01.02.02.SequenceModeling_VanillaRNN.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Vanilla RNN Implementation  \n",
    "\n",
    "RNN을 이용한 sequence modeling을 실습하기 전에, 가장 기초적인 형태의 RNN 레이어를 실제로 구현해 보겠습니다.  \n",
    "향후 실습과정의 대부분은 keras를 기반으로 진행됩니다.  \n",
    "따라서 이번 장의 RNN 레이어 구현 과정을 통해 keras의 기본개념을 익히는 과정도 함께 진행해 보겠습니다.\n",
    "\n",
    "(참고)  \n",
    "https://keras.io/layers/recurrent/   \n",
    "케라스 창시자에게 배우는 딥러닝(프랑소와 숄레 저/박해선 역) 3.2.2절"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### keras 기본 워크플로우  \n",
    "\n",
    "지난 절에서 우리는 Multi-layer Perceptron을 활용한 sequence modeling을 구현하면서 이미 keras를 활용한 바 있습니다.  \n",
    "대략 다음과 같은 구조를 가지고 있었습니다.\n",
    "\n",
    "```\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense\n",
    "\n",
    "# step 1. define Dataset\n",
    "df = pd.read_csv(dataset_filepath)\n",
    "train_data = [([df.iat[i+j,1] for j in range(5)], df.iat[i+5,1]) for i in range(115)]\n",
    "test_data = [([df.iat[i+j,1] for j in range(5)], df.iat[i+5,1]) for i in range(115, 139)]\n",
    "\n",
    "# step 2. define Model with Layers\n",
    "model = Sequential()\n",
    "model.add(Dense(100, activation='relu', input_dim=5))\n",
    "model.add(Dense(1))\n",
    "\n",
    "# step 3. define how to train with loss/optimizer\n",
    "model.compile(optimizer='adam', loss='mse')\n",
    "\n",
    "# step 4. fit model to train\n",
    "model.fit(X_train, y_train, epochs=1000, verbose=1, validation_data=(X_val, y_val))\n",
    "\n",
    "# step 5. inference\n",
    "cur_y_hat = model.predict(x_input_reshape, verbose=0)\n",
    "```"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "위 소스코드는 keras의 창시자인 프랑소아 숄레의 저서에서 소개된 keras의 기본 워크플로우를 잘 보여 줍니다.  \n",
    "\n",
    "step 1 : [dataset]       입력 텐서와 타깃 텐서로 이루어진 훈련데이터를 정의한다.  \n",
    "step 2 : [model, layer]  입력과 타깃을 매핑하는 층(layer)으로 이루어진 네트워크(model)을 정의한다.  \n",
    "step 3 : [model.compile] 손실 함수, 옵티마이저, 모니터링하기 위한 측정 지표를 선택하여 학습 과정을 설정한다.  \n",
    "step 4 : [model.fit]     훈련 데이터에 대해 모델의 fit() 메소드를 반복적으로 호출한다.  \n",
    "step 5 : [model.predict] 훈련이 끝난 model을 활용하여 테스트데이터를 해석한다."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "MLP 대신 RNN을 활용하기 위한 작업은 위 step 2에 해당하는 작업입니다.  \n",
    "그러나 모델 구조가 바뀌면 step 1의 데이터구조, step 3의 측정 지표 등이 함께 영향을 받게 되므로 종합적인 검토가 필요합니다."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### sequential model  \n",
    "\n",
    "keras가 tensorflow, pytorch 등 딥러닝 프레임워크들에 비해 High Level 프레임워크라고 평가받게 되는 가장 중요한 특징이 keras의 model에 있습니다.  \n",
    "keras.model 클래스 안에는 keras 워크플로우를 구성하는 compile, fit, predict 등 주요한 메소드들이 모두 포함되어 있습니다. 즉 keras.model은 keras의 가장 핵심 클래스라고 할 수 있습니다.   (참고) https://keras.io/models/model/  \n",
    "이 클래스를 통해 keras는 딥러닝 워크플로우를 표준화하여 내부에 미리 구현함으로써, 사용자들로 하여금 최소한의 코딩만으로 손쉽게 딥러닝 모델을 구현하여 운영할 수 있게 됩니다.  \n",
    "\n",
    "특히 가장 많이 활용되는 대표적인 model인 sequence model은 이전 layer의 출력이 다음 layer의 입력이 되게끔 layer들이 sequence 구조로 배치된 아키텍처를 구현하였습니다. 이번 실습에서는 sequence model 내부의 Dense layer를 간단한 RNN 레이어로 바꾸어 보는 작업을 진행해 보겠습니다."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import keras\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense, RNN\n",
    "from keras import backend as K"
   ]
  },
  {
   "attachments": {
    "image.png": {
     "image/png": "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"
    }
   },
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![image.png](attachment:image.png)"
   ]
  },
  {
   "attachments": {
    "image.png": {
     "image/png": "iVBORw0KGgoAAAANSUhEUgAAAc0AAABNCAYAAADEgj9xAAAgAElEQVR4Ae2dd3wUx9nHv3enK+oFdYEQSIheJIoQpncwYJoB45K4xL3Ejh337jdxYjt2cNziuBv3igtuFNM7iCYkJNR7Lydd3/ezd0K6O52kEwgk7L1/bnf2mWee+e3sPDPPPPOMTBAEAeknISAhICEgISAhICHQIQLyDikkAgkBCQEJAQkBCQEJASsCktKUGoKEgISAhICEgISAmwhIStNNoCQyCQEJAQkBCQEJAUlpSm1AQkBCQEJAQkBCwE0EJKXpJlASmYSAhICEgISAhICkNKU2ICEgISAhICEgIeAmApLSdBMoiUxCQEJAQkBCQEJAUppSG5AQkBCQEJAQkBBwEwFJaboJlEQmISAhICEgISAhIClNqQ1ICEgISAhICEgIuImApDTdBEoikxCQEJAQkBCQEJCUptQGJAQkBCQEJAQkBNxEQFKabgIlkUkISAhICEgISAhISlNqAxICEgISAhICEgJuIuDhJp1EJiHQAQICtaXZpGUWoa0pJ1frzYw504nykXWQT3r820RAoK4oi/ScAqqryqgweHLRzHlEeXdcW6O2jBNpWdTV1JKfW0ffi6aQFBfUcUaJQkLgPCAgKc3zAPLvoghBy7avX+Nfb/9MelYBtf0X88u06UT9LiovVbI1Ag3s+PQlnvtkA8dPFaIcNJvPJrmnNPMP/sDj/3yDo8cyKC+P4r53EyWl2RpgKaWbEJDMs90E/G+vWA2Js67i//7xEON7lVBTqUUul2aZv7337G6N1IxacA1PPvMAY01VlJTVgpu9Teigqdz/yNP8aW44FbUlmE1uZnRXNIlOQuAsEJBa41mAJ2W1Q0DmQXjMEJImLmby2GhAsHsoXf7+EPAgrP9QkpIvYdrYENQKwe0m4R3ch9FjxrNo9lRAJbWl31/j6dE17rlK02LGbLK4AZ6A0ewOnRusJJIuQMCIj38ECO53kl1QqMSixyLggV+kBoWH0OlhlIdnIKDoUTWzmExYLD1rQCiIfaXUBZ63dtIjlaagK2XL5l0U1+rdAMJIYfputh8tc4NWIjk/CPw2zLImgwGz2Z0O0ozOaD4/0F6Ipfw2mgP66kw2bUlBa+hZGsqkr2D3r3spb7wQG8eFJ3MPVJoG9nz2DltPVuMZ6OkGoioCZLX88NEaNufUu0EvkZx7BNxRNOdeirMpob7wIBt+SaFR70ZdBD1Zhzax+XDp2RT5283rBoQ9vvKGata/8TYn6kCj6VmzX6Xak/qMb3jxw83U9Sx93uNf65kI2OOUZuXhL/hiv45LV11MkJsjVP9Bc1g51oMP1/yPMsOZwCDlkRCwQ0BbxGevfkWp3AsvLzc+EZkXwap6fvjoRXYUNNgxki5/Kwgc//F1dmn7s2LhaJTdUqk6tm/8kZ1Zda1Ll/sye8XlyFL+xwfbT7V+LqV0KQJu9AhdWl77zAwlvPbetwRMW0K8vytSE3u+fp331x3A6PR42LxriS3ZyZvrs5yeSLfdhUDLBMNAcU42RZUXgv3Iwt5PXyVVk8CSuUPddfgkZORiLh1m4v0171Ju6i7Ee3a5Le3BSGFuFkVV7iy/iH5ALTnryvLJyS/jfE6oTJX7+c+6wyQuWUSIqx7TVMH3b7/Itzvzu/gFNJJ5ZDefvP0KLz12D5dffycfHCx2XYb/QK6bF8+W196hQOuaRErtGgR61D7N4o0fsy8nmqceH+a6drpTvH7XfaRf/AIrFyU60qh6c/GiQfzls7dZPe9x+ohOd9KvGxGQofGCmvTNfLYll9DIQHL2rSFDGMFdd/+R6OZN7vV8/+bL/JxhZGBcBGqZjoxDhcTP+CN/WBQHunxeevlDalS+RHh7UJafiV9cMldetohmFl1YS13BBt7cnM3iR/6Cjyu+QjXrXv43ZX2Wcc2iYdgbQ0Yvuoavv3qEtZsWcscsaYeqI3wyNBqoS93A59sLCO4dyKk9L5KlGsEdt/+RGC9Havs7hdoLLAV88e5PGHuFQeFRnjuiY9GNNzFzWIg96Tm4Fji0di1lnlOZP7yXS/76nJ08fcsDDHpxIguSe7ukOaNEi56ivAJq9RYspgZyasBL1XaX3XvKVUS+8ife2JzDIxf3PaMipUwdI+Bq3NRxrnNCoeWrdVsIGjaVQW19QAUn2FETyOCxw1yaSAZPnoRn7g6+31t5TiSUmHYCAbUHZSdS2LQ1i+C4QYwaM45JyTHs+e8N3PLcR9Q3Tx7kBIVGEO1n4OuX7+PGm2/jua8PoQxoagRyDRTv4uHbbuLmx/9NaoWCXiG9zpFPpYlt735JTcRcpvT3c1lZc8lhXnngWX48WeagMK3EPgOYNzOM9Z+8T5nkF+SAn8JDTXXGQb7bm0/4gEEkJCYxaXw0v/79fv76xGfUOlDb33hg1uaz58eN6IL6MnR4IsmTJqE48R7X3PkYh0rPMdCNWXz0YxoJU6a4HkQBFSePkaKIJ2FkH3vBz/5a5kvilPlcd8Mt3HbtErw8ZRgt7cyxPWOYelEvNn7xBTVnX7rEoQ0EeozSFCoPs+N4OfHTRrXujJqELzi5j9IAFaOHDHBZHXn4WCaEmfh1wy6Xz6XE84OATCYHXSV7t+4keNISLpk6jj6hYYycdRvXLh7Et8+9y/6i0wZ2L8YvuJI773uYv913Db3MFnovWM1lkyNtwqqC+eOqRST1S+axf/2PZ/7+FJfOvAjNuahK9RE+2X2CUdOmtMlfe+oY+2TRJCTEuZRgzLSJCMe3sum4ZCNrBkjmgYelgF83nCImeSFzp4wjKjSUhNl/5taFSj798BX2F7uyadvm8Vl7fyTHfxQrFk1nWEw40UOS+cudKyj+5QM+2XCyuZhzcVGbuplDNWpGJvVvk3166k5M/SNJ6BvcJs0ZPZAp8PK2tXR9rdZqpba3bLjimTxhMvqjW9jjEk9XOaS0ziLQY5RmxbGDpOuDGRPrbG4xUVNdSXV9Ffv370FDAEEeRuprq9G3GmT6M3ZcONmp+yjvLBK/EXqLuRUo571mMg8VnNpKvnow4+MD7MqXEdJ/JNSmkZHrPLfwIGH5Qzx81RjyvniOzw5XNeUTOJ6Rz6InnuavS5MIdseh2q7EzlyWHtrKqUp/khMj2syWcWI3+qhejI53PatQxiYz1r+ajVtS2uTxe3ug0KgoO3qMak0fxg9wjCEbPTIIKgtIz3Xh4CIT11gy2FpoITlpOPaGSd/wwfSnmhMZx8/p+uaJvYeo9uvPsHCnrlLQU1VZSXV1PgcOHCHMNwylvp66ujpMzVaU8/+mg4YMpa8pi30HKs5/4T2mRAHzOdy46tQSuq/WOWlHMfhE0jvYaSxlKOGrV//JX+68jb+9c4AGk4F1L97DX596g8OlrV1l+8SHY848Rtbv0D5RlbmNz7/dh6EbP1qxBQkyCxh9GDx4qENHJz6L9A/Fn2zKa1y9IB+uvus+EqsO8PiTryGqTeOpX/ihOIxVSye3aYHoqlZ77NAe6gIHEu/cBi2NFObnkleYyY7dB/BTBGCuKKSoqASd8xhFHkligj8nDh9Emmva3oxCMGM0hdFvRL9WryowfCTUm6gudLFdTBC3duiIHRBLlNMuD41fL8S4UyX1ta2cAlsVcsYJZtLS0vAOiyXYXmOLbbziBK8//Sh3/+UOXvupDJM2m5cf/TOPv/Q1+d3pQO0fR2QkpJ9IO+NaX/AZLXXs2fwd64+em2U6p6ZwFnAJJhobjSAT0DWIDdmToEB/PNxUy3n5qWiCZuDv7MCjDGf5zfexUHuS65dso27k9Tz/9HKQKfHxbe38HRI+CIt5N7nFJsb6d131zgKZ85LVUH6QF1/eROIVN6C0G3dYLBbkcseXYJ2NyhU0h4YVBMQgJ46xYi2Ik1aFwjGvYLEgc+AnWCOk2JchIJra4omJ8G1Vdw+5AgUWLILrtRnNkKU8+X+rmXf3y7z2SRIxNekMTF7arqNIq0LOKMHAyYx0/PrMxH5ubGWlzefL11/hUGE+27/LwxLciy9feoQfo8Zx4x3XM8ghg4y+sUFotx6j0AQDelgTFN+9XOGogSxmE8g9WtoDYDFbkDu8e9voXcxr17xo3R6wBoRQKOyojEaIjCAswnGWKWKrUNjMj4LLIBLioDiA+Kg4hzLFfDKF3Dogc5nN7v0LZgM6MRiBYEKrrUOm9CMwwNtNr+gq8otzCewX2cpcLwsawo0PPIE++ycmb9jL+Kse4NmrRoKHBt+2fDLs5Dpnl4oAQsN92F2QiYXJbtbznEnjgrGAvlFnjQ5laNTSaJLjFxSEZ1d+J3Ifhvf14Yn33sA35C4mhjm2dxdCdSqpa0S11PPjS8/w4b4CfH3U5KbtpCxmBe//+z76u+XiqKMq14inIphW3axMgbdfAN711VTWlpCclERQYOuP73StNT69MFJPaZVo7hHDcP0OfqZavnrhXfQxS5iTENrcwTTmH+Wb9T9T7RVFXEQICI2U5OVSpvckafZykvqJL8dMxu6f+GVfJgG9YwkP8MRYX0ZmTgGKkIEsWDCPiNOdgKGabV+9x45SH0YNisbSUEFeSSPDLprHhCGhTkCrkMnsOs6mp4JgC6fm6pmNRMbMq+/lmi+/5f6bb+aeR5/lkbFtm0udCj2L2zJyU834TYxA7czFJ44bHnoaU+kOZu/YSfTK+3jpnmkIMjkeLr6g0IiBNOq3UlguMCC8NQbO7M/Xva4yi++//I5iZQgDokLxkBsoL8inuMZEwqzlTIwXvUMFCo9s5dtNh/CKiCMq2AtzQzW5OTkYvPswZ8kS+vmdrlMDe776kG35RuLj49FYqsgprCE6cS6zE5vWpE9XzgPEpW7nn9A0eHLRVJpIZchcjbxPtyNnhnb3lvpc3n36GTYXWvBV60g9ug/fGQ+w9rGVnG7SduStLy01VOYoCBge0NrxTK7ELyCQsrocjAoLE8eNJjDI5T651nzPaYoHURpPystqETd4udX9nlN57JmbObz+ZdZ8dACNnzcVp46Rre/LE/97k1kxLhqHfdZOXcvxiZvGqlE7ePeDz0i4c2WX4uDik++UdDZiUyXpR06SeXwP+w5korOAz6LZyFtPBNtg3khVqgFzpKy5w3cmLDqRQmVJCIvj23flV3l6I5N7YDC2Nt068/yt3BfuWsu3ef488sBkB6/iRm0l+adOsnXbGr7alg3yICZePJtRI0cwqEHcIyd+UiaqywvIPLGP7/5xN6n5etQBA5m3eArDE4LRnfbXEcGy6CnNTWfjh99y34FsAvolMnfufGLGukJSsN9e54qg7bTAEdx105X87/KXyDN7cg6XMVtkMNVTVilHE+Gi45PJ8FCq0JamU9OoZeboEShVziaRFlaePkE0GuuorjNDeNd8Yi3cz/zKpKumMDeHbbte55afDgM+JM5cQHJiPNHNISsFasvzycw4ztbnH2Fndg0y3xjmXzyTUSM1iMaklp+RqvxTbPv0c+7eloYmcjgL5s9g1bAZLSSnr8RQxGe4bCAOtM7kZ64pIvVIJscz97P3mC1a09Bko8M30i5fYz2Vx82woG2qzANHkJn7M6ivi3bTWMp3n39LTk0jtD0qsGMuI3HeSsb3a3tSYEfcxqUHvhpfPEt1VntPG0TdlGwiN+swWVnHSN25lyITKH1nYLG3SnShZIlzLmXDHY/zyYbJXD2j6wbeXfNFq6K57b8fsHL7s0yfew/H6iGpXyxhbfcrTtAIWM0z7XwbuUcyyFYOZ8gAB1uYEx/RbiO3fpwmk/NiU2vS30SKqZy17/xA8JzHiHMaPgcNnMxdf5/Mqm+fYfe2v1IXt5j/fvoGgx2mUmrGLLiOMQuuJY48bnxpI+MWPsrnb13W2rSjCWPZ3S8yf3oSU6/5gFvfepcrE7rYY1B8KUI9Nb7xTB/ej4/+/hhXzhvP/IFOlevql6eto9xoQNXmsA3yDx9D19CPIbHt11nj5Y0FGXqDqGG65hPriur6RCZw6+MJXHtwJLt+upIc5SyeW/shUx2MBHIGTVvNP6at5pvelSy691P6L7yTj9fe7mK07s/c2/+PuZdMYNyVz3LpI29yz8zW65ZdIfuZ8FBGJfGPr7/nwJtXMP3atdSgYGJcf/eVpqjo2+1G9Bw7VERdeCLxrvpkQxUHtm1gb6ELJyeXFVLhn7DgLJWmzGrhkZmM59RByqX4HSaqWXDz6yxYsYuFo2fzbW4dPgPHnjtrjFc8S2YP557//ZdJEx8lzqHf61DYNgm69IsuPFFIqXU934/42KGt1gHalAIV3r2VKNRtzTSNHMk8gHnIRQwIbF9kY2MtgmDEU9NFCLUtdI94UrX7Y37NCOLhp0a0KY9CoUYcBzeqfdoZ8FqI0Ni283tp/NtRHVBWb2LFPY9zxblQmEBRygZ+LI7hubdf4OrZy3jw6Q+Z+Na1uN452Wa1O/lAHLG1M2rDzInDuRQHjWRwdAcmlCY75JnOkJwFFxqrKawyEBIWiqoLlmfMChWinsxR+OLZjlVM7W2b8XgG+VsP6HKW6/R9VUEDS6++l1t6kMI8LRsYOXWovMkpK46hA2JaHnV05aHGu4+Ceoc1fLtMhhL2F6URMGIVka5w9B/Iw6+utctwPi6N1Oq11Hn5uD846ECsmtJ8dOpQwlo5nHSQsY3H2vw0Dtfb3OT6TBlIVAefUxts3EqOmzOLoDfu4/vtRdw+3dXIxi02DkSuXrUDgfs3Bg6fyMLq6Ozdl77D+rfb8Try9aJXvAKjrhKXjmemYlKPHWfEkKGEW2evZtK2fMOJcgdbkZWltr4WmcUDP++eZc13rG9X3TXyzac/oBw9i9HtLHargnzx8gSdNoM6F06KNmkUqJtMjnqDtm2PRGMhG49qmTDFMRpOV9UIQzG/rM8jeeoMRiQu4qm753Jo7eM8/2N2lxXhkpGnDwEeauRtmdHM1RzMTcF/1BD6djAea6ivRI65GU+X5XUisfr4t/ztvQ8oad3cO8GlhdTDy8cW19lSSnVd26HsfFQ2w7iuus71dymyNFXwc2oRYxIT3VsnbBHj/FyZStmVWWgzVfYZTGx/h2l1+zJ4+BMcZ6ChutblcMpYfpLs7CKSRgyzDSp0FRzc/AN5uvbZntunRoq1Nah9PVuvzZ9JwUItX33wDGtT2gjfdwY8s9MP0VgrOgKqSOof2zVytiWH30gWJniz6evvu8ybveuUpqmC9Nz9VpOAuncwQ/q3b8JyrKOcsIg+NNaVUevCHCJUZ5O735uB8YNsAFek8N32LCwuHATqa3JQe/emT9g52f7uKHZ339UcZv3uEoZMmtCuEVDp74U48TboC9C1tdRrLGVP5lFrjYz6Ktr67kv27aJeHcPwiLZWGm3mIZGRK/0jOgBZ3UhcPQTSN6+jNnIU0wbYBj1zbnyYy/uV8O8nH+PkudzDofIh1NeCoaANN/W6bDLT8hkzdLgtMoy5lj3ff0q6i50z2rpalHIv/Hy6YFoIaCsy2ZORQUN7E+FOtEWlxgt/kZelloZGFx+clVcDJ4/YgoSYTmnbVJqVGUcp14UxbHBbykjW4gB02n/ITtbTDmGubEzNz1y0leZ21NHQvCqHkyUZ1hJDh0bRP7R9S5WdaFZHwvCwXlSWl+Dqs6nPyaTiVG+GDBE3v0BVxi5+PljOuTgERSGXIyBi2UGbEnRU1hoYGtW7i2aaOtLTD5Be7XI64wiXW3dmcvZlU2uNZdGXITGdmPm7xd+ZyIMx05PIOLCeI120eb/LlKapKpPslDyrxOHBAxgQ1pnGCf3jhmHQZVF4ek+7Xd1lchNGix9+nr0QBAMbvtiE9+AZDApo3YCyUzMwBw6kX2d0tl1ZF9Jlyf6tHNQFcFGC6432p+vi6emNjz8IeoHqetfTlaKsQ2RWWazrVjp9neuZprmKHSeLiEkYQ3v6oL7epnhEhyznn66xzhoyTe7ClbLm6Oc8/P4+YqePa/ZWlAeM48FHb0O3fS3/eH3bOVynCaZPHNQVlLqsu74ki+LUUAYNibVWqT5jF5tSdfi6GJvVVGXhExRLpLvH9DiD5HQvV6jwUqtbrzE70bl7K/f2wm+g6DhdSK3VIax1Tm3BDvYUYDXjGuobXWIiOpGlHUshZOTkJgtQaz6gp77cIDppt9rqIlJrteIHb33YKrO+UVwLNCO4qLnFoEN8Kurh1q2shVVlznGKT9oOChgYFU9kW2O9lix2Vxri4uMoL0+jysUOKUFhxGQKJsDHH0Eo4ZvvjzDgojmEdKG5UdzSI27ULy7Jw1CUT3lmAeK2IXErmctf7SkKiwzEDxzs8nHnE2Wo1Z6oHbYfdZ5Lcw5LOdszcrDaN6JiiY5ta7DVnOOsL6JGTKRvZS7f7rXpp7Nl2GVKsy43mxNNEa3C+48nxtqZCOgaG2jU6Tvs7CKGjyJCX8b+7OrWdQoYwxW3J5Gz9S1eeOE1jqtHsWrOIBefUj3HDtYROyKZ3q31aWu+F3jKoV2bUAQNZHBE+5VVqJSIljbBaEbbOowSGGs4diCbCXOmWdc+axoycXWMZHXmcYoqepE4IswFcma01WXkHtvCvoPiyP44O3bsobi0GqNZQN9QTUnuUTbtS7Gayvbv3kluSTn1jXoay3PY8NFzXH7ZtXzy5QGKS+yiBVnMNMg80GPi7Sfu5PnPdpBXUoW+nYOATYZGGhp1tN7DZ6GxsQHXWT2JG9yP8spUXM01BbUMuUcAakFJQ0MJ6745SOz4GUS0MtWaOHmkEE3vYUS5UKgugDvvSTKlFypxuVIoRu/S9FDH3k25jB43zao0G0zl1LswPRhLj7I3O4ikJFdrRQJ1lRXkHNvGzkMVkJfLwR27Ka6qsUbyMtZXUpJ7jO07DwF5pBzcSU5xJQ16MyZdHcUFp9i+fRtQy8FDm0nLr0LbaEYw6akqLmDH1h2IObOOHmLPyQKqa12bIQpOZGELPhVAv4Gjbevi1j3lDej0rgeQ9i9kWOIIZEUZnHARYMd/5DxWXhHO9o/W8MJzHyAMm8u80c4Rzey5dfa6js9v+QNzpizmkQ/3M2niKKp3/pfL5iez5I6/kevCsl536igl5mjGjemhAdtr8skobpr5D+nPsBi19Z02NDRyrs5xl4XFMyqijiMbDojDs7P+tTdI6xTz3MzjHBIHPzJP+k4cglBRwPfffsLXv+6htMZA3NhFrLx0IaNjg1waVBR9xpAcrSZl8xEYM8mxbLkvSx9+iXFFFQgyDUHh4fiqXej72nT2lMoZf3myC4XqyLKn3YkOTJU1WsxilAEXP4Xah+AgP1q8sytJ2V5EeMjSjhfSlf6o1DLQN1KhFTsXx968On03aV5jmTIoH4sMdIZGdKL5xGHELJB2aD/B45cT6bLVaPlszR2s+aWK2KiFLFkOxz56lCWfDOeV957GY91/uO35HwiMi2DxsqVwdC3XrPiKax68A/2Bf/LZbg0e/SYwyaeI9Ru2c/mYS2zFl+/nlY9+Zvjk2UQHerJ/3XP8+EoMd7zwKBcPd3QNEhrL2fXzetZt+JkTRQZ6D5vM6stWkDwgGHTFfPPOf1m7+TghQ2fwpz9cxog+jueYDBuVhOqLnaSXQ5iTpUIdPZ1r7/qG9Rvf4JXMXgQPncvy8U57Ea3vrYSjqQKJ85Na7zl28V67JUmhwtPqjGyhRtv6uDZ9xj4yfKNIGoE1OInJrMUk2icdmo2FYzu2EzhyGq79oupZ9+y1vLxTR8SouSwZBsffv5+VXw7iP6+8jOynZ7nmxV1ER45iybKhGI+/w8rVn3PnCy8yvvITlj7wMX16B7N02SXIKjZz7SWbWP3kS1w3IpV7L3mInOjezFi+BA9FEY9fs4jBy+7nn39e7thksZCadYISEWTfUAaM6Ut1USrffvoxP+0+itakIWH6ClYuncWAENdT0LBR0xgmX8++fflMned4golC3Z8/v/QWBWX1eCi9CYkM7WLTrDczH/w/EnRmNL7eeKtViAPC+vo6BLW/y9n98b3HMPWZQlLf9gfS3dLugLq8NEoybN7E/XoFUXxoG+u/+ZDdJwpQ9xnFylVXMm1sLK669zOXOZQBg818mfErFZZLCHWhOjrD22X31xkGNloLOWm7rFpcpvEmqvEwj9x0LzsM0SybNZu+mT+y5omreeWtebzz+ccsG+bYWVl5yEJZviyRB775gTzTJPo4SSZX+xMd42IvlJ2w+Qc2U+0dzeyxrjozO8IedWkkZ+c6/r3m3/z3061ozaDSeKLARKPdJsmwBQ+x/8sniTqNS0MhBwuMKFfEtOvZaK2q2hcfLx8wGChynllY6tmzK5eRM6cSVf69NTJHaZ2eerEvtVOaQkUKhwpCuXihq1mFWIovK//yKotul+HpqUGhEDDqdOjNcnx9PGDZnXw5/1ZUnp4oFXLMJj06sTPw9kKY+A7LBRVq8dgjixGj4NFSdHACL6z9FQ+VyjpgMBn16A0yPH0cHb3M9Tm8/Jcb+fcuI5esupQ5fbJ47d/38vobX/DmO89g/vGfvLDHwupVMzn6wXMs+nIj73//DhPt9kVFjp5CvPcnbN+fz6Q5jh2kTB7I6ideZXZlLcg1BAT54eFijc6SfZDDhkAunzy8R7UyB2GUnnj7iz2HQE2j02qdpZHd+/PpM2Ae8b0a8PKFRu0RasUlLbsxiqX2BDuLvJi01PXhCeIe0GUPvMN8s9zWHmQCRn2jrT2IfC55gB9mmdFY24MMs1FPo8GCp48PSuGP/PzdZag1nqg85FjMBhobTKh9fFF6hPDcz+sR1Bo81R5gFiORNSJXebe0mebKVpGdvtfqxOMZ7Isi7Xtu//vrlEZMYvGUuWRsfpenb1vG/z6+mS8+fZbEXnYN/jSPoARWzQnlvc2bMMy7stW3pvINoZ9vV84uTxcs/ssJ6B3tFKEqgODQNr5BoZItB46QsOgqwly0TXvO3XVdlJZFTplYuozjP77F9Vn7mTR9KtNm9uHnt//Nslff49pn/8s/bmBj4GsAABt5SURBVJrRrmd35+SXER07hqy9hRTWQmgHuxY74n26C+6IroPn5aTtstmLFcZy3l/zNstvf4Dvbr+YAPHlNU6g6MA6/vPr9zz94vvMfuVGfF1o+6HzryTi66dZt7eUW5I7a+vW8su3x4mecR1D7T7uDgRvemxBW1NNndaAGEus8+1N3MgvxycwCB9NZyDVsev9J7j8+ucwRg9ieFIyhqoCDqTmWgMRjBw7CE+5EbMgp3diPF52mAkVJaQIehL6ujKVOtVaqcZD7QMWA3qto51Nm7mXdIUvK/qqUFV6WSPcmIxG9A6HThhJ2bkPrxFjiPBsCx0ZGm8/h8mI0kfV4lGp8SbAbqaiVCrRnB7cq+wHQ0rHjkmuxMenpTNTKk/PkuzqKGj55Knbee1oDP/7+lmmxtgUam/tDhY+toFbl8/CN2YuH3y1lnHe6dz58f+Rc+Bj3vzhHib+we5c1l6jWDFxIG9t2ErjnMtaB1VQaAgOsauEnQinLw9t/RWP6JFMindU6qef94h/uQZPTbi4uYdSraODh740nZM1emaNDEaVCyo1WExGdEbRjHS6AQpkHjqEd6/+xIW1vBvHusnQ+Pg7tgelXXtQ+xBkZ9q2tofmrbhiCM7TjUMcvClRN8OuwjfIbgO4hxKl2o7WXoiaU6TvsfbQyItSePnV9dz60Cu8uXyMdR1UPyOEQ9sXs2HTmzy/9hLeun26i/VRBZMvu4LX7v2WLcWXM9M5cLt9ed18rU1Zz9HqIdw+Y2Q3S9J28ccyD2Pzg5fhPXgKz73+EhcPsWmxJYm9mDXlel68/3YGD9/ITZPd6NvaLsrhSVRYbzyyD3KqQsco+47Igcq9m8708G1zrClgX6bNjdBk8iJ55d08e8fFdh1mAMGRYoiufEoPb6bAcCODmj8CO7YBw7n9qoG88NX7FCfeRbjdR2VH5fKy/MDn7CvrxQ33j3f5vP3EOrZ8tIZ3f0lD6X0GDheCGb3Bl6V33s+l49p3yrGXI+uHZ7n8ln/Sa/EjvPDkrUyIDcJQuJ87r17B6ymRPPHxJhb187COlJ1VVX35KWT6KuIC3Rg2aXxQeYuKqQi9vTlOMJJyLJve/WbZHD68YvDyAn19HWWiq2aTI4u5JpO9uRbGLBva7KBjX4/uvq7c/h7Pr6vknrXvNytMUaZeUV5WBVxVUUHSHau5qLecPa89zqsbcqwKoLzaef+Niml/uIJ3HvyE7QXLmdnZDWT6PNZvLGXMpTfRST+48wuhXIOPn2iNKSTbSWlmHTmCvP80okXrntoT30Bx6dNMpdWBzPZBWgylpKTW0mf+pQ5K8fxWouPS6nNzOJBnW7fUmkK5+S+PcO/yhOaM6oBggsI9IbeenJTd1DPdaVZnI/WJn8nNs77luw/XMf7OxW2eq9nMuDsuzLV89fUmYi+5mtHieKhH/mrJybCd/uPhF8+DzzzfrDBFcYPjExma4MWhncf54tNv+OPk65oGrmayUzZbozqZ1T4kTppFfKgrBdJ2pb1DwwkrqaC8vB5iO5fXmWuXKE1d/nGO1NoceAITlvPYXUtaFKZYYmMFVZX51rJFT9j2fkNm38DUvY/y9qc7ufWKZLcaaEP+Id79dDfJN9xNG4ert1ek1f8uNDqOhDGeKNTKzs80rac4eBHu7/7LMJTs5u4Hn8Zj1kN89c4jRDYN2FWRo3nyvut5e+HTfLfpOIv6jXApT31tNpiqCfJzZ1qtJMwa7MGItqRlDctYmUFqqYUZTaZIpUqNUmwRBgNaY8uiZvaRw5h8+zOozVlFB/Ce08cNfPz6axgGX8HyBPvIxWYKD57ebxrKxRNtnWV1da7NGcB7IAsuEl1IHX+e/adz84T1fPn2OpIeXNapdcm0n9ZS7JfAfTPbPnvRsbTuuvNA5WVrN2ar3bVJjoZT7MzQMmG1bQsFSm+UYjQFk4FGvWjGtSnNmvRdlMnjmd+nrVlmd9XLsdyirIOcMNhcP4Yvvo07L2tRmCKlsa6A2mrb9yCTteciomHWils48uSLfLppGFdNi+tZg0dBz6F1b5GinshNl1zkCEJPumvIJ3uXTQ8EhV7EtNHO1kQBQSFaERpIT99Nke46+lu7VAtVRal89MxjfHHEk1d3Hu+00vTyDSCIWqqqHS1tZwJPlyjNnLSD1FaIIzoZUy9diXOgGFNlCcViqEvASzUQO2tba5mVYSy/8Ta+3pBFrdaEj3cHIgomqsoKiLjoKpZOONMQXt6MnncFo+e1FufcpBjY8NIzfH0klrfW3t2sME+XFTQ0mRFyFdl7UjFeM8LFWg001BnAqCLYvw3T1Glm1n81fbxF25cOnd25RTl7diGPnkhMEwuZVwAe4nEDVVp0DWLjUlo3r6ekVDBg4aLW5kqHMrrpRtChiU3m2hELHcO8mUs4crjStik9LJmBTfuGE+Y/zF0F3+M3chGrRrsy/6iYfPmtHHroJT7+YTBXzx3iVgdZfuwHPthazpI7HiTKznrYFagE+nmh1GgIcOdVu1WgEo2vaPkBU23LbDt3xz504SOJ9W9Shppe+PqIUTEaybdGcBEHJTVs2VFA/0luBj13S55zQWQm88h+bDrTlwWrlxLh1JVUZZdRXiiW7YGfZ79227c8eDhXX385v6QVoTPF4u1qQftcVMMNnmZDHSWGcFZdsYh+zSZuNzK6QyILxEujROvrhkWrA366vAy2Z9tcfn2Hj6K3kyXRXJNHbblt8lVQ10CZlialqSRh7q3cvvcXDpUZGNzffnDcQaFNj2VWB8uGpsGfe3naonJqRm2RtZdu5tSBHMStWBDN2CGDWhFXZmWwt0hMlhMyYjThHTh2qcKHsmxFPEbRlbPDn4ygwTO5LMHpDXSYr/sILJUHePvT9UTOfYJlg1w4RckViGe11DadFehqPG9oMGIyyhC3PHf8U+KpFIdsgjWQuHV1qiGLLfkWkpfaDTQ8FHiKx0Y1GjE2uePXpv1KsWc8F0e7P4vuWJ6upPDn0jv+hsLTfl0UqMxlb421R8QrcRD9m2bJIcPn8tyaue0KIAsawB//fBmbMxrRGQW87c9ac5VTMFBdb2H80huZ3r/zJ+vUpf7M/U+9QXkbk53GklQO5+v5Y04mvm18O8Gxs7j/0WvdVNhygpoWCXUlddaN+6rGQnbn15I42e6wZ7kccY8oZgvVTYvc2pO7yLGEsjr2bIKKuwKxq9NqSN1TbLMqqBMZM9jRsUss7dSJTNLEMYMsiOgRQzuMTNNr6BSWxeoQ3OmWuro67fCTKXyZunQlalcdRTv5bI9q+PLf/+TjnZltUOo5dvAAhu1Xkx3p69LqJTp9LbnlEVZOarJQtMGpKCuF403tKO6iAdbtbfakVQWVlDVtpfQ8bfVqJqjj8PEyhJjJxHX+E0Pp54OXnxzr9oBmnmd2cfZKU6hmX1a+bbNqWH/6DmhtUM9M3cUpUT65hoSJ8S4W21sLLxcX/1snt06RKfA8FyE4WpfUZSm52zawNcPIksfmO5qxm0oQais4roM4r4A2MbBGRJG1Hy21RWAVAdaQaAIN+nqrO0fu/gMQMIjYQLtpkZcPsV6ebDJqqSkTzbN6du6tITZpYptytJTRTVcyBT7+rUfBtfkpFJTZ1tlHxUUR5lZjaqmDf3wSC/qbEVxEo2mhOn2loHfiHOKUbWi002Rt/Cu8/QgPD0duFrfxO/bICrWGWnM+qjoPQsL7EKwy4ngWgXjOpUBYWK9ObHeQE+TlbS2pIU9n3UNdlnGCamU/5jU5UVlF1fii8RF7qEoMtaIZ08jewwVEDJtGkMpRzjaq1n3JdQXsLLA5ATF0MH2bz7c7LZKeY+l7sc6zg4IYNdq9yDQemp43eJR7qM/i+/TAPziUsJBa8XBTx9Ynk6NES06mmsbgYGKjg2loMnefRhEsmC0BBPl3/IHlnzyExbo6F0hy/9ZLGAUnM5ujfkUH+BFmP59oLGRfYS5BY4dzJr5YFqMRk0GwRcNoEf6Mrs5eadYVcKrAZnsNigtnYKSzfaCWQ9t2W4WTe45g4lBbRJUzkvY3kUkg5cguik3DmTbEtdNQeWE6pUoT4wf1adM0qPJU4KF09/gtGWpb0F60ei1GUwOp+aUMHHcxqtMOkSK2ShV+SiUIOgSzDEPJQfLMXlwcf65c6s/dCy1My6SqVJyF+zKy7zB8zqCPV3i4qQRlCjRnNMq31d8rOomHnh2HxRYozQEUcXBUsOFvpH9VzgP/eIZ4L+d3bru3PwTcgUEbN55qT+uAzaBvQI9A+tHDRCf8CR/79qAQPVNFJWHCbLQgnk+ZV6jioov7NfvRtsG+25N1hSfILbfu0GTwyL5EOy/9605xeIfNKaVX2FjGxLUeeHV7Jc6LAN5Mv/wOpq4SozI5D9lAJpTx+N1ZVMx4gL9dPJDWhxCI7U8uZu3gV8/JHYW2yRWDGRTjvDRiITPvWFNgEQ/6Ro6hl93nZyhJpaCwhjFXj7S2PVGO02EWOyjY+tjcqMeoE1pCOrqTqQ2as1aadQUnyDtmc1vvGzqYPk5WMqpO8Otem5nMb/RERsWKCzMWdPUNyD29UbXs1m9DxPORbCLr8B6OZJaAux2lg1gCFos3w8cnExvW0VaDRvKytiLEziEiwjVtxv79oFAzLrm1qft0sd5+XggeRsqsHqAdm8p8Inyta6NmUwGnsrIo0w1gom2V/TRLUPgSrhRlqqaxJo/9R3PwGjCW8I7Mky0cuu1KrxP36nmitH68Rk4czaZAtEN79qHvcOe9hAKNWq11b5+6p9RNJpraXWt2D7k4A1A0HXgtxva1h9n53v5Z29fivl1xIG+QaSktOMZx/QBWDnJujxoG+orrRzoaMqrJPJmDKX5s0zpT27x7wpOCtKOU5Yv2bhnDeg8i0AEz0J48yNYjtk32vZOnEC8qVYuRhgYDKm9vl3twe0K9zpUMcnFZxtVP5oG19TWFxGytqNxtfw2UVYvDM2B4PNF9nNYlLRWkpeyxSaAIZuDUsQ6e2RWnTlKaF8HSAcHUV+aQlp6HURnKwGGxBKrbkN2uPoJMQEBp3SNul3xGl2etNEszcsiwBsL1pPfgsQQ7Nc7c3T+yK09c3JUzfUYysWoQcn/h5utfYfGaj1gU3/G0/oxq1qlMdWx8/z88/12K6yjjbvHy5/q/v8XtC+M7oDagq61B1TeYXi6nP2X8/MsBlEFzmTmi7dGvxisKs8KfSp173mCeXh7W7Rfa2iL279pH34lL8HZ6V8iUaKz7TE0UZh4lx+DHxFWdOa2mg6qfk8cWcrav5aFn3idq7n08ecM0lLIaUguOW9ez5JEhxPV3GtXmbeT6a9Yw/9m1XDbS3gZ0TgQ8a6ZmiwVBEGOQnjWrZgYKP41NMRhz2fFzGgNGjUNc8nH8KfAWLQ/I0BamsPdoMBMvaW1Wc8zTE+4ETh3Po1jESx5N7JBBTktCAgc2fcNR61hfzbw5yVYnssJNr/DnNcd55INXGeY8fugJ1eoOGQQzFkGcFJxt4xOsfMQq+PUPI8zXsfOxlBxl96+2AJZ+cZNYPtmxH806lkZOr97Un/qRD07IGTdlLEdefoi/ayfz5su3OsxKXcHUWFtNhTyUXoFOytoVcQdpZ600M04dQdz1hsqHmKQBTmYbA7vX76VIK4BqAOMSxljH0pn7NrKrSs1NvXqCwhSFD+CqJ9/ksscs1uPl3XGtccTVNtryULmz3qFG6elPoK8PAS5Mevq0LXyzP5Ppf3mt3Q/XPzgGhdKPoyXiifQdR0DyCVAjxiUoOLCFPVNXsizO2V4l1sgbn4E+8Gs9G9Z/ztB//I/onvKKHAFvvhPqT7Lmwbt5/9dSQtJ7c/X10xiozeVUyjErTXivEAY4RdA+uPFTDpR4cUuf32/P6B3qaw02kXf4CzZXLebpeFdLBRoC1L4oMLLzl49IWvwf4vxcNNrmt9FTLuo5kpNqPZlFFhFK3DAnJyBzMZvXHbYFofefwoRRtoHAge2bSZMNJ9R5hamnVOuClkODt7dtRtg30BfnJdCcfRvZXGnb1jT9+j9xUbi9UjWQkXeU2pKjZJTfwdO3zCNQpSRo3iCev/JdNv75ei4dbOeb4QInXXUtNZHBhESc/SD5LJVmA1mpe60iKrxDSOrnHN6phj3FhVbvPM2QoYxNFr2rGvjy8y30mXwHI2xe7y6qeL6TZNa1m/PTHXgydMgsqj4pptQA4t7qlp+BL/77MjkhV/HSVRNbkl1dBYcy1KyiLlNUmh3/VJpAVEooLFEwOmkIYija1j8VGn9bk6iQD2B6Qs9fy9TnHWf3dhsGgVExBMqh7tgONh601S4wYCDB9ksGdUf51yubGPGHV0jqolNIWuPY81OUGiWeKqit82Hg0IGEuVQUCrx81IjhNSqVkYwfHt+GAbln1VdoKCD3ZJOfRXAUw6OcOkptMb9W2GY1QTPGkihudtel8tUvWUy4+mFCXX4bPauOF540/iROGoDsi/3U1Dod4GEu5v23P6fGDEETruPBK6c7Vs9STPquAwSMuoqbr1pAUJMjhtziiUVbTo1W3O7YvtIsKMmhJtKf6ID26RwLdn3XyiDjmqyN1MZMDm+zHU6q6TWaoa02CXkR1cv2NYb3jmZcMGR99U/eOxLAzdcsOAuPrzbkuUCSx168jCnaTLYddjxP49gXT/HSLnjylScYF9LBeMYnhtERSvSp2U2L6+1X3t8/BoNFRtzCFVyc6BSNvDmrHF9fI8i9uOya1bRa4mqm6zkXCt8g/LzBI2o8jz9/F6HmEl589XPMgwZb10TKy0upbzIdmEqO8cztt5LiM5e7/jD5glAA5wppb18NFjX4JS9l9SS7bUdOBWqi5BjksGDl1ST16DBHLYI35KVybI/NzyIkciytdgGpg+jbNDgcHT+QKDn8+NyT7FNM4OZLHAMgtHCVrs4WgcRF1zAlzIPcLT+wNe10kJUq1v3rLp7+IpWwwXN4/tl7GRPiuEYplKWzNyWE+auuYlSvFpVVcDyXUstQ+vdxmHm4EFMgN/cEA8L7EH321lknU7+L4tpNMiuxWGyVT5g7n/5OAzrR3Lf8j9fxzc8pbN71E/fceRcZ+w+z4sG/c8ng369pzH/YEh64ZR9r/vUofW67gZFhen79+Rs2bsvmqvv/xfUTnMxJLl9CIKNm9+H9jSmUGaF3R9NkiwVZUF9WrFhNkGObdOAutwj4xkxl2YIRDuk99UYZlcB1dy/l5AfF7P7oaXbkHuGIbgLvvruYXf/9K0+8/T5/Wu1Fcl8N+Sd2kWceyL9eepKxwR0MSnpqhbtILoVMg1wTwPzLryS6nT5HqRIIiRjJissmtOnJ3UUidRkbmUyFwbrUr+SiRbNp5San7sMdt1zD7sNPsP/bd7mz/lcOHNNzz1N/ZXgr4i4T63fPyKv/DP7x8lPc89f/484rV7Fn+mDqcvfy4Td7SVh8E/c+cD8Lx7ZeJqhOO8p+7wDuH2fv0NfIQfGQkAkzGWinSF2DXEbGkSriRk5ot+9znbd16tn1HD6x3PD8GqL2NjBn9QL7QxCaS+o97ipef9uHz3/aR2GNniV3P81Ni8Y0P/99XqiZessDGII+IftUKpYygVpzEKv+eitzRrU1C2yN1IiESeg/3MiRMjO9I9vRhIDCO4KZl1zNFTPb34/mFzmDK/40hfEdzXRbi9M9KTI/lt7zKuqQN9hwrACPmIk8ec1NTIzxZnL828SP/Zw9adnUaRvpN/EKHrr6Ogb1mGWB7oFMLNVDHcboeX9i9bxh7QrhEzKEpdctZKbV671d0h7z0Kt/Mg++9TQzsyO5fNUYJz8LUUw5Iy59gLeV0Xy3J41qnQe3/f0Rlie1vzm/x1TwghVEzril9/JO5FA+WPcrxfUNqCKTePSlP3PJ4oXEuHKzAHKy9yIL8GZQVEvfKDScZPfWDMYtfriNowrtQCo/weEiHxJvHX2Ws8QmnoL061YEjHXlQmFJzRnJYCnbISxLGCM89Hl2h/lN9WVCRk5Fh3R1ZQVCQbW+QzqJ4PwhkP/zU8JFN/1ZSG/owjIbKoSM/FLB0AHLhvICIb+6vgMq6fFvFgFLifDo7dOFW7850U1VNAgf3ZgkxA+/Qjhu1y3V7n9diPOOFZ5cXyQIxhxh/ZYtQn4bzbR4w/PC+NErhM2lpi6pQ4uB2E4xS5fnDwEPn15EhLYxxOpADFlwAgvGenN82zabJ2A79ApvMaJHx7Ynn+BIIv3PfrG8HVGkR51EQO3jT5CvF3Jx32lX/TyDiI0KcRnX2L4Iz16RRPn/fpdS7LH4XV7LNAQEBuAlb9+Sde6wKedQqg7vIcnE2HVLh7eso7LfCGZNDqcu9TAZh/PEE+Rc/Mzs2biVkLEzGOu0VuqC2K0kSWm6BVNPJdIwd/UlVGb8wv7qzm+U6am1kuRyRCBw5Gr+c9ftdOjv4JhNupMQ6AIEfLnqlue5s4O4sl1QkGsWJensKssnavQgh4D61dVVhIZGEikzsCMlk7BRkwi1U6rNzGqPse6wlsmLZ7sMWdpM14kLSWl2AqyeSBo+aRUXB1ayfv3RniieJFMXIKDwDCI6LMwx5GEX8JVYSAh0jICMoJBown1daaSOc58thaGmFO+AfkxLGOzAauycqxkbkMdba14jV5PIrKTWDkRihtSfP6U6MJHl09r35XBg3sGNTDTydkAjPe7hCGT/+CJ/fb+Wp15/gHiNtMmsh78uSTwJAQkBNxEQTDqqqhvxCQhA5XAcm0B9TRl1DTICg4PRuAqHqcvh8bsfwXvpo9w9vesiWUkzTTdfXk8mi5l9JfMH5/PaxzusgSR6sqySbBICEgISAu4iIPPQEBQc6KQwxdwyfPxDiYgIca0wsXB43ZeUe0/jiqldpzDFkiWl6e7b68l0sgAu/8MVBJ38lvUptmATPVlcSTYJAQkBCYFzh4BAUdpG1u0ycOXNq87oKLH2ZJOUZnvoXEDPlFEXcf2qmchrKi4gqSVRJQQkBCQEuhgBwUxFiYHkqy5nXF934oF3rnxpTbNzeEnUEgISAhICEgI9GQHRTcfx/LwulVaaaXYpnBIzCQEJAQkBCYFuReAcKkyxXpLS7Na3KxUuISAhICEgIXAhISApzQvpbUmySghICEgISAh0KwKS0uxW+KXCJQQkBCQEJAQuJAQkpXkhvS1JVgkBCQEJAQmBbkVAUprdCr9UuISAhICEgITAhYSApDQvpLclySohICEgISAh0K0ISEqzW+GXCpcQkBCQEJAQuJAQkJTmhfS2JFklBCQEJAQkBLoVAUlpdiv8UuESAhICEgISAhcSApLSvJDeliSrhICEgISAhEC3IiApzW6FXypcQkBCQEJAQuBCQkBSmhfS25JklRCQEJAQkBDoVgQkpdmt8EuFSwhICEgISAhcSAj8P/BK6OqL7LzMAAAAAElFTkSuQmCC"
    }
   },
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![image.png](attachment:image.png)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# from https://keras.io/layers/recurrent/\n",
    "\n",
    "class MinimalRNNCell(keras.layers.Layer):\n",
    "\n",
    "    def __init__(self, units, **kwargs):\n",
    "        self.units = units\n",
    "        self.state_size = units\n",
    "        super(MinimalRNNCell, self).__init__(**kwargs)\n",
    "\n",
    "    def build(self, input_shape):\n",
    "        # Whx \n",
