch 709/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.3788 - categorical_accuracy: 0.8833\n",
      "Epoch 710/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.3788 - categorical_accuracy: 0.8831\n",
      "Epoch 711/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.3790 - categorical_accuracy: 0.8832\n",
      "Epoch 712/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.3787 - categorical_accuracy: 0.8836\n",
      "Epoch 713/1000\n",
      "1536/1536 [==============================] - 0s 121us/sample - loss: 0.3785 - categorical_accuracy: 0.8830\n",
      "Epoch 714/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.3783 - categorical_accuracy: 0.8834\n",
      "Epoch 715/1000\n",
      "1536/1536 [==============================] - 0s 121us/sample - loss: 0.3785 - categorical_accuracy: 0.8831\n",
      "Epoch 716/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.3784 - categorical_accuracy: 0.8833\n",
      "Epoch 717/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.3782 - categorical_accuracy: 0.8833\n",
      "Epoch 718/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.3780 - categorical_accuracy: 0.8831\n",
      "Epoch 719/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.3778 - categorical_accuracy: 0.8838\n",
      "Epoch 720/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.3778 - categorical_accuracy: 0.8832\n",
      "Epoch 721/1000\n",
      "1536/1536 [==============================] - 0s 136us/sample - loss: 0.3779 - categorical_accuracy: 0.8837\n",
      "Epoch 722/1000\n",
      "1536/1536 [==============================] - 0s 130us/sample - loss: 0.3779 - categorical_accuracy: 0.8834\n",
      "Epoch 723/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.3779 - categorical_accuracy: 0.8834\n",
      "Epoch 724/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.3777 - categorical_accuracy: 0.8833\n",
      "Epoch 725/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.3773 - categorical_accuracy: 0.8836\n",
      "Epoch 726/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.3778 - categorical_accuracy: 0.8832\n",
      "Epoch 727/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.3775 - categorical_accuracy: 0.8835\n",
      "Epoch 728/1000\n",
      "1536/1536 [==============================] - 0s 137us/sample - loss: 0.3778 - categorical_accuracy: 0.8832\n",
      "Epoch 729/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.3774 - categorical_accuracy: 0.8833\n",
      "Epoch 730/1000\n",
      "1536/1536 [==============================] - 0s 133us/sample - loss: 0.3771 - categorical_accuracy: 0.8841\n",
      "Epoch 731/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.3774 - categorical_accuracy: 0.8835\n",
      "Epoch 732/1000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.3772 - categorical_accuracy: 0.8836\n",
      "Epoch 733/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.3771 - categorical_accuracy: 0.8836\n",
      "Epoch 734/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.3769 - categorical_accuracy: 0.8839\n",
      "Epoch 735/1000\n",
      "1536/1536 [==============================] - 0s 121us/sample - loss: 0.3772 - categorical_accuracy: 0.8835\n",
      "Epoch 736/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.3768 - categorical_accuracy: 0.8835\n",
      "Epoch 737/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.3770 - categorical_accuracy: 0.8837\n",
      "Epoch 738/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.3807 - categorical_accuracy: 0.8822\n",
      "Epoch 739/1000\n",
      "1536/1536 [==============================] - 0s 120us/sample - loss: 0.3827 - categorical_accuracy: 0.8819\n",
      "Epoch 740/1000\n",
      "1536/1536 [==============================] - 0s 118us/sample - loss: 0.3804 - categorical_accuracy: 0.8827\n",
      "Epoch 741/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.3788 - categorical_accuracy: 0.8830\n",
      "Epoch 742/1000\n",
      "1536/1536 [==============================] - 0s 128us/sample - loss: 0.3775 - categorical_accuracy: 0.8836\n",
      "Epoch 743/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.3771 - categorical_accuracy: 0.8840\n",
      "Epoch 744/1000\n",
      "1536/1536 [==============================] - 0s 121us/sample - loss: 0.3765 - categorical_accuracy: 0.8836\n",
      "Epoch 745/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.3761 - categorical_accuracy: 0.8840\n",
      "Epoch 746/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.3761 - categorical_accuracy: 0.8842\n",
      "Epoch 747/1000\n",
      "1536/1536 [==============================] - 0s 120us/sample - loss: 0.3759 - categorical_accuracy: 0.8841\n",
      "Epoch 748/1000\n",
      "1536/1536 [==============================] - 0s 121us/sample - loss: 0.3760 - categorical_accuracy: 0.8838\n",
      "Epoch 749/1000\n",
      "1536/1536 [==============================] - 0s 120us/sample - loss: 0.3755 - categorical_accuracy: 0.8840\n",
      "Epoch 750/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.3758 - categorical_accuracy: 0.8836\n",
      "Epoch 751/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.3758 - categorical_accuracy: 0.8838\n",
      "Epoch 752/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.3756 - categorical_accuracy: 0.8842\n",
      "Epoch 753/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.3752 - categorical_accuracy: 0.8840\n",
      "Epoch 754/1000\n",
      "1536/1536 [==============================] - 0s 128us/sample - loss: 0.3753 - categorical_accuracy: 0.8838\n",
      "Epoch 755/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.3753 - categorical_accuracy: 0.8841\n",
      "Epoch 756/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.3753 - categorical_accuracy: 0.8839\n",
      "Epoch 757/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.3756 - categorical_accuracy: 0.8842\n",
      "Epoch 758/1000\n",
      "1536/1536 [==============================] - 0s 119us/sample - loss: 0.3757 - categorical_accuracy: 0.8843\n",
      "Epoch 759/1000\n",
      "1536/1536 [==============================] - 0s 121us/sample - loss: 0.3754 - categorical_accuracy: 0.8843\n",
      "Epoch 760/1000\n",
      "1536/1536 [==============================] - 0s 121us/sample - loss: 0.3751 - categorical_accuracy: 0.8841\n",
      "Epoch 761/1000\n",
      "1536/1536 [==============================] - 0s 121us/sample - loss: 0.3758 - categorical_accuracy: 0.8843\n",
      "Epoch 762/1000\n",
      "1536/1536 [==============================] - 0s 119us/sample - loss: 0.3759 - categorical_accuracy: 0.8838\n",
      "Epoch 763/1000\n",
      "1536/1536 [==============================] - 0s 118us/sample - loss: 0.3765 - categorical_accuracy: 0.8834\n",
      "Epoch 764/1000\n",
      "1536/1536 [==============================] - 0s 120us/sample - loss: 0.3756 - categorical_accuracy: 0.8839\n",
      "Epoch 765/1000\n",
      "1536/1536 [==============================] - 0s 121us/sample - loss: 0.3752 - categorical_accuracy: 0.8844\n",
      "Epoch 766/1000\n",
      "1536/1536 [==============================] - 0s 121us/sample - loss: 0.3750 - categorical_accuracy: 0.8845\n",
      "Epoch 767/1000\n",
      "1536/1536 [==============================] - 0s 121us/sample - loss: 0.3744 - categorical_accuracy: 0.8844\n",
      "Epoch 768/1000\n",
      "1536/1536 [==============================] - 0s 120us/sample - loss: 0.3745 - categorical_accuracy: 0.8847\n",
      "Epoch 769/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.3744 - categorical_accuracy: 0.8847\n",
      "Epoch 770/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.3745 - categorical_accuracy: 0.8844\n",
      "Epoch 771/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.3744 - categorical_accuracy: 0.8846\n",
      "Epoch 772/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.3752 - categorical_accuracy: 0.8839\n",
      "Epoch 773/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.3745 - categorical_accuracy: 0.8851\n",
      "Epoch 774/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.3741 - categorical_accuracy: 0.8849\n",
      "Epoch 775/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.3740 - categorical_accuracy: 0.8849\n",
      "Epoch 776/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.3746 - categorical_accuracy: 0.8841\n",
      "Epoch 777/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.3745 - categorical_accuracy: 0.8841\n",
      "Epoch 778/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.3743 - categorical_accuracy: 0.8849\n",
      "Epoch 779/1000\n",
      "1536/1536 [==============================] - 0s 121us/sample - loss: 0.3743 - categorical_accuracy: 0.8844\n",
      "Epoch 780/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.3739 - categorical_accuracy: 0.8847\n",
      "Epoch 781/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.3734 - categorical_accuracy: 0.8849\n",
      "Epoch 782/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.3734 - categorical_accuracy: 0.8849\n",
      "Epoch 783/1000\n",
      "1536/1536 [==============================] - 0s 128us/sample - loss: 0.3745 - categorical_accuracy: 0.8847\n",
      "Epoch 784/1000\n",
      "1536/1536 [==============================] - 0s 133us/sample - loss: 0.3742 - categorical_accuracy: 0.8841\n",
      "Epoch 785/1000\n",
      "1536/1536 [==============================] - 0s 133us/sample - loss: 0.3740 - categorical_accuracy: 0.8845\n",
      "Epoch 786/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.3741 - categorical_accuracy: 0.8849\n",
      "Epoch 787/1000\n",
      "1536/1536 [==============================] - 0s 136us/sample - loss: 0.3744 - categorical_accuracy: 0.8847\n",
      "Epoch 788/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.3743 - categorical_accuracy: 0.8852\n",
      "Epoch 789/1000\n",
      "1536/1536 [==============================] - 0s 128us/sample - loss: 0.3742 - categorical_accuracy: 0.8847\n",
      "Epoch 790/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.3742 - categorical_accuracy: 0.8842\n",
      "Epoch 791/1000\n",
      "1536/1536 [==============================] - 0s 121us/sample - loss: 0.3742 - categorical_accuracy: 0.8839\n",
      "Epoch 792/1000\n",
      "1536/1536 [==============================] - 0s 120us/sample - loss: 0.3735 - categorical_accuracy: 0.8853\n",
      "Epoch 793/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.3730 - categorical_accuracy: 0.8852\n",
      "Epoch 794/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.3726 - categorical_accuracy: 0.8855\n",
      "Epoch 795/1000\n",
      "1536/1536 [==============================] - 0s 121us/sample - loss: 0.3728 - categorical_accuracy: 0.8849\n",
      "Epoch 796/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.3731 - categorical_accuracy: 0.8846\n",
      "Epoch 797/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.3731 - categorical_accuracy: 0.8850\n",
      "Epoch 798/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.3735 - categorical_accuracy: 0.8849\n",
      "Epoch 799/1000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.3736 - categorical_accuracy: 0.8848\n",
      "Epoch 800/1000\n",
      "1536/1536 [==============================] - 0s 121us/sample - loss: 0.3735 - categorical_accuracy: 0.8849\n",
      "Epoch 801/1000\n",
      "1536/1536 [==============================] - 0s 120us/sample - loss: 0.3733 - categorical_accuracy: 0.8852\n",
      "Epoch 802/1000\n",
      "1536/1536 [==============================] - 0s 120us/sample - loss: 0.3730 - categorical_accuracy: 0.8845\n",
      "Epoch 803/1000\n",
      "1536/1536 [==============================] - 0s 120us/sample - loss: 0.3735 - categorical_accuracy: 0.8850\n",
      "Epoch 804/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.3729 - categorical_accuracy: 0.8853\n",
      "Epoch 805/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.3733 - categorical_accuracy: 0.8848\n",
      "Epoch 806/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.3742 - categorical_accuracy: 0.8843\n",