    "        self.kernel = self.add_weight(shape=(input_shape[-1], self.units),\n",
    "                                      initializer='uniform',\n",
    "                                      name='kernel')\n",
    "        # Whh\n",
    "        self.recurrent_kernel = self.add_weight(\n",
    "            shape=(self.units, self.units),\n",
    "            initializer='uniform',\n",
    "            name='recurrent_kernel')\n",
    "        # bh\n",
    "        self.bias = self.add_weight(\n",
    "            shape=(1, self.units),\n",
    "            initializer='uniform',\n",
    "            name='bias')\n",
    "        self.built = True\n",
    "\n",
    "    def call(self, inputs, states):\n",
    "        prev_output = states[0]\n",
    "        h = K.dot(inputs, self.kernel)\n",
    "        output = h + K.dot(prev_output, self.recurrent_kernel) + self.bias\n",
    "#         output = keras.activations.sigmoid(output)\n",
    "        return output, [output]\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "keras의 layer는 반드시 build와 call 메소드를 구현해야 합니다.  \n",
    "build 메소드 안에서 layer는 add_weight 메소드를 활용하여 이 layer의 파라미터 구성을 정의합니다.  \n",
    "call 메소드는 해당 layer의 실제 로직을 구현합니다. layer의 입력과 출력, build에서 정의한 파라미터들 사이에 어떤 연산을 거치게 되는지를 정의합니다. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tensor(\"input_1:0\", shape=(?, ?, 5), dtype=float32)\n"
     ]
    }
   ],
   "source": [
    "x = keras.Input((None, 5))\n",
    "print(x)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From C:\\Users\\uhmpp\\Anaconda3\\envs\\tf13\\lib\\site-packages\\tensorflow\\python\\framework\\op_def_library.py:263: colocate_with (from tensorflow.python.framework.ops) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Colocations handled automatically by placer.\n",
      "Tensor(\"rnn_1/TensorArrayReadV3:0\", shape=(?, 32), dtype=float32)\n"
     ]
    }
   ],
   "source": [
    "cell = MinimalRNNCell(32)\n",
    "layer = RNN(cell)\n",
    "h = layer(x)\n",
    "print(h)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "MinimalRNNCell을 사용한 가장 간단한 vanilla RNN을 구현해 보았습니다.  \n",
    "이 RNN layer를 활용하여 이전 장과 동일한 time series forcasting을 수행해 보겠습니다."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "from matplotlib.pylab import rcParams\n",
    "rcParams['figure.figsize'] = 15, 6    # matlab 차트의 기본 크기를 15,6으로 지정해 줍니다.\n",
    "from pandas.plotting import register_matplotlib_converters\n",
    "register_matplotlib_converters()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# csv 포맷의 데이터 경로를 지정합니다.\n",
    "dataset_filepath = 'datasets/airline-passengers.csv'\n",
    "# csv 파일을 읽어들여 DataFrame 객체를 생성합니다.\n",
    "df = pd.read_csv(dataset_filepath)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_data = [([df.iat[i+j,1] for j in range(5)], df.iat[i+5,1]) for i in range(115)]\n",
    "test_data = [([df.iat[i+j,1] for j in range(5)], df.iat[i+5,1]) for i in range(115, 139)]\n",
    "X_train = np.array([X for (X,y) in train_data])\n",
    "X_test = np.array([X for (X,y) in test_data])\n",
    "y_train = np.array([y for (X,y) in train_data])\n",
    "y_test = np.array([y for (X,y) in test_data])\n",
    "X_val = X_test\n",
    "y_val = y_test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "# RNN의 경우 Dense와 입출력 tensor shape가 다르다.\n",
    "X_train = np.reshape(X_train, (X_train.shape[0], X_train.shape[1], 1))\n",
    "X_val = np.reshape(X_test, (X_test.shape[0], X_test.shape[1], 1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "# to reproduce same training result\n",
    "from numpy.random import seed\n",
    "seed(1)\n",
    "from tensorflow import set_random_seed\n",
    "set_random_seed(2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "rnn_12 (RNN)                 (None, 50)                2600      \n",
      "_________________________________________________________________\n",
      "dense_11 (Dense)             (None, 1)                 51        \n",
      "=================================================================\n",
      "Total params: 2,651\n",
      "Trainable params: 2,651\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "cell = MinimalRNNCell(50)\n",
    "# cell.build((5, 1))\n",
    "rnn_layer = RNN(cell)\n",
    "# rnn_layer.build((5, 1))\n",
    "\n",
    "# define model\n",
    "model = Sequential()\n",
    "#model.add(Dense(100, activation='relu', input_dim=5))   \n",
    "model.add(rnn_layer)  # Dense Layer 대신 RNN Layer를 사용\n",
    "model.add(Dense(1))\n",
    "model.build((None, 5, 1))   # model의 input shape를 지정해 주어야 summary가 가능하다.\n",
    "model.summary()\n",
    "\n",
    "model.compile(optimizer='adam', loss='mse')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 115 samples, validate on 24 samples\n",
      "Epoch 1/1000\n",
      "115/115 [==============================] - 1s 5ms/step - loss: 68592.5199 - val_loss: 180959.7500\n",
      "Epoch 2/1000\n",
      "115/115 [==============================] - 0s 321us/step - loss: 58132.4001 - val_loss: 152895.4375\n",
      "Epoch 3/1000\n",
      "115/115 [==============================] - 0s 321us/step - loss: 48698.7080 - val_loss: 121581.4375\n",
      "Epoch 4/1000\n",
      "115/115 [==============================] - 0s 312us/step - loss: 37261.9125 - val_loss: 84556.1172\n",
      "Epoch 5/1000\n",
      "115/115 [==============================] - 0s 304us/step - loss: 23934.1561 - val_loss: 40845.9336\n",
      "Epoch 6/1000\n",
      "115/115 [==============================] - 0s 326us/step - loss: 8777.3802 - val_loss: 7364.8203\n",
      "Epoch 7/1000\n",
      "115/115 [==============================] - 0s 286us/step - loss: 3539.4134 - val_loss: 22303.2500\n",
      "Epoch 8/1000\n",
      "115/115 [==============================] - 0s 278us/step - loss: 6921.8547 - val_loss: 12179.7998\n",
      "Epoch 9/1000\n",
      "115/115 [==============================] - 0s 270us/step - loss: 3131.7274 - val_loss: 6697.6812\n",
      "Epoch 10/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 2405.6031 - val_loss: 10205.8643\n",
      "Epoch 11/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 3281.3100 - val_loss: 9323.0107\n",
      "Epoch 12/1000\n",
      "115/115 [==============================] - 0s 321us/step - loss: 2564.3533 - val_loss: 6324.6094\n",
      "Epoch 13/1000\n",
      "115/115 [==============================] - 0s 269us/step - loss: 1800.2301 - val_loss: 6540.2876\n",
      "Epoch 14/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 2067.2679 - val_loss: 7103.5117\n",
      "Epoch 15/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 2064.7730 - val_loss: 6118.5864\n",
      "Epoch 16/1000\n",
      "115/115 [==============================] - 0s 269us/step - loss: 1835.2359 - val_loss: 6109.1050\n",
      "Epoch 17/1000\n",
      "115/115 [==============================] - 0s 269us/step - loss: 1864.2667 - val_loss: 6214.3672\n",
      "Epoch 18/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 1832.6445 - val_loss: 6007.6431\n",
      "Epoch 19/1000\n",
      "115/115 [==============================] - 0s 295us/step - loss: 1775.7076 - val_loss: 5950.6304\n",
      "Epoch 20/1000\n",
      "115/115 [==============================] - 0s 269us/step - loss: 1812.3186 - val_loss: 6052.6328\n",
      "Epoch 21/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 1797.1588 - val_loss: 5899.6538\n",
      "Epoch 22/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 1745.7569 - val_loss: 5886.6665\n",
      "Epoch 23/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 1787.8682 - val_loss: 5938.8833\n",
      "Epoch 24/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 1753.6574 - val_loss: 5850.2847\n",
      "Epoch 25/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 1721.4698 - val_loss: 5853.1035\n",
      "Epoch 26/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 1779.3305 - val_loss: 5909.3237\n",
      "Epoch 27/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 1767.9087 - val_loss: 5789.4995\n",
      "Epoch 28/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 1724.4551 - val_loss: 5783.8794\n",
      "Epoch 29/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 1720.7278 - val_loss: 5761.6133\n",
      "Epoch 30/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 1710.5725 - val_loss: 5735.4648\n",
      "Epoch 31/1000\n",
      "115/115 [==============================] - 0s 269us/step - loss: 1710.5014 - val_loss: 5727.6919\n",
      "Epoch 32/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 1728.3690 - val_loss: 5736.7910\n",
      "Epoch 33/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 1694.1520 - val_loss: 5688.7344\n",
      "Epoch 34/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 1698.0172 - val_loss: 5730.2407\n",
      "Epoch 35/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 1713.1020 - val_loss: 5675.8384\n",
      "Epoch 36/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 1680.5711 - val_loss: 5633.4390\n",
      "Epoch 37/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 1694.4858 - val_loss: 5641.1519\n",
      "Epoch 38/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 1691.9106 - val_loss: 5586.5591\n",
      "Epoch 39/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 1665.4968 - val_loss: 5566.7915\n",
      "Epoch 40/1000\n",
      "115/115 [==============================] - 0s 295us/step - loss: 1659.5924 - val_loss: 5547.7891\n",
      "Epoch 41/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 1663.8358 - val_loss: 5530.1328\n",
      "Epoch 42/1000\n",
      "115/115 [==============================] - 0s 250us/step - loss: 1661.7015 - val_loss: 5512.6479\n",
      "Epoch 43/1000\n",
      "115/115 [==============================] - 0s 269us/step - loss: 1643.9380 - val_loss: 5496.8633\n",
      "Epoch 44/1000\n",
      "115/115 [==============================] - 0s 278us/step - loss: 1648.6989 - val_loss: 5516.2383\n",
      "Epoch 45/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 1660.3088 - val_loss: 5460.1489\n",
      "Epoch 46/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 1632.5876 - val_loss: 5458.0444\n",
      "Epoch 47/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 1642.6924 - val_loss: 5428.0352\n",
      "Epoch 48/1000\n",
      "115/115 [==============================] - 0s 286us/step - loss: 1616.3805 - val_loss: 5442.3960\n",
      "Epoch 49/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 1626.7778 - val_loss: 5429.4722\n",
      "Epoch 50/1000\n",
      "115/115 [==============================] - 0s 269us/step - loss: 1624.0364 - val_loss: 5359.2007\n",
      "Epoch 51/1000\n",
      "115/115 [==============================] - 0s 296us/step - loss: 1604.7267 - val_loss: 5338.8882\n",
      "Epoch 52/1000\n",
      "115/115 [==============================] - 0s 269us/step - loss: 1607.8120 - val_loss: 5319.0142\n",
      "Epoch 53/1000\n",
      "115/115 [==============================] - 0s 269us/step - loss: 1590.3621 - val_loss: 5298.9761\n",
      "Epoch 54/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 1610.8960 - val_loss: 5282.4429\n",
      "Epoch 55/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 1596.9101 - val_loss: 5287.0059\n",
      "Epoch 56/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 1577.9928 - val_loss: 5238.7104\n",
      "Epoch 57/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 1602.1218 - val_loss: 5277.4507\n",
      "Epoch 58/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 1586.5831 - val_loss: 5203.0679\n",
      "Epoch 59/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 1560.8106 - val_loss: 5183.8579\n",
      "Epoch 60/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 1583.4739 - val_loss: 5182.0703\n",
      "Epoch 61/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 1557.6099 - val_loss: 5161.3730\n",
      "Epoch 62/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 1558.9794 - val_loss: 5134.3682\n",
      "Epoch 63/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 1555.4878 - val_loss: 5112.3618\n",
      "Epoch 64/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 1542.9023 - val_loss: 5083.5933\n",