      "Epoch 807/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.3741 - categorical_accuracy: 0.8844\n",
      "Epoch 808/1000\n",
      "1536/1536 [==============================] - 0s 118us/sample - loss: 0.3736 - categorical_accuracy: 0.8843\n",
      "Epoch 809/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.3733 - categorical_accuracy: 0.8844\n",
      "Epoch 810/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.3731 - categorical_accuracy: 0.8850\n",
      "Epoch 811/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.3730 - categorical_accuracy: 0.8851\n",
      "Epoch 812/1000\n",
      "1536/1536 [==============================] - 0s 120us/sample - loss: 0.3727 - categorical_accuracy: 0.8854\n",
      "Epoch 813/1000\n",
      "1536/1536 [==============================] - 0s 119us/sample - loss: 0.3724 - categorical_accuracy: 0.8849\n",
      "Epoch 814/1000\n",
      "1536/1536 [==============================] - 0s 117us/sample - loss: 0.3723 - categorical_accuracy: 0.8851\n",
      "Epoch 815/1000\n",
      "1536/1536 [==============================] - 0s 121us/sample - loss: 0.3724 - categorical_accuracy: 0.8853\n",
      "Epoch 816/1000\n",
      "1536/1536 [==============================] - 0s 113us/sample - loss: 0.3728 - categorical_accuracy: 0.8848\n",
      "Epoch 817/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.3722 - categorical_accuracy: 0.8851\n",
      "Epoch 818/1000\n",
      "1536/1536 [==============================] - 0s 120us/sample - loss: 0.3727 - categorical_accuracy: 0.8853\n",
      "Epoch 819/1000\n",
      "1536/1536 [==============================] - 0s 117us/sample - loss: 0.3725 - categorical_accuracy: 0.8854\n",
      "Epoch 820/1000\n",
      "1536/1536 [==============================] - 0s 118us/sample - loss: 0.3724 - categorical_accuracy: 0.8855\n",
      "Epoch 821/1000\n",
      "1536/1536 [==============================] - 0s 120us/sample - loss: 0.3727 - categorical_accuracy: 0.8851\n",
      "Epoch 822/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.3732 - categorical_accuracy: 0.8843\n",
      "Epoch 823/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.3735 - categorical_accuracy: 0.8850\n",
      "Epoch 824/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.3738 - categorical_accuracy: 0.8848\n",
      "Epoch 825/1000\n",
      "1536/1536 [==============================] - 0s 128us/sample - loss: 0.3732 - categorical_accuracy: 0.8846\n",
      "Epoch 826/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.3729 - categorical_accuracy: 0.8847\n",
      "Epoch 827/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.3725 - categorical_accuracy: 0.8848\n",
      "Epoch 828/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.3728 - categorical_accuracy: 0.8848\n",
      "Epoch 829/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.3730 - categorical_accuracy: 0.8848\n",
      "Epoch 830/1000\n",
      "1536/1536 [==============================] - 0s 128us/sample - loss: 0.3729 - categorical_accuracy: 0.8847\n",
      "Epoch 831/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.3719 - categorical_accuracy: 0.8855\n",
      "Epoch 832/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.3721 - categorical_accuracy: 0.8852\n",
      "Epoch 833/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.3731 - categorical_accuracy: 0.8851\n",
      "Epoch 834/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.3721 - categorical_accuracy: 0.8855\n",
      "Epoch 835/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.3718 - categorical_accuracy: 0.8857\n",
      "Epoch 836/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.3717 - categorical_accuracy: 0.8856\n",
      "Epoch 837/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.3716 - categorical_accuracy: 0.8862\n",
      "Epoch 838/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.3718 - categorical_accuracy: 0.8854\n",
      "Epoch 839/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.3712 - categorical_accuracy: 0.8859\n",
      "Epoch 840/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.3710 - categorical_accuracy: 0.8858\n",
      "Epoch 841/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.3714 - categorical_accuracy: 0.8854\n",
      "Epoch 842/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.3716 - categorical_accuracy: 0.8856\n",
      "Epoch 843/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.3714 - categorical_accuracy: 0.8856\n",
      "Epoch 844/1000\n",
      "1536/1536 [==============================] - 0s 121us/sample - loss: 0.3713 - categorical_accuracy: 0.8856\n",
      "Epoch 845/1000\n",
      "1536/1536 [==============================] - 0s 121us/sample - loss: 0.3712 - categorical_accuracy: 0.8857\n",
      "Epoch 846/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.3713 - categorical_accuracy: 0.8860\n",
      "Epoch 847/1000\n",
      "1536/1536 [==============================] - 0s 121us/sample - loss: 0.3713 - categorical_accuracy: 0.8860\n",
      "Epoch 848/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.3713 - categorical_accuracy: 0.8856\n",
      "Epoch 849/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.3709 - categorical_accuracy: 0.8856\n",
      "Epoch 850/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.3709 - categorical_accuracy: 0.8860\n",
      "Epoch 851/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.3713 - categorical_accuracy: 0.8859\n",
      "Epoch 852/1000\n",
      "1536/1536 [==============================] - 0s 128us/sample - loss: 0.3714 - categorical_accuracy: 0.8854\n",
      "Epoch 853/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.3718 - categorical_accuracy: 0.8858\n",
      "Epoch 854/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.3711 - categorical_accuracy: 0.8853\n",
      "Epoch 855/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.3716 - categorical_accuracy: 0.8858\n",
      "Epoch 856/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.3721 - categorical_accuracy: 0.8854\n",
      "Epoch 857/1000\n",