      "Epoch 65/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 1531.8942 - val_loss: 5072.4556\n",
      "Epoch 66/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 1529.4576 - val_loss: 5049.1187\n",
      "Epoch 67/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 1516.0095 - val_loss: 5048.6904\n",
      "Epoch 68/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 1541.9402 - val_loss: 5037.4360\n",
      "Epoch 69/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 1491.0259 - val_loss: 5022.5957\n",
      "Epoch 70/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 1519.4235 - val_loss: 5092.7383\n",
      "Epoch 71/1000\n",
      "115/115 [==============================] - 0s 217us/step - loss: 1538.2155 - val_loss: 4951.1938\n",
      "Epoch 72/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 1489.6364 - val_loss: 4923.6558\n",
      "Epoch 73/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 1490.0308 - val_loss: 4909.9077\n",
      "Epoch 74/1000\n",
      "115/115 [==============================] - 0s 278us/step - loss: 1487.3935 - val_loss: 4902.1738\n",
      "Epoch 75/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 1486.9294 - val_loss: 4866.9819\n",
      "Epoch 76/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 1468.2753 - val_loss: 4848.4287\n",
      "Epoch 77/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 1459.6174 - val_loss: 4839.1187\n",
      "Epoch 78/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 1460.5518 - val_loss: 4811.8657\n",
      "Epoch 79/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 1457.5212 - val_loss: 4795.7881\n",
      "Epoch 80/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 1446.1968 - val_loss: 4789.5547\n",
      "Epoch 81/1000\n",
      "115/115 [==============================] - 0s 217us/step - loss: 1529.9214 - val_loss: 4825.2446\n",
      "Epoch 82/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 1427.7400 - val_loss: 4804.5176\n",
      "Epoch 83/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 1496.1190 - val_loss: 4786.2939\n",
      "Epoch 84/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 1423.8198 - val_loss: 4746.5195\n",
      "Epoch 85/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 1456.0858 - val_loss: 4770.4170\n",
      "Epoch 86/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 1440.6564 - val_loss: 4670.8218\n",
      "Epoch 87/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 1414.5688 - val_loss: 4660.8970\n",
      "Epoch 88/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 1413.5816 - val_loss: 4645.9409\n",
      "Epoch 89/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 1401.6834 - val_loss: 4612.0005\n",
      "Epoch 90/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 1412.2429 - val_loss: 4603.6689\n",
      "Epoch 91/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 1394.2488 - val_loss: 4596.3931\n",
      "Epoch 92/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 1387.7228 - val_loss: 4556.6406\n",
      "Epoch 93/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 1402.1256 - val_loss: 4592.8203\n",
      "Epoch 94/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 1420.3272 - val_loss: 4540.8105\n",
      "Epoch 95/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 1354.1629 - val_loss: 4608.7573\n",
      "Epoch 96/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 1447.1024 - val_loss: 4526.9888\n",
      "Epoch 97/1000\n",
      "115/115 [==============================] - 0s 232us/step - loss: 1394.1086 - val_loss: 4691.5078\n",
      "Epoch 98/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 1418.7174 - val_loss: 4466.7827\n",
      "Epoch 99/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 1401.1699 - val_loss: 4601.7944\n",
      "Epoch 100/1000\n",
      "115/115 [==============================] - 0s 248us/step - loss: 1370.5058 - val_loss: 4422.0786\n",
      "Epoch 101/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 1380.3852 - val_loss: 4522.1011\n",
      "Epoch 102/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 1348.8988 - val_loss: 4374.0142\n",
      "Epoch 103/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 1387.5540 - val_loss: 4471.5005\n",
      "Epoch 104/1000\n",
      "115/115 [==============================] - 0s 231us/step - loss: 1329.8015 - val_loss: 4367.0327\n",
      "Epoch 105/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 1355.6007 - val_loss: 4430.1860\n",
      "Epoch 106/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 1309.4955 - val_loss: 4317.1567\n",
      "Epoch 107/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 1336.6425 - val_loss: 4379.4019\n",
      "Epoch 108/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 1325.6614 - val_loss: 4285.8091\n",
      "Epoch 109/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 1300.9497 - val_loss: 4293.0063\n",
      "Epoch 110/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 1302.9803 - val_loss: 4228.9438\n",
      "Epoch 111/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 1285.2579 - val_loss: 4211.3491\n",
      "Epoch 112/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 1281.4245 - val_loss: 4203.0913\n",
      "Epoch 113/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 1293.3659 - val_loss: 4258.4546\n",
      "Epoch 114/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 1272.7185 - val_loss: 4177.3667\n",
      "Epoch 115/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 1274.2353 - val_loss: 4167.8877\n",
      "Epoch 116/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 1304.2326 - val_loss: 4126.4121\n",
      "Epoch 117/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 1284.1841 - val_loss: 4304.2886\n",
      "Epoch 118/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 1279.2045 - val_loss: 4087.0813\n",
      "Epoch 119/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 1234.7748 - val_loss: 4130.8403\n",
      "Epoch 120/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 1265.0416 - val_loss: 4064.5408\n",
      "Epoch 121/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 1269.4178 - val_loss: 4118.3047\n",
      "Epoch 122/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 1245.1632 - val_loss: 4013.0237\n",
      "Epoch 123/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 1219.0549 - val_loss: 4014.6409\n",
      "Epoch 124/1000\n",
      "115/115 [==============================] - 0s 312us/step - loss: 1222.2996 - val_loss: 3975.8652\n",
      "Epoch 125/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 1231.5084 - val_loss: 3960.4866\n",
      "Epoch 126/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 1244.8361 - val_loss: 4058.5469\n",
      "Epoch 127/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 1248.4272 - val_loss: 3927.5730\n",
      "Epoch 128/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 1229.8899 - val_loss: 3899.7605\n",
      "Epoch 129/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 1217.8796 - val_loss: 3920.7004\n",
      "Epoch 130/1000\n",
      "115/115 [==============================] - 0s 286us/step - loss: 1202.9251 - val_loss: 3912.5940\n",
      "Epoch 131/1000\n",
      "115/115 [==============================] - 0s 321us/step - loss: 1197.6727 - val_loss: 3855.5178\n",
      "Epoch 132/1000\n",
      "115/115 [==============================] - 0s 304us/step - loss: 1177.8149 - val_loss: 3818.5527\n",
      "Epoch 133/1000\n",
      "115/115 [==============================] - 0s 278us/step - loss: 1166.3122 - val_loss: 3800.1829\n",
      "Epoch 134/1000\n",
      "115/115 [==============================] - 0s 304us/step - loss: 1167.5217 - val_loss: 3793.7737\n",
      "Epoch 135/1000\n",
      "115/115 [==============================] - 0s 269us/step - loss: 1194.7632 - val_loss: 3766.9524\n",
      "Epoch 136/1000\n",
      "115/115 [==============================] - 0s 269us/step - loss: 1174.2736 - val_loss: 3813.2336\n",
      "Epoch 137/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 1154.6552 - val_loss: 3711.1523\n",
      "Epoch 138/1000\n",
      "115/115 [==============================] - 0s 266us/step - loss: 1143.9578 - val_loss: 3696.1418\n",
      "Epoch 139/1000\n",
      "115/115 [==============================] - 0s 286us/step - loss: 1165.0301 - val_loss: 3708.7336\n",
      "Epoch 140/1000\n",
      "115/115 [==============================] - 0s 295us/step - loss: 1123.6327 - val_loss: 3642.9939\n",
      "Epoch 141/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 1139.9170 - val_loss: 3680.6116\n",
      "Epoch 142/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 1120.4912 - val_loss: 3614.6816\n",
      "Epoch 143/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 1106.0457 - val_loss: 3605.4463\n",
      "Epoch 144/1000\n",
      "115/115 [==============================] - 0s 248us/step - loss: 1100.0978 - val_loss: 3548.9421\n",
      "Epoch 145/1000\n",
      "115/115 [==============================] - 0s 269us/step - loss: 1091.2102 - val_loss: 3523.8845\n",
      "Epoch 146/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 1082.4686 - val_loss: 3504.2012\n",
      "Epoch 147/1000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "115/115 [==============================] - 0s 251us/step - loss: 1076.8282 - val_loss: 3478.0969\n",
      "Epoch 148/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 1068.4695 - val_loss: 3447.0762\n",
      "Epoch 149/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 1095.8298 - val_loss: 3423.0989\n",
      "Epoch 150/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 1049.9824 - val_loss: 3549.4172\n",
      "Epoch 151/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 1077.6367 - val_loss: 3383.0508\n",
      "Epoch 152/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 1035.5247 - val_loss: 3354.5300\n",
      "Epoch 153/1000\n",
      "115/115 [==============================] - 0s 217us/step - loss: 1081.6348 - val_loss: 3433.6902\n",
      "Epoch 154/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 1023.7942 - val_loss: 3385.6345\n",
      "Epoch 155/1000\n",
      "115/115 [==============================] - 0s 217us/step - loss: 1080.1172 - val_loss: 3461.0938\n",
      "Epoch 156/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 1025.5640 - val_loss: 3266.6980\n",
      "Epoch 157/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 1006.5477 - val_loss: 3294.2585\n",
      "Epoch 158/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 1017.3860 - val_loss: 3211.4036\n",
      "Epoch 159/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 1027.4103 - val_loss: 3404.2307\n",
      "Epoch 160/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 1012.1509 - val_loss: 3172.6921\n",
      "Epoch 161/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 1018.4503 - val_loss: 3289.3496\n",
      "Epoch 162/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 995.4656 - val_loss: 3143.0820\n",
      "Epoch 163/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 974.8610 - val_loss: 3241.6787\n",
      "Epoch 164/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 994.3723 - val_loss: 3098.5750\n",
      "Epoch 165/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 977.0736 - val_loss: 3097.9736\n",
      "Epoch 166/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 934.3449 - val_loss: 3123.0203\n",
      "Epoch 167/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 986.6407 - val_loss: 3142.9246\n",
      "Epoch 168/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 944.3227 - val_loss: 3056.1379\n",
      "Epoch 169/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 978.4919 - val_loss: 3066.0247\n",
      "Epoch 170/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 937.6324 - val_loss: 3016.1602\n",
      "Epoch 171/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 955.0762 - val_loss: 3124.8181\n",
      "Epoch 172/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 941.4312 - val_loss: 2995.3730\n",
      "Epoch 173/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 954.8541 - val_loss: 3042.7170\n",