      "1536/1536 [==============================] - 0s 121us/sample - loss: 0.3718 - categorical_accuracy: 0.8853\n",
      "Epoch 858/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.3715 - categorical_accuracy: 0.8852\n",
      "Epoch 859/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.3715 - categorical_accuracy: 0.8853\n",
      "Epoch 860/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.3712 - categorical_accuracy: 0.8854\n",
      "Epoch 861/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.3706 - categorical_accuracy: 0.8856\n",
      "Epoch 862/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.3710 - categorical_accuracy: 0.8855\n",
      "Epoch 863/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.3706 - categorical_accuracy: 0.8854\n",
      "Epoch 864/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.3707 - categorical_accuracy: 0.8858\n",
      "Epoch 865/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.3703 - categorical_accuracy: 0.8859\n",
      "Epoch 866/1000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1536/1536 [==============================] - 0s 121us/sample - loss: 0.3700 - categorical_accuracy: 0.8857\n",
      "Epoch 867/1000\n",
      "1536/1536 [==============================] - 0s 120us/sample - loss: 0.3702 - categorical_accuracy: 0.8863\n",
      "Epoch 868/1000\n",
      "1536/1536 [==============================] - 0s 121us/sample - loss: 0.3705 - categorical_accuracy: 0.8858\n",
      "Epoch 869/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.3708 - categorical_accuracy: 0.8856\n",
      "Epoch 870/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.3708 - categorical_accuracy: 0.8854\n",
      "Epoch 871/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.3709 - categorical_accuracy: 0.8854\n",
      "Epoch 872/1000\n",
      "1536/1536 [==============================] - 0s 120us/sample - loss: 0.3707 - categorical_accuracy: 0.8857\n",
      "Epoch 873/1000\n",
      "1536/1536 [==============================] - 0s 121us/sample - loss: 0.3708 - categorical_accuracy: 0.8859\n",
      "Epoch 874/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.3703 - categorical_accuracy: 0.8858\n",
      "Epoch 875/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.3702 - categorical_accuracy: 0.8860\n",
      "Epoch 876/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.3701 - categorical_accuracy: 0.8855\n",
      "Epoch 877/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.3703 - categorical_accuracy: 0.8856\n",
      "Epoch 878/1000\n",
      "1536/1536 [==============================] - 0s 121us/sample - loss: 0.3704 - categorical_accuracy: 0.8862\n",
      "Epoch 879/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.3705 - categorical_accuracy: 0.8862\n",
      "Epoch 880/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.3705 - categorical_accuracy: 0.8858\n",
      "Epoch 881/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.3705 - categorical_accuracy: 0.8856\n",
      "Epoch 882/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.3705 - categorical_accuracy: 0.8856\n",
      "Epoch 883/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.3702 - categorical_accuracy: 0.8859\n",
      "Epoch 884/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.3700 - categorical_accuracy: 0.8858\n",
      "Epoch 885/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.3698 - categorical_accuracy: 0.8859\n",
      "Epoch 886/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.3700 - categorical_accuracy: 0.8856\n",
      "Epoch 887/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.3711 - categorical_accuracy: 0.8856\n",
      "Epoch 888/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.3705 - categorical_accuracy: 0.8856\n",
      "Epoch 889/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.3702 - categorical_accuracy: 0.8857\n",
      "Epoch 890/1000\n",
      "1536/1536 [==============================] - 0s 121us/sample - loss: 0.3705 - categorical_accuracy: 0.8858\n",
      "Epoch 891/1000\n",
      "1536/1536 [==============================] - 0s 121us/sample - loss: 0.3699 - categorical_accuracy: 0.8856\n",
      "Epoch 892/1000\n",
      "1536/1536 [==============================] - 0s 121us/sample - loss: 0.3698 - categorical_accuracy: 0.8859\n",
      "Epoch 893/1000\n",
      "1536/1536 [==============================] - 0s 121us/sample - loss: 0.3703 - categorical_accuracy: 0.8858\n",
      "Epoch 894/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.3699 - categorical_accuracy: 0.8863\n",
      "Epoch 895/1000\n",
      "1536/1536 [==============================] - 0s 120us/sample - loss: 0.3702 - categorical_accuracy: 0.8860\n",
      "Epoch 896/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.3702 - categorical_accuracy: 0.8859\n",
      "Epoch 897/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.3700 - categorical_accuracy: 0.8858\n",
      "Epoch 898/1000\n",
      "1536/1536 [==============================] - 0s 128us/sample - loss: 0.3694 - categorical_accuracy: 0.8858\n",
      "Epoch 899/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.3692 - categorical_accuracy: 0.8863\n",
      "Epoch 900/1000\n",
      "1536/1536 [==============================] - 0s 120us/sample - loss: 0.3692 - categorical_accuracy: 0.8861\n",
      "Epoch 901/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.3731 - categorical_accuracy: 0.8848\n",
      "Epoch 902/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.3723 - categorical_accuracy: 0.8850\n",
      "Epoch 903/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.3712 - categorical_accuracy: 0.8854\n",
      "Epoch 904/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.3706 - categorical_accuracy: 0.8853\n",
      "Epoch 905/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.3706 - categorical_accuracy: 0.8859\n",
      "Epoch 906/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.3704 - categorical_accuracy: 0.8857\n",
      "Epoch 907/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.3698 - categorical_accuracy: 0.8858\n",
      "Epoch 908/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.3696 - categorical_accuracy: 0.8863\n",
      "Epoch 909/1000\n",
      "1536/1536 [==============================] - 0s 119us/sample - loss: 0.3695 - categorical_accuracy: 0.8859\n",
      "Epoch 910/1000\n",
      "1536/1536 [==============================] - 0s 119us/sample - loss: 0.3689 - categorical_accuracy: 0.8864\n",
      "Epoch 911/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.3694 - categorical_accuracy: 0.8858\n",
      "Epoch 912/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.3695 - categorical_accuracy: 0.8860\n",
      "Epoch 913/1000\n",
      "1536/1536 [==============================] - 0s 117us/sample - loss: 0.3689 - categorical_accuracy: 0.8863\n",
      "Epoch 914/1000\n",
      "1536/1536 [==============================] - 0s 119us/sample - loss: 0.3688 - categorical_accuracy: 0.8861\n",
      "Epoch 915/1000\n",
      "1536/1536 [==============================] - 0s 118us/sample - loss: 0.3688 - categorical_accuracy: 0.8866\n",
      "Epoch 916/1000\n",
      "1536/1536 [==============================] - 0s 119us/sample - loss: 0.3688 - categorical_accuracy: 0.8864\n",
      "Epoch 917/1000\n",
      "1536/1536 [==============================] - 0s 119us/sample - loss: 0.3692 - categorical_accuracy: 0.8859\n",
      "Epoch 918/1000\n",
      "1536/1536 [==============================] - 0s 121us/sample - loss: 0.3691 - categorical_accuracy: 0.8863\n",
      "Epoch 919/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.3692 - categorical_accuracy: 0.8860\n",
      "Epoch 920/1000\n",
      "1536/1536 [==============================] - 0s 121us/sample - loss: 0.3692 - categorical_accuracy: 0.8863\n",
      "Epoch 921/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.3699 - categorical_accuracy: 0.8858\n",
      "Epoch 922/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.3701 - categorical_accuracy: 0.8856\n",
      "Epoch 923/1000\n",
      "1536/1536 [==============================] - 0s 121us/sample - loss: 0.3695 - categorical_accuracy: 0.8859\n",
      "Epoch 924/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.3692 - categorical_accuracy: 0.8860\n",
      "Epoch 925/1000\n",
      "1536/1536 [==============================] - 0s 130us/sample - loss: 0.3705 - categorical_accuracy: 0.8856\n",
      "Epoch 926/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.3715 - categorical_accuracy: 0.8847\n",
      "Epoch 927/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.3707 - categorical_accuracy: 0.8857\n",
      "Epoch 928/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.3739 - categorical_accuracy: 0.8837\n",
      "Epoch 929/1000\n",
      "1536/1536 [==============================] - 0s 130us/sample - loss: 0.3742 - categorical_accuracy: 0.8841\n",
      "Epoch 930/1000\n",
      "1536/1536 [==============================] - 0s 128us/sample - loss: 0.3725 - categorical_accuracy: 0.8845\n",
      "Epoch 931/1000\n",
      "1536/1536 [==============================] - 0s 130us/sample - loss: 0.3707 - categorical_accuracy: 0.8857\n",
      "Epoch 932/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.3698 - categorical_accuracy: 0.8857\n",
      "Epoch 933/1000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1536/1536 [==============================] - 0s 121us/sample - loss: 0.3692 - categorical_accuracy: 0.8856\n",
      "Epoch 934/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.3695 - categorical_accuracy: 0.8863\n",
      "Epoch 935/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.3697 - categorical_accuracy: 0.8859\n",
      "Epoch 936/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.3692 - categorical_accuracy: 0.8859\n",
      "Epoch 937/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.3698 - categorical_accuracy: 0.8857\n",
      "Epoch 938/1000\n",
      "1536/1536 [==============================] - 0s 120us/sample - loss: 0.3696 - categorical_accuracy: 0.8860\n",
      "Epoch 939/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.3695 - categorical_accuracy: 0.8865\n",
      "Epoch 940/1000\n",
      "1536/1536 [==============================] - 0s 119us/sample - loss: 0.3695 - categorical_accuracy: 0.8860\n",
      "Epoch 941/1000\n",
      "1536/1536 [==============================] - 0s 126us/sample - loss: 0.3688 - categorical_accuracy: 0.8863\n",
      "Epoch 942/1000\n",
      "1536/1536 [==============================] - 0s 128us/sample - loss: 0.3686 - categorical_accuracy: 0.8864\n",
      "Epoch 943/1000\n",
      "1536/1536 [==============================] - 0s 118us/sample - loss: 0.3681 - categorical_accuracy: 0.8866\n",
      "Epoch 944/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.3686 - categorical_accuracy: 0.8866\n",
      "Epoch 945/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.3687 - categorical_accuracy: 0.8858\n",
      "Epoch 946/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.3693 - categorical_accuracy: 0.8858\n",
      "Epoch 947/1000\n",
      "1536/1536 [==============================] - 0s 121us/sample - loss: 0.3693 - categorical_accuracy: 0.8860\n",
      "Epoch 948/1000\n",
      "1536/1536 [==============================] - 0s 121us/sample - loss: 0.3687 - categorical_accuracy: 0.8862\n",
      "Epoch 949/1000\n",
      "1536/1536 [==============================] - 0s 120us/sample - loss: 0.3690 - categorical_accuracy: 0.8859\n",
      "Epoch 950/1000\n",
      "1536/1536 [==============================] - 0s 125us/sample - loss: 0.3685 - categorical_accuracy: 0.8863\n",
      "Epoch 951/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.3688 - categorical_accuracy: 0.8856\n",
      "Epoch 952/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.3691 - categorical_accuracy: 0.8858\n",
      "Epoch 953/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.3694 - categorical_accuracy: 0.8861\n",
      "Epoch 954/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.3681 - categorical_accuracy: 0.8863\n",
      "Epoch 955/1000\n",