      "Epoch 174/1000\n",
      "115/115 [==============================] - 0s 217us/step - loss: 919.0720 - val_loss: 3011.0203\n",
      "Epoch 175/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 945.0706 - val_loss: 3048.5095\n",
      "Epoch 176/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 946.2282 - val_loss: 2958.0984\n",
      "Epoch 177/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 939.3682 - val_loss: 2961.0808\n",
      "Epoch 178/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 911.3805 - val_loss: 2959.6497\n",
      "Epoch 179/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 913.3130 - val_loss: 2930.3008\n",
      "Epoch 180/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 898.7049 - val_loss: 2968.7815\n",
      "Epoch 181/1000\n",
      "115/115 [==============================] - 0s 286us/step - loss: 911.9114 - val_loss: 2923.3586\n",
      "Epoch 182/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 895.2552 - val_loss: 2904.8684\n",
      "Epoch 183/1000\n",
      "115/115 [==============================] - 0s 239us/step - loss: 905.1249 - val_loss: 2935.8889\n",
      "Epoch 184/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 909.9877 - val_loss: 2888.1770\n",
      "Epoch 185/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 906.9942 - val_loss: 2881.2258\n",
      "Epoch 186/1000\n",
      "115/115 [==============================] - 0s 239us/step - loss: 894.4381 - val_loss: 2880.7952\n",
      "Epoch 187/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 894.1308 - val_loss: 2864.2302\n",
      "Epoch 188/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 902.0787 - val_loss: 2921.1453\n",
      "Epoch 189/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 891.4411 - val_loss: 2851.8699\n",
      "Epoch 190/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 891.3516 - val_loss: 2853.1956\n",
      "Epoch 191/1000\n",
      "115/115 [==============================] - 0s 226us/step - loss: 886.9732 - val_loss: 2846.4727\n",
      "Epoch 192/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 899.6330 - val_loss: 2842.9319\n",
      "Epoch 193/1000\n",
      "115/115 [==============================] - 0s 269us/step - loss: 881.0024 - val_loss: 2826.1350\n",
      "Epoch 194/1000\n",
      "115/115 [==============================] - 0s 278us/step - loss: 885.7870 - val_loss: 2821.7441\n",
      "Epoch 195/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 883.9385 - val_loss: 2823.9817\n",
      "Epoch 196/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 877.9631 - val_loss: 2811.7546\n",
      "Epoch 197/1000\n",
      "115/115 [==============================] - 0s 224us/step - loss: 885.2806 - val_loss: 2809.5789\n",
      "Epoch 198/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 864.5594 - val_loss: 2806.1973\n",
      "Epoch 199/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 898.1379 - val_loss: 2808.9504\n",
      "Epoch 200/1000\n",
      "115/115 [==============================] - 0s 248us/step - loss: 865.7862 - val_loss: 2791.6680\n",
      "Epoch 201/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 870.6826 - val_loss: 2821.4021\n",
      "Epoch 202/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 870.7540 - val_loss: 2771.2317\n",
      "Epoch 203/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 861.4793 - val_loss: 2771.2913\n",
      "Epoch 204/1000\n",
      "115/115 [==============================] - 0s 248us/step - loss: 864.3650 - val_loss: 2760.2136\n",
      "Epoch 205/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 855.0198 - val_loss: 2764.9607\n",
      "Epoch 206/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 863.1075 - val_loss: 2779.0615\n",
      "Epoch 207/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 858.8787 - val_loss: 2740.6575\n",
      "Epoch 208/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 853.7985 - val_loss: 2740.0117\n",
      "Epoch 209/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 864.5805 - val_loss: 2729.6191\n",
      "Epoch 210/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 853.2120 - val_loss: 2724.9426\n",
      "Epoch 211/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 848.9000 - val_loss: 2721.6174\n",
      "Epoch 212/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 858.3828 - val_loss: 2718.6245\n",
      "Epoch 213/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 856.7189 - val_loss: 2711.9128\n",
      "Epoch 214/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 845.4054 - val_loss: 2710.6631\n",
      "Epoch 215/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 851.3178 - val_loss: 2728.2590\n",
      "Epoch 216/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 843.9447 - val_loss: 2693.5461\n",
      "Epoch 217/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 854.4904 - val_loss: 2702.7959\n",
      "Epoch 218/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 843.9804 - val_loss: 2697.7344\n",
      "Epoch 219/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 845.1426 - val_loss: 2695.1926\n",
      "Epoch 220/1000\n",
      "115/115 [==============================] - 0s 278us/step - loss: 834.4229 - val_loss: 2673.3298\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 221/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 846.9499 - val_loss: 2693.2910\n",
      "Epoch 222/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 840.2560 - val_loss: 2662.1997\n",
      "Epoch 223/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 840.0604 - val_loss: 2729.8152\n",
      "Epoch 224/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 848.3500 - val_loss: 2665.5110\n",
      "Epoch 225/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 825.3600 - val_loss: 2653.7107\n",
      "Epoch 226/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 833.0031 - val_loss: 2657.5916\n",
      "Epoch 227/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 834.3919 - val_loss: 2639.5645\n",
      "Epoch 228/1000\n",
      "115/115 [==============================] - 0s 269us/step - loss: 842.0837 - val_loss: 2663.4363\n",
      "Epoch 229/1000\n",
      "115/115 [==============================] - 0s 304us/step - loss: 843.6108 - val_loss: 2630.8521\n",
      "Epoch 230/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 826.5254 - val_loss: 2627.0193\n",
      "Epoch 231/1000\n",
      "115/115 [==============================] - 0s 248us/step - loss: 826.6885 - val_loss: 2639.7893\n",
      "Epoch 232/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 839.5172 - val_loss: 2621.5024\n",
      "Epoch 233/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 829.6431 - val_loss: 2630.2947\n",
      "Epoch 234/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 845.7360 - val_loss: 2613.4177\n",
      "Epoch 235/1000\n",
      "115/115 [==============================] - 0s 217us/step - loss: 806.3669 - val_loss: 2657.7468\n",
      "Epoch 236/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 860.2963 - val_loss: 2699.3660\n",
      "Epoch 237/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 831.9451 - val_loss: 2595.4966\n",
      "Epoch 238/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 822.5705 - val_loss: 2603.6926\n",
      "Epoch 239/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 822.2622 - val_loss: 2598.0457\n",
      "Epoch 240/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 812.9608 - val_loss: 2590.7947\n",
      "Epoch 241/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 811.2559 - val_loss: 2580.6931\n",
      "Epoch 242/1000\n",
      "115/115 [==============================] - 0s 224us/step - loss: 828.5596 - val_loss: 2574.3108\n",
      "Epoch 243/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 809.1810 - val_loss: 2594.9363\n",
      "Epoch 244/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 816.4135 - val_loss: 2600.6516\n",
      "Epoch 245/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 809.9427 - val_loss: 2559.8455\n",
      "Epoch 246/1000\n",
      "115/115 [==============================] - 0s 231us/step - loss: 807.9869 - val_loss: 2566.5237\n",
      "Epoch 247/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 815.0433 - val_loss: 2558.4004\n",
      "Epoch 248/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 804.1239 - val_loss: 2552.8672\n",
      "Epoch 249/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 811.0204 - val_loss: 2545.9370\n",
      "Epoch 250/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 798.3261 - val_loss: 2540.6350\n",
      "Epoch 251/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 805.7177 - val_loss: 2551.0857\n",
      "Epoch 252/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 811.0045 - val_loss: 2527.8984\n",
      "Epoch 253/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 820.6648 - val_loss: 2523.5774\n",
      "Epoch 254/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 795.3247 - val_loss: 2537.3762\n",
      "Epoch 255/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 802.8861 - val_loss: 2545.5461\n",
      "Epoch 256/1000\n",
      "115/115 [==============================] - 0s 269us/step - loss: 801.1161 - val_loss: 2513.7722\n",
      "Epoch 257/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 796.1501 - val_loss: 2505.3147\n",
      "Epoch 258/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 798.3105 - val_loss: 2508.5574\n",
      "Epoch 259/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 790.2402 - val_loss: 2499.1619\n",
      "Epoch 260/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 797.2111 - val_loss: 2493.5576\n",
      "Epoch 261/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 787.7346 - val_loss: 2492.5383\n",
      "Epoch 262/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 788.3405 - val_loss: 2490.1201\n",
      "Epoch 263/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 808.6016 - val_loss: 2482.1042\n",
      "Epoch 264/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 799.6661 - val_loss: 2498.8579\n",
      "Epoch 265/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 787.9634 - val_loss: 2481.0791\n",
      "Epoch 266/1000\n",
      "115/115 [==============================] - 0s 269us/step - loss: 795.7208 - val_loss: 2471.9836\n",
      "Epoch 267/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 781.1996 - val_loss: 2473.6345\n",
      "Epoch 268/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 782.9095 - val_loss: 2494.0129\n",
      "Epoch 269/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 788.7829 - val_loss: 2483.5261\n",
      "Epoch 270/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 787.2389 - val_loss: 2466.8718\n",
      "Epoch 271/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 784.2807 - val_loss: 2452.2131\n",
      "Epoch 272/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 810.5678 - val_loss: 2499.2463\n",
      "Epoch 273/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 784.8463 - val_loss: 2445.7971\n",
      "Epoch 274/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 783.6668 - val_loss: 2447.4854\n",
      "Epoch 275/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 793.1429 - val_loss: 2474.2795\n",
      "Epoch 276/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 779.3396 - val_loss: 2440.0439\n",
      "Epoch 277/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 785.7834 - val_loss: 2441.4890\n",
      "Epoch 278/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 777.4983 - val_loss: 2430.2869\n",
      "Epoch 279/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 772.3168 - val_loss: 2481.7454\n",
      "Epoch 280/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 799.4870 - val_loss: 2452.4045\n",
      "Epoch 281/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 784.2989 - val_loss: 2459.9739\n",
      "Epoch 282/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 787.2432 - val_loss: 2416.6655\n",
      "Epoch 283/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 767.9145 - val_loss: 2422.2314\n",