      "1536/1536 [==============================] - 0s 121us/sample - loss: 0.3679 - categorical_accuracy: 0.8864\n",
      "Epoch 956/1000\n",
      "1536/1536 [==============================] - 0s 122us/sample - loss: 0.3681 - categorical_accuracy: 0.8860\n",
      "Epoch 957/1000\n",
      "1536/1536 [==============================] - 0s 130us/sample - loss: 0.3681 - categorical_accuracy: 0.8866\n",
      "Epoch 958/1000\n",
      "1536/1536 [==============================] - 0s 124us/sample - loss: 0.3682 - categorical_accuracy: 0.8866\n",
      "Epoch 959/1000\n",
      "1536/1536 [==============================] - 0s 121us/sample - loss: 0.3681 - categorical_accuracy: 0.8866\n",
      "Epoch 960/1000\n",
      "1536/1536 [==============================] - 0s 121us/sample - loss: 0.3688 - categorical_accuracy: 0.8860\n",
      "Epoch 961/1000\n",
      "1536/1536 [==============================] - 0s 147us/sample - loss: 0.3689 - categorical_accuracy: 0.8860\n",
      "Epoch 962/1000\n",
      "1536/1536 [==============================] - 0s 143us/sample - loss: 0.3691 - categorical_accuracy: 0.8859\n",
      "Epoch 963/1000\n",
      "1536/1536 [==============================] - 0s 152us/sample - loss: 0.3703 - categorical_accuracy: 0.8850\n",
      "Epoch 964/1000\n",
      "1536/1536 [==============================] - 0s 127us/sample - loss: 0.3713 - categorical_accuracy: 0.8849\n",
      "Epoch 965/1000\n",
      "1536/1536 [==============================] - 0s 132us/sample - loss: 0.3703 - categorical_accuracy: 0.8856\n",
      "Epoch 966/1000\n",
      "1536/1536 [==============================] - 0s 135us/sample - loss: 0.3700 - categorical_accuracy: 0.8857\n",
      "Epoch 967/1000\n",
      "1536/1536 [==============================] - 0s 130us/sample - loss: 0.3698 - categorical_accuracy: 0.8856\n",
      "Epoch 968/1000\n",
      "1536/1536 [==============================] - 0s 135us/sample - loss: 0.3718 - categorical_accuracy: 0.8849\n",
      "Epoch 969/1000\n",
      "1536/1536 [==============================] - 0s 132us/sample - loss: 0.3703 - categorical_accuracy: 0.8855\n",
      "Epoch 970/1000\n",
      "1536/1536 [==============================] - 0s 123us/sample - loss: 0.3686 - categorical_accuracy: 0.8858\n",
      "Epoch 971/1000\n",
      "1536/1536 [==============================] - 0s 144us/sample - loss: 0.3679 - categorical_accuracy: 0.8863\n",
      "Epoch 972/1000\n",
      "1536/1536 [==============================] - 0s 145us/sample - loss: 0.3681 - categorical_accuracy: 0.8867\n",
      "Epoch 973/1000\n",
      "1536/1536 [==============================] - 0s 137us/sample - loss: 0.3682 - categorical_accuracy: 0.8869\n",
      "Epoch 974/1000\n",
      "1536/1536 [==============================] - 0s 128us/sample - loss: 0.3682 - categorical_accuracy: 0.8863\n",
      "Epoch 975/1000\n",
      "1536/1536 [==============================] - 0s 130us/sample - loss: 0.3690 - categorical_accuracy: 0.8865\n",
      "Epoch 976/1000\n",
      "1536/1536 [==============================] - 0s 146us/sample - loss: 0.3684 - categorical_accuracy: 0.8862\n",
      "Epoch 977/1000\n",
      "1536/1536 [==============================] - 0s 132us/sample - loss: 0.3687 - categorical_accuracy: 0.8861\n",
      "Epoch 978/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.3688 - categorical_accuracy: 0.8860\n",
      "Epoch 979/1000\n",
      "1536/1536 [==============================] - 0s 129us/sample - loss: 0.3700 - categorical_accuracy: 0.8856\n",
      "Epoch 980/1000\n",
      "1536/1536 [==============================] - 0s 141us/sample - loss: 0.3686 - categorical_accuracy: 0.8859\n",
      "Epoch 981/1000\n",
      "1536/1536 [==============================] - 0s 135us/sample - loss: 0.3684 - categorical_accuracy: 0.8864\n",
      "Epoch 982/1000\n",
      "1536/1536 [==============================] - 0s 147us/sample - loss: 0.3677 - categorical_accuracy: 0.8866\n",
      "Epoch 983/1000\n",
      "1536/1536 [==============================] - 0s 148us/sample - loss: 0.3673 - categorical_accuracy: 0.8868\n",
      "Epoch 984/1000\n",
      "1536/1536 [==============================] - 0s 137us/sample - loss: 0.3671 - categorical_accuracy: 0.8868\n",
      "Epoch 985/1000\n",
      "1536/1536 [==============================] - 0s 137us/sample - loss: 0.3678 - categorical_accuracy: 0.8868\n",
      "Epoch 986/1000\n",
      "1536/1536 [==============================] - 0s 136us/sample - loss: 0.3685 - categorical_accuracy: 0.8861\n",
      "Epoch 987/1000\n",
      "1536/1536 [==============================] - 0s 143us/sample - loss: 0.3690 - categorical_accuracy: 0.8859\n",
      "Epoch 988/1000\n",
      "1536/1536 [==============================] - 0s 128us/sample - loss: 0.3687 - categorical_accuracy: 0.8861\n",
      "Epoch 989/1000\n",
      "1536/1536 [==============================] - 0s 132us/sample - loss: 0.3679 - categorical_accuracy: 0.8863\n",
      "Epoch 990/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.3678 - categorical_accuracy: 0.8863\n",
      "Epoch 991/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.3681 - categorical_accuracy: 0.8861\n",
      "Epoch 992/1000\n",
      "1536/1536 [==============================] - 0s 136us/sample - loss: 0.3687 - categorical_accuracy: 0.8862\n",
      "Epoch 993/1000\n",
      "1536/1536 [==============================] - 0s 139us/sample - loss: 0.3682 - categorical_accuracy: 0.8860\n",
      "Epoch 994/1000\n",
      "1536/1536 [==============================] - 0s 134us/sample - loss: 0.3675 - categorical_accuracy: 0.8867\n",
      "Epoch 995/1000\n",
      "1536/1536 [==============================] - 0s 128us/sample - loss: 0.3679 - categorical_accuracy: 0.8860\n",
      "Epoch 996/1000\n",
      "1536/1536 [==============================] - 0s 131us/sample - loss: 0.3671 - categorical_accuracy: 0.8872\n",
      "Epoch 997/1000\n",
      "1536/1536 [==============================] - 0s 138us/sample - loss: 0.3677 - categorical_accuracy: 0.8867\n",
      "Epoch 998/1000\n",
      "1536/1536 [==============================] - 0s 128us/sample - loss: 0.3674 - categorical_accuracy: 0.8861\n",
      "Epoch 999/1000\n",
      "1536/1536 [==============================] - 0s 137us/sample - loss: 0.3671 - categorical_accuracy: 0.8867\n",
      "Epoch 1000/1000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1536/1536 [==============================] - 0s 132us/sample - loss: 0.3668 - categorical_accuracy: 0.8871\n"
     ]
    }
   ],
   "source": [
    "model = Model({'dims': [X_train.shape[1:], Y_train.shape[1:]], \n",
    "               'X_train': X_train,\n",
    "               'Y_train': Y_train,\n",
    "               'epochs':1000, \n",
    "               'learning_rate': 0.01,\n",
    "               'batch_size': 64})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Original Name: serendipaceratops - Predicted Name: sacanoittleratops\n",
      "Original Name: peishansaurus - Predicted Name: palbtansaurus\n",
      "Original Name: pneumatoarthrus - Predicted Name: paeumatorlaoais\n",
      "Original Name: vulcanodon - Predicted Name: veaormoson\n",
      "Original Name: dimodosaurus - Predicted Name: dalaconaurus\n",
      "Original Name: coronosaurus - Predicted Name: cheysasaurus\n",
      "Original Name: shuangbaisaurus - Predicted Name: saaengaegsaurus\n",
      "Original Name: eugongbusaurus - Predicted Name: eurnngjusaurus\n",
      "Original Name: canardia - Predicted Name: chmlsoia\n",
      "Original Name: megalosaurus - Predicted Name: magamosaurus\n",
      "Original Name: kittysaurus - Predicted Name: kotaosaurus\n",
      "Original Name: saichania - Predicted Name: sausoinoa\n",
      "Original Name: arcusaurus - Predicted Name: anchlaurus\n",
      "Original Name: dandakosaurus - Predicted Name: dastacosaurus\n",
      "Original Name: siamodracon - Predicted Name: sanmodracon\n",
      "Original Name: leptospondylus - Predicted Name: lapeosaondylus\n",
      "Original Name: cheneosaurus - Predicted Name: chisonsaurus\n",
      "Original Name: opisthocoelicaudia - Predicted Name: ortsteosoelasaudia\n",
      "Original Name: gongpoquansaurus - Predicted Name: gangbaquansaurus\n",
      "Original Name: megadontosaurus - Predicted Name: magamontosaurus\n",
      "Original Name: adamantisaurus - Predicted Name: anrsantisaurus\n",
      "Original Name: quilmesaurus - Predicted Name: qialmisaurus\n",
      "Original Name: notohypsilophodon - Predicted Name: natocarsilophoson\n",
      "Original Name: diabloceratops - Predicted Name: dalneonoratops\n",
      "Original Name: pyroraptor - Predicted Name: palhraptor\n",
      "Original Name: loncosaurus - Predicted Name: langosaurus\n",
      "Original Name: aragosaurus - Predicted Name: ancsosaurus\n",
      "Original Name: mantellodon - Predicted Name: martirooson\n",
      "Original Name: chuxiongosaurus - Predicted Name: chiaiosgisaurus\n",
      "Original Name: brasileosaurus - Predicted Name: baacyleosaurus\n",
      "Original Name: serikornis - Predicted Name: sacasornis\n",
      "Original Name: zhenyuanlong - Predicted Name: zhujgaannong\n",
      "Original Name: shidaisaurus - Predicted Name: saaeossaurus\n",
      "Original Name: mantellisaurus - Predicted Name: martiroosaurus\n",
      "Original Name: auroraceratops - Predicted Name: ansornshratops\n",
      "Original Name: archaeornis - Predicted Name: anchaeornis\n",
      "Original Name: yibinosaurus - Predicted Name: yuzinosaurus\n",
      "Original Name: zupaysaurus - Predicted Name: zhninsaurus\n",
      "Original Name: vectensia - Predicted Name: velaosoha\n",
      "Original Name: adeopapposaurus - Predicted Name: anrtseloosaurus\n",
      "Original Name: velociraptor - Predicted Name: velocoraptor\n",
      "Original Name: syntarsus - Predicted Name: saltansas\n",
      "Original Name: pachyrhinosaurus - Predicted Name: palhysoinosaurus\n",
      "Original Name: byronosaurus - Predicted Name: barasosaurus\n",
      "Original Name: avaceratops - Predicted Name: anileratops\n",
      "Original Name: geminiraptor - Predicted Name: garbmuaaptor\n",
      "Original Name: parrosaurus - Predicted Name: palaosaurus\n",
      "Original Name: hierosaurus - Predicted Name: hesrosaurus\n",
      "Original Name: protorosaurus - Predicted Name: paotorosaurus\n",
      "Original Name: edmontonia - Predicted Name: eumontonia\n",
      "Original Name: trimucrodon - Predicted Name: taigiceodon\n",
      "Original Name: xenoposeidon - Predicted Name: xinopespidon\n",
      "Original Name: gobivenator - Predicted Name: ganicenator\n",
      "Original Name: achelousaurus - Predicted Name: aneerossaurus\n",
      "Original Name: palaeoctonus - Predicted Name: palaeoseonys\n",
      "Original Name: leonerasaurus - Predicted Name: lapcorocaurus\n",
      "Original Name: barilium - Predicted Name: baranous\n",
      "Original Name: euhelopus - Predicted Name: eurelosus\n",
      "Original Name: byranjaffia - Predicted Name: barandaffia\n",
      "Original Name: variraptor - Predicted Name: vercnaptor\n",
      "Original Name: acanthopholis - Predicted Name: anenchopaolos\n",
      "Original Name: graciliraptor - Predicted Name: gaycolisaptor\n",
      "Original Name: velocipes - Predicted Name: velocorls\n",
      "Original Name: chialingosaurus - Predicted Name: chintengosaurus\n",
      "Original Name: dysganus - Predicted Name: dassynas\n",
      "Original Name: rhabdodon - Predicted Name: raopioron\n",
      "Original Name: spinosaurus - Predicted Name: sainosaurus\n",
      "Original Name: muyelensaurus - Predicted Name: maienanoaurus\n",
      "Original Name: asiatosaurus - Predicted Name: antalosaurus\n",
      "Original Name: ouranosaurus - Predicted Name: orranosaurus\n",
      "Original Name: gojirasaurus - Predicted Name: ganiaapaurus\n",
      "Original Name: animantarx - Predicted Name: antaintosa\n",
      "Original Name: delapparentia - Predicted Name: daraniosentia\n",
      "Original Name: olorotitan - Predicted Name: orodanitan\n",
      "Original Name: lusotitan - Predicted Name: lasatetan\n",
      "Original Name: abydosaurus - Predicted Name: anrposaurus\n",