      "Epoch 284/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 771.3338 - val_loss: 2435.0613\n",
      "Epoch 285/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 779.6135 - val_loss: 2421.0771\n",
      "Epoch 286/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 763.7327 - val_loss: 2408.4417\n",
      "Epoch 287/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 783.8367 - val_loss: 2412.9543\n",
      "Epoch 288/1000\n",
      "115/115 [==============================] - 0s 232us/step - loss: 771.6055 - val_loss: 2398.0271\n",
      "Epoch 289/1000\n",
      "115/115 [==============================] - 0s 278us/step - loss: 774.1116 - val_loss: 2432.1746\n",
      "Epoch 290/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 767.9886 - val_loss: 2392.9165\n",
      "Epoch 291/1000\n",
      "115/115 [==============================] - 0s 248us/step - loss: 780.5343 - val_loss: 2395.0330\n",
      "Epoch 292/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 763.3013 - val_loss: 2390.2976\n",
      "Epoch 293/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 765.3648 - val_loss: 2413.9624\n",
      "Epoch 294/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 764.1009 - val_loss: 2386.9255\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 295/1000\n",
      "115/115 [==============================] - 0s 236us/step - loss: 763.9308 - val_loss: 2378.9407\n",
      "Epoch 296/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 766.0498 - val_loss: 2374.6672\n",
      "Epoch 297/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 761.7018 - val_loss: 2383.8560\n",
      "Epoch 298/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 763.4985 - val_loss: 2368.5315\n",
      "Epoch 299/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 756.6524 - val_loss: 2367.7695\n",
      "Epoch 300/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 755.9994 - val_loss: 2366.2800\n",
      "Epoch 301/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 756.0092 - val_loss: 2366.2278\n",
      "Epoch 302/1000\n",
      "115/115 [==============================] - 0s 256us/step - loss: 774.4814 - val_loss: 2356.4043\n",
      "Epoch 303/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 756.3517 - val_loss: 2364.9402\n",
      "Epoch 304/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 764.0061 - val_loss: 2416.3169\n",
      "Epoch 305/1000\n",
      "115/115 [==============================] - 0s 217us/step - loss: 766.9532 - val_loss: 2348.9363\n",
      "Epoch 306/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 755.2665 - val_loss: 2352.7444\n",
      "Epoch 307/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 754.5949 - val_loss: 2352.6445\n",
      "Epoch 308/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 757.3100 - val_loss: 2358.2627\n",
      "Epoch 309/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 754.3813 - val_loss: 2341.6189\n",
      "Epoch 310/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 766.1303 - val_loss: 2339.9358\n",
      "Epoch 311/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 757.9314 - val_loss: 2401.2927\n",
      "Epoch 312/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 754.9038 - val_loss: 2333.5366\n",
      "Epoch 313/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 764.9911 - val_loss: 2345.1482\n",
      "Epoch 314/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 751.8708 - val_loss: 2337.4324\n",
      "Epoch 315/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 749.4798 - val_loss: 2348.7517\n",
      "Epoch 316/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 760.5416 - val_loss: 2382.3269\n",
      "Epoch 317/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 756.3101 - val_loss: 2325.0859\n",
      "Epoch 318/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 765.4656 - val_loss: 2332.1699\n",
      "Epoch 319/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 732.8774 - val_loss: 2377.0994\n",
      "Epoch 320/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 760.2405 - val_loss: 2396.2302\n",
      "Epoch 321/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 757.4564 - val_loss: 2317.6836\n",
      "Epoch 322/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 747.4739 - val_loss: 2317.1653\n",
      "Epoch 323/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 742.5801 - val_loss: 2319.7795\n",
      "Epoch 324/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 743.4698 - val_loss: 2334.3904\n",
      "Epoch 325/1000\n",
      "115/115 [==============================] - 0s 278us/step - loss: 742.7059 - val_loss: 2305.3284\n",
      "Epoch 326/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 741.2651 - val_loss: 2308.2771\n",
      "Epoch 327/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 755.4106 - val_loss: 2302.6238\n",
      "Epoch 328/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 741.2742 - val_loss: 2362.6724\n",
      "Epoch 329/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 750.2483 - val_loss: 2322.5327\n",
      "Epoch 330/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 733.1418 - val_loss: 2295.1448\n",
      "Epoch 331/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 748.7820 - val_loss: 2294.3135\n",
      "Epoch 332/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 742.1989 - val_loss: 2290.0947\n",
      "Epoch 333/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 740.4431 - val_loss: 2315.8640\n",
      "Epoch 334/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 737.2194 - val_loss: 2281.0989\n",
      "Epoch 335/1000\n",
      "115/115 [==============================] - 0s 278us/step - loss: 745.0843 - val_loss: 2281.5022\n",
      "Epoch 336/1000\n",
      "115/115 [==============================] - 0s 269us/step - loss: 747.7485 - val_loss: 2335.1689\n",
      "Epoch 337/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 742.3973 - val_loss: 2274.6228\n",
      "Epoch 338/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 735.1465 - val_loss: 2272.8669\n",
      "Epoch 339/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 736.3202 - val_loss: 2282.4316\n",
      "Epoch 340/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 731.8827 - val_loss: 2268.9392\n",
      "Epoch 341/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 743.3754 - val_loss: 2276.1360\n",
      "Epoch 342/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 728.7953 - val_loss: 2267.4133\n",
      "Epoch 343/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 733.3782 - val_loss: 2265.6704\n",
      "Epoch 344/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 731.8527 - val_loss: 2266.1340\n",
      "Epoch 345/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 727.6797 - val_loss: 2264.5508\n",
      "Epoch 346/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 739.6517 - val_loss: 2259.6008\n",
      "Epoch 347/1000\n",
      "115/115 [==============================] - 0s 269us/step - loss: 727.6848 - val_loss: 2264.1829\n",
      "Epoch 348/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 727.7018 - val_loss: 2261.1152\n",
      "Epoch 349/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 730.1621 - val_loss: 2267.2878\n",
      "Epoch 350/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 737.0023 - val_loss: 2254.9580\n",
      "Epoch 351/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 724.3000 - val_loss: 2258.5085\n",
      "Epoch 352/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 728.5792 - val_loss: 2306.1882\n",
      "Epoch 353/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 738.3644 - val_loss: 2247.7327\n",
      "Epoch 354/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 726.8158 - val_loss: 2243.5022\n",
      "Epoch 355/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 726.0380 - val_loss: 2243.2329\n",
      "Epoch 356/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 726.5641 - val_loss: 2241.8440\n",
      "Epoch 357/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 769.4889 - val_loss: 2350.2034\n",
      "Epoch 358/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 780.8398 - val_loss: 2259.9534\n",
      "Epoch 359/1000\n",
      "115/115 [==============================] - 0s 269us/step - loss: 736.4682 - val_loss: 2238.3560\n",
      "Epoch 360/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 738.0377 - val_loss: 2312.5103\n",
      "Epoch 361/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 724.7106 - val_loss: 2230.6790\n",
      "Epoch 362/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 722.3908 - val_loss: 2247.7346\n",
      "Epoch 363/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 751.4824 - val_loss: 2228.9404\n",
      "Epoch 364/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 724.1838 - val_loss: 2346.3093\n",
      "Epoch 365/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 752.4990 - val_loss: 2253.7478\n",
      "Epoch 366/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 713.2469 - val_loss: 2281.9534\n",
      "Epoch 367/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 762.6305 - val_loss: 2221.5029\n",
      "Epoch 368/1000\n",
      "115/115 [==============================] - 0s 278us/step - loss: 717.7605 - val_loss: 2225.4851\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 369/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 713.8035 - val_loss: 2232.7039\n",
      "Epoch 370/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 728.4074 - val_loss: 2227.6123\n",
      "Epoch 371/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 719.1557 - val_loss: 2249.4717\n",
      "Epoch 372/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 732.0840 - val_loss: 2212.5098\n",
      "Epoch 373/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 712.3106 - val_loss: 2222.6528\n",
      "Epoch 374/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 724.9928 - val_loss: 2206.1736\n",
      "Epoch 375/1000\n",
      "115/115 [==============================] - 0s 217us/step - loss: 708.2915 - val_loss: 2220.6116\n",
      "Epoch 376/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 718.7105 - val_loss: 2226.8328\n",
      "Epoch 377/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 705.3670 - val_loss: 2203.9949\n",
      "Epoch 378/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 735.0878 - val_loss: 2263.4485\n",
      "Epoch 379/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 726.5218 - val_loss: 2245.4248\n",
      "Epoch 380/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 719.1151 - val_loss: 2225.5779\n",
      "Epoch 381/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 718.6454 - val_loss: 2194.2092\n",
      "Epoch 382/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 710.8243 - val_loss: 2194.9656\n",
      "Epoch 383/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 704.0888 - val_loss: 2193.6877\n",
      "Epoch 384/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 709.5619 - val_loss: 2193.6155\n",
      "Epoch 385/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 707.0806 - val_loss: 2200.5168\n",
      "Epoch 386/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 724.5685 - val_loss: 2244.8066\n",
      "Epoch 387/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 709.9596 - val_loss: 2193.1809\n",
      "Epoch 388/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 720.1702 - val_loss: 2180.2839\n",
      "Epoch 389/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 707.4090 - val_loss: 2271.3293\n",
      "Epoch 390/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 715.4409 - val_loss: 2177.9297\n",
      "Epoch 391/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 699.6108 - val_loss: 2183.7856\n",
      "Epoch 392/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 713.2396 - val_loss: 2182.6228\n",
      "Epoch 393/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 695.5211 - val_loss: 2239.2058\n",
      "Epoch 394/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 727.9823 - val_loss: 2242.7332\n",
      "Epoch 395/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 687.2953 - val_loss: 2188.9871\n",
      "Epoch 396/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 725.1366 - val_loss: 2204.6423\n",
      "Epoch 397/1000\n",
      "115/115 [==============================] - 0s 278us/step - loss: 689.1822 - val_loss: 2255.7390\n",
      "Epoch 398/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 727.7128 - val_loss: 2247.7100\n",
      "Epoch 399/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 685.7492 - val_loss: 2167.9680\n",
      "Epoch 400/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 716.9056 - val_loss: 2250.4221\n",
      "Epoch 401/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 702.0703 - val_loss: 2212.4197\n",
      "Epoch 402/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 765.1807 - val_loss: 2307.2021\n",
      "Epoch 403/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 733.4373 - val_loss: 2204.0295\n",
      "Epoch 404/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 708.5149 - val_loss: 2154.3870\n",
      "Epoch 405/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 692.8072 - val_loss: 2219.5039\n",
      "Epoch 406/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 703.7845 - val_loss: 2159.2666\n",
      "Epoch 407/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 683.0443 - val_loss: 2150.1184\n",
      "Epoch 408/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 695.7601 - val_loss: 2148.1221\n",
      "Epoch 409/1000\n",
      "115/115 [==============================] - 0s 269us/step - loss: 688.5685 - val_loss: 2173.6956\n",
      "Epoch 410/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 685.9740 - val_loss: 2149.1331\n",
      "Epoch 411/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 686.6211 - val_loss: 2145.3225\n",
      "Epoch 412/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 687.5269 - val_loss: 2140.4666\n",
      "Epoch 413/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 698.2751 - val_loss: 2165.5720\n",
      "Epoch 414/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 686.3722 - val_loss: 2143.1731\n",
      "Epoch 415/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 713.8390 - val_loss: 2135.6304\n",
      "Epoch 416/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 670.6358 - val_loss: 2239.9563\n",
      "Epoch 417/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 723.7688 - val_loss: 2193.9827\n",
      "Epoch 418/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 668.8987 - val_loss: 2197.5613\n",
      "Epoch 419/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 735.1002 - val_loss: 2148.0447\n",
      "Epoch 420/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 711.4841 - val_loss: 2250.6455\n",
      "Epoch 421/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 717.8985 - val_loss: 2131.8997\n",
      "Epoch 422/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 677.7985 - val_loss: 2143.3350\n",
      "Epoch 423/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 671.6546 - val_loss: 2131.5015\n",
      "Epoch 424/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 679.7178 - val_loss: 2131.5325\n",
      "Epoch 425/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 668.0323 - val_loss: 2150.0312\n",
      "Epoch 426/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 672.9413 - val_loss: 2167.8420\n",
      "Epoch 427/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 678.0474 - val_loss: 2130.3821\n",
      "Epoch 428/1000\n",
      "115/115 [==============================] - 0s 278us/step - loss: 673.1053 - val_loss: 2129.9307\n",
      "Epoch 429/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 675.2578 - val_loss: 2135.6331\n",
      "Epoch 430/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 674.9564 - val_loss: 2141.0781\n",
      "Epoch 431/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 666.5013 - val_loss: 2132.2139\n",
      "Epoch 432/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 671.3395 - val_loss: 2131.3704\n",
      "Epoch 433/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 669.8217 - val_loss: 2129.9026\n",
      "Epoch 434/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 666.0807 - val_loss: 2152.3525\n",
      "Epoch 435/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 673.1645 - val_loss: 2152.0098\n",
      "Epoch 436/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 669.5724 - val_loss: 2130.8796\n",
      "Epoch 437/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 695.4910 - val_loss: 2134.5291\n",
      "Epoch 438/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 689.5012 - val_loss: 2202.5393\n",
      "Epoch 439/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 681.7606 - val_loss: 2136.6658\n",
      "Epoch 440/1000\n",
      "115/115 [==============================] - 0s 269us/step - loss: 667.4565 - val_loss: 2132.3464\n",
      "Epoch 441/1000\n",
      "115/115 [==============================] - 0s 278us/step - loss: 670.1060 - val_loss: 2138.7729\n",
      "Epoch 442/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 670.3067 - val_loss: 2130.1270\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 443/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 671.7245 - val_loss: 2182.8425\n",
      "Epoch 444/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 675.0024 - val_loss: 2132.2610\n",
      "Epoch 445/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 671.3202 - val_loss: 2134.7874\n",
      "Epoch 446/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 665.9913 - val_loss: 2140.4583\n",
      "Epoch 447/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 662.8364 - val_loss: 2138.6143\n",
      "Epoch 448/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 674.0794 - val_loss: 2132.5234\n",
      "Epoch 449/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 665.7353 - val_loss: 2173.3665\n",
      "Epoch 450/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 702.2735 - val_loss: 2133.5300\n",
      "Epoch 451/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 656.4582 - val_loss: 2197.6565\n",
      "Epoch 452/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 679.3682 - val_loss: 2153.3916\n",
      "Epoch 453/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 665.8096 - val_loss: 2130.5779\n",
      "Epoch 454/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 663.2883 - val_loss: 2129.1726\n",
      "Epoch 455/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 662.7525 - val_loss: 2140.9014\n",
      "Epoch 456/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 670.3423 - val_loss: 2135.0840\n",
      "Epoch 457/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 675.3087 - val_loss: 2158.1064\n",
      "Epoch 458/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 659.6332 - val_loss: 2128.9666\n",
      "Epoch 459/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 670.9567 - val_loss: 2128.1135\n",
      "Epoch 460/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 655.6052 - val_loss: 2163.2727\n",
      "Epoch 461/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 675.1036 - val_loss: 2176.3552\n",
      "Epoch 462/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 659.1576 - val_loss: 2124.8401\n",
      "Epoch 463/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 667.5545 - val_loss: 2158.7312\n",
      "Epoch 464/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 671.8310 - val_loss: 2136.1169\n",
      "Epoch 465/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 671.9906 - val_loss: 2182.7346\n",
      "Epoch 466/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 666.2882 - val_loss: 2120.2288\n",
      "Epoch 467/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 662.4210 - val_loss: 2129.2405\n",
      "Epoch 468/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 676.4379 - val_loss: 2126.6416\n",
      "Epoch 469/1000\n",
      "115/115 [==============================] - 0s 278us/step - loss: 669.6071 - val_loss: 2121.4353\n",
      "Epoch 470/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 690.0527 - val_loss: 2177.0364\n",
      "Epoch 471/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 674.8736 - val_loss: 2125.7004\n",
      "Epoch 472/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 664.4496 - val_loss: 2122.0059\n",
      "Epoch 473/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 669.6329 - val_loss: 2147.1816\n",
      "Epoch 474/1000\n",
      "115/115 [==============================] - 0s 217us/step - loss: 670.6504 - val_loss: 2130.6309\n",
      "Epoch 475/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 672.2002 - val_loss: 2128.9702\n",
      "Epoch 476/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 663.4693 - val_loss: 2182.1121\n",
      "Epoch 477/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 667.4212 - val_loss: 2127.7029\n",
      "Epoch 478/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 666.3621 - val_loss: 2127.1140\n",
      "Epoch 479/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 665.0068 - val_loss: 2132.4255\n",
      "Epoch 480/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 666.0715 - val_loss: 2124.8992\n",
      "Epoch 481/1000\n",
      "115/115 [==============================] - 0s 269us/step - loss: 662.7147 - val_loss: 2122.6748\n",
      "Epoch 482/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 664.1703 - val_loss: 2128.6545\n",
      "Epoch 483/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 660.6486 - val_loss: 2118.4800\n",
      "Epoch 484/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 661.9927 - val_loss: 2117.2961\n",
      "Epoch 485/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 667.1236 - val_loss: 2129.3240\n",
      "Epoch 486/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 663.1894 - val_loss: 2134.5200\n",
      "Epoch 487/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 662.5589 - val_loss: 2117.9763\n",
      "Epoch 488/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 670.5671 - val_loss: 2117.7217\n",
      "Epoch 489/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 662.9246 - val_loss: 2141.7307\n",
      "Epoch 490/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 664.3244 - val_loss: 2117.4854\n",
      "Epoch 491/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 659.4342 - val_loss: 2121.7717\n",
      "Epoch 492/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 669.1079 - val_loss: 2116.3809\n",
      "Epoch 493/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 665.2221 - val_loss: 2134.7991\n",
      "Epoch 494/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 682.7765 - val_loss: 2150.6218\n",
      "Epoch 495/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 671.6912 - val_loss: 2134.6233\n",
      "Epoch 496/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 669.1047 - val_loss: 2126.2666\n",
      "Epoch 497/1000\n",
      "115/115 [==============================] - 0s 286us/step - loss: 678.0007 - val_loss: 2162.9802\n",
      "Epoch 498/1000\n",
      "115/115 [==============================] - 0s 278us/step - loss: 679.1245 - val_loss: 2127.8513\n",
      "Epoch 499/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 663.7486 - val_loss: 2142.1001\n",
      "Epoch 500/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 674.0580 - val_loss: 2137.0325\n",
      "Epoch 501/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 661.0293 - val_loss: 2121.4163\n",
      "Epoch 502/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 664.4803 - val_loss: 2117.2708\n",
      "Epoch 503/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 661.9417 - val_loss: 2128.1650\n",
      "Epoch 504/1000\n",
      "115/115 [==============================] - 0s 269us/step - loss: 659.9800 - val_loss: 2118.6848\n",
      "Epoch 505/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 678.4986 - val_loss: 2114.3538\n",
      "Epoch 506/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 667.0396 - val_loss: 2127.3318\n",