      "Original Name: malawisaurus - Predicted Name: marataaaurus\n",
      "Original Name: morelladon - Predicted Name: manrnloson\n",
      "Original Name: breviceratops - Predicted Name: baaniaeratops\n",
      "Original Name: nteregosaurus - Predicted Name: naaronasaurus\n",
      "Original Name: proceratosaurus - Predicted Name: paoteratosaurus\n",
      "Original Name: cerasinops - Predicted Name: chrataoops\n",
      "Original Name: stenotholus - Predicted Name: saegotholus\n",
      "Original Name: dromaeosauroides - Predicted Name: daomieosauruides\n",
      "Original Name: ankistrodon - Predicted Name: anthstroson\n",
      "Original Name: hungarosaurus - Predicted Name: heagarosaurus\n",
      "Original Name: losillasaurus - Predicted Name: lansneisaurus\n",
      "Original Name: gavinosaurus - Predicted Name: galenosaurus\n",
      "Original Name: mandschurosaurus - Predicted Name: martaahurosaurus\n",
      "Original Name: borealosaurus - Predicted Name: barealosaurus\n",
      "Original Name: struthiomimus - Predicted Name: saealhoosumus\n",
      "Original Name: erlikosaurus - Predicted Name: euiiaosaurus\n",
      "Original Name: lophorhothon - Predicted Name: lanuosnosaon\n",
      "Original Name: wyomingraptor - Predicted Name: waomingdaptor\n",
      "Original Name: elrhazosaurus - Predicted Name: euiiorosaurus\n",
      "Original Name: teleocrater - Predicted Name: tanousaoter\n",
      "Original Name: shunosaurus - Predicted Name: saaessaurus\n",
      "Original Name: soriatitan - Predicted Name: sanalnitan\n",
      "Original Name: basutodon - Predicted Name: baratipon\n",
      "Original Name: rahiolisaurus - Predicted Name: rahinmasaurus\n"
     ]
    }
   ],
   "source": [
    "for i in range(100):\n",
    "    name = ''\n",
    "    for idx in np.argmax(X_train[i,:,:], axis=1):\n",
    "        name = name + idx_to_char[idx]\n",
    "        if idx==0:\n",
    "            break\n",
    "    name_prediction = '' + name[0]\n",
    "    for idx in np.argmax(model.predict(X_train)[i,:,:], axis=1):\n",
    "        name_prediction = name_prediction + idx_to_char[idx]\n",
    "        if idx==0:\n",
    "            break\n",
    "    print('Original Name: {} - Predicted Name: {}'.format(name.split()[0], name_prediction.split()[0]))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "-----"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "def InferenceModel(params):\n",
    "    dims = params['dims']\n",
    "    model = params['model']\n",
    "    f1 = tf.keras.layers.SimpleRNN(units=50,\n",
    "                                  activation='tanh',\n",
    "                                  return_sequences=True,\n",
    "                                  return_state=True, \n",
    "                                  name='rnn')\n",
    "    f2 = tf.keras.layers.Lambda(lambda x: x[0])\n",
    "    f3 = tf.keras.layers.Lambda(lambda x: x[1])\n",
    "    f4 = tf.keras.layers.Dense(units=dims[1], \n",
    "                               activation='softmax', \n",
    "                               name='dense')\n",
    "    x0 = tf.keras.Input(shape=(1,dims[0]))\n",
    "    a0 = tf.keras.Input(shape=(50))\n",
    "    z0 = f1([x0,a0])\n",
    "    x1 = f2(z0)\n",
    "    a1 = f3(z0)\n",
    "    x2 = f4(x1)\n",
    "    inference_model = tf.keras.Model([x0,a0], [x2,a1]) \n",
    "    inference_model.get_layer(name='rnn').set_weights(model.get_layer(name='rnn').get_weights())\n",
    "    inference_model.get_layer(name='dense').set_weights(model.get_layer(name='dense').get_weights())\n",
    "    return inference_model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "inference_model = InferenceModel({'dims': [X_train.shape[2], Y_train.shape[2]],\n",
    "                                  'model': model})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Sampling names...\n",
      "Generated name 1/30: ebechnosaurus\n",
      "Generated name 2/30: ankerorna\n",
      "Generated name 3/30: uenisaurus\n",
      "Generated name 4/30: engesaurus\n",
      "Generated name 5/30: uewtogosaurus\n",
      "Generated name 6/30: anawasaurus\n",
      "Generated name 7/30: abienchasaurus\n",
      "Generated name 8/30: aeasceritosaurus\n",
      "Generated name 9/30: icernodon\n",
      "Generated name 10/30: amnanosaurus\n",
      "Generated name 11/30: ancoelisti\n",
      "Generated name 12/30: abbenssaursausaurus\n",
      "Generated name 13/30: inosaurus\n",
      "Generated name 14/30: iniangosaurus\n",
      "Generated name 15/30: raptorniptetis\n",
      "Generated name 16/30: abzerachus\n",
      "Generated name 17/30: uerypensa\n",
      "Generated name 18/30: oloraterys\n",
      "Generated name 19/30: errosia\n",
      "Generated name 20/30: aeilodon\n",
      "Generated name 21/30: ilaniang\n",
      "Generated name 22/30: canosaurus\n",
      "Generated name 23/30: ankerodon\n",
      "Generated name 24/30: aloskus\n",
      "Generated name 25/30: ichiosaurus\n",
      "Generated name 26/30: uectopcopieus\n",
      "Generated name 27/30: eratosaurus\n",
      "Generated name 28/30: uenidonafanosaurus\n",
      "Generated name 29/30: uengheda\n",
      "Generated name 30/30: enjolus\n"
     ]
    }
   ],
   "source": [
    "num_names = 30\n",
    "num_samples = max_length_name + 10\n",
    "print('Sampling names...')\n",
    "for i in range(num_names):\n",
    "    x = np.zeros(shape=(1,1,vocab_size))\n",
    "    a = np.zeros(shape=(1,50))\n",
    "    word = ''\n",
    "    for _ in range(num_samples):\n",
    "        z = inference_model.predict([x, a])\n",
    "        idx = np.random.choice(range(vocab_size), p = z[0].ravel())\n",
    "        word = word + idx_to_char[idx]\n",
    "        if idx==0:\n",
    "            break\n",
    "        x = np.zeros(shape=(1,1,vocab_size), dtype='float32')\n",
    "        x[0,0,idx] = 1.0\n",
    "        a = z[1]\n",
    "    print('Generated name {}/{}: {}'.format(i+1, num_names, word.split()[0]))"
   ]
  }
 ],
 "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": "3.6.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}