      "Epoch 507/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 692.2279 - val_loss: 2198.7336\n",
      "Epoch 508/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 672.2321 - val_loss: 2113.7939\n",
      "Epoch 509/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 679.6267 - val_loss: 2137.7693\n",
      "Epoch 510/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 663.1180 - val_loss: 2144.1909\n",
      "Epoch 511/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 689.3680 - val_loss: 2169.6516\n",
      "Epoch 512/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 679.2564 - val_loss: 2170.1340\n",
      "Epoch 513/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 685.2101 - val_loss: 2111.9917\n",
      "Epoch 514/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 680.2992 - val_loss: 2209.4678\n",
      "Epoch 515/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 673.9747 - val_loss: 2114.5139\n",
      "Epoch 516/1000\n",
      "115/115 [==============================] - 0s 373us/step - loss: 676.8387 - val_loss: 2116.7380\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 517/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 663.1392 - val_loss: 2135.3726\n",
      "Epoch 518/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 668.1759 - val_loss: 2134.3401\n",
      "Epoch 519/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 660.5336 - val_loss: 2115.6306\n",
      "Epoch 520/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 667.0921 - val_loss: 2119.4380\n",
      "Epoch 521/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 664.7607 - val_loss: 2124.3159\n",
      "Epoch 522/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 662.5089 - val_loss: 2131.3262\n",
      "Epoch 523/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 669.8232 - val_loss: 2117.0071\n",
      "Epoch 524/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 673.0061 - val_loss: 2114.1487\n",
      "Epoch 525/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 670.6371 - val_loss: 2158.9841\n",
      "Epoch 526/1000\n",
      "115/115 [==============================] - 0s 226us/step - loss: 670.2744 - val_loss: 2115.4407\n",
      "Epoch 527/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 659.5693 - val_loss: 2110.3171\n",
      "Epoch 528/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 670.3871 - val_loss: 2116.4065\n",
      "Epoch 529/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 659.8876 - val_loss: 2146.6091\n",
      "Epoch 530/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 672.3098 - val_loss: 2139.6704\n",
      "Epoch 531/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 654.8863 - val_loss: 2128.3494\n",
      "Epoch 532/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 671.6020 - val_loss: 2111.9478\n",
      "Epoch 533/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 664.3910 - val_loss: 2138.2869\n",
      "Epoch 534/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 664.2747 - val_loss: 2132.7083\n",
      "Epoch 535/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 694.4672 - val_loss: 2114.4402\n",
      "Epoch 536/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 681.3637 - val_loss: 2216.1240\n",
      "Epoch 537/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 685.2549 - val_loss: 2113.9783\n",
      "Epoch 538/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 660.6094 - val_loss: 2114.1682\n",
      "Epoch 539/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 661.4263 - val_loss: 2113.8140\n",
      "Epoch 540/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 663.7520 - val_loss: 2112.1111\n",
      "Epoch 541/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 663.1754 - val_loss: 2115.0708\n",
      "Epoch 542/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 662.9521 - val_loss: 2129.6050\n",
      "Epoch 543/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 663.5549 - val_loss: 2114.5459\n",
      "Epoch 544/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 674.8177 - val_loss: 2118.6384\n",
      "Epoch 545/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 663.4461 - val_loss: 2123.5955\n",
      "Epoch 546/1000\n",
      "115/115 [==============================] - 0s 278us/step - loss: 667.0244 - val_loss: 2112.8381\n",
      "Epoch 547/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 663.9165 - val_loss: 2121.8250\n",
      "Epoch 548/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 668.5084 - val_loss: 2173.0305\n",
      "Epoch 549/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 664.5896 - val_loss: 2112.7593\n",
      "Epoch 550/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 661.9356 - val_loss: 2126.4634\n",
      "Epoch 551/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 668.4940 - val_loss: 2125.2644\n",
      "Epoch 552/1000\n",
      "115/115 [==============================] - 0s 269us/step - loss: 666.2292 - val_loss: 2119.3418\n",
      "Epoch 553/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 659.9935 - val_loss: 2124.3525\n",
      "Epoch 554/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 672.7126 - val_loss: 2109.3899\n",
      "Epoch 555/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 687.2971 - val_loss: 2244.0720\n",
      "Epoch 556/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 680.7277 - val_loss: 2111.8379\n",
      "Epoch 557/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 701.8387 - val_loss: 2123.6531\n",
      "Epoch 558/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 703.3726 - val_loss: 2280.2837\n",
      "Epoch 559/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 691.5555 - val_loss: 2110.8376\n",
      "Epoch 560/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 665.7567 - val_loss: 2126.2637\n",
      "Epoch 561/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 682.0672 - val_loss: 2120.4377\n",
      "Epoch 562/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 660.8134 - val_loss: 2110.2781\n",
      "Epoch 563/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 660.2816 - val_loss: 2109.1365\n",
      "Epoch 564/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 664.8373 - val_loss: 2116.8083\n",
      "Epoch 565/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 660.9944 - val_loss: 2105.8782\n",
      "Epoch 566/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 671.6847 - val_loss: 2116.9492\n",
      "Epoch 567/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 662.0704 - val_loss: 2105.1829\n",
      "Epoch 568/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 663.0924 - val_loss: 2104.1057\n",
      "Epoch 569/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 660.2898 - val_loss: 2116.4531\n",
      "Epoch 570/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 668.5237 - val_loss: 2123.5576\n",
      "Epoch 571/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 659.5528 - val_loss: 2107.1716\n",
      "Epoch 572/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 676.1727 - val_loss: 2106.7139\n",
      "Epoch 573/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 657.5248 - val_loss: 2125.5195\n",
      "Epoch 574/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 702.3443 - val_loss: 2204.2952\n",
      "Epoch 575/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 671.0605 - val_loss: 2103.0447\n",
      "Epoch 576/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 690.9913 - val_loss: 2192.0652\n",
      "Epoch 577/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 685.7162 - val_loss: 2143.9285\n",
      "Epoch 578/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 676.8761 - val_loss: 2132.6824\n",
      "Epoch 579/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 664.3892 - val_loss: 2110.1641\n",
      "Epoch 580/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 664.3546 - val_loss: 2105.7190\n",
      "Epoch 581/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 665.0045 - val_loss: 2101.3440\n",
      "Epoch 582/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 664.0086 - val_loss: 2115.7715\n",
      "Epoch 583/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 663.7279 - val_loss: 2110.9919\n",
      "Epoch 584/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 672.0164 - val_loss: 2106.1279\n",
      "Epoch 585/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 666.8967 - val_loss: 2136.0486\n",
      "Epoch 586/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 677.1341 - val_loss: 2140.3018\n",
      "Epoch 587/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 664.9837 - val_loss: 2111.1677\n",
      "Epoch 588/1000\n",
      "115/115 [==============================] - 0s 269us/step - loss: 661.8242 - val_loss: 2113.9524\n",
      "Epoch 589/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 662.4599 - val_loss: 2111.7886\n",
      "Epoch 590/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 661.6716 - val_loss: 2111.6414\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 591/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 660.9848 - val_loss: 2124.9976\n",
      "Epoch 592/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 663.7500 - val_loss: 2128.6428\n",
      "Epoch 593/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 668.9858 - val_loss: 2119.1875\n",
      "Epoch 594/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 670.9716 - val_loss: 2151.6897\n",
      "Epoch 595/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 661.4289 - val_loss: 2111.9463\n",
      "Epoch 596/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 668.5651 - val_loss: 2111.7009\n",
      "Epoch 597/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 663.1073 - val_loss: 2120.1150\n",
      "Epoch 598/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 675.3999 - val_loss: 2120.1130\n",
      "Epoch 599/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 662.0295 - val_loss: 2120.3367\n",
      "Epoch 600/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 700.6509 - val_loss: 2171.2336\n",
      "Epoch 601/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 658.2430 - val_loss: 2139.2483\n",
      "Epoch 602/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 683.6221 - val_loss: 2120.6667\n",
      "Epoch 603/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 663.4899 - val_loss: 2127.4749\n",
      "Epoch 604/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 662.8141 - val_loss: 2139.5249\n",
      "Epoch 605/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 661.1218 - val_loss: 2105.8193\n",
      "Epoch 606/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 662.9017 - val_loss: 2106.7810\n",
      "Epoch 607/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 662.3387 - val_loss: 2117.5784\n",
      "Epoch 608/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 662.7344 - val_loss: 2113.0295\n",
      "Epoch 609/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 669.1357 - val_loss: 2111.5339\n",
      "Epoch 610/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 660.1518 - val_loss: 2166.0618\n",
      "Epoch 611/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 671.4672 - val_loss: 2115.8936\n",
      "Epoch 612/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 658.8033 - val_loss: 2111.6682\n",
      "Epoch 613/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 671.9420 - val_loss: 2112.0330\n",
      "Epoch 614/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 664.4089 - val_loss: 2142.4011\n",
      "Epoch 615/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 672.4867 - val_loss: 2159.3132\n",
      "Epoch 616/1000\n",
      "115/115 [==============================] - 0s 269us/step - loss: 678.4926 - val_loss: 2114.6814\n",
      "Epoch 617/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 669.9566 - val_loss: 2107.6367\n",
      "Epoch 618/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 677.9694 - val_loss: 2174.8904\n",
      "Epoch 619/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 676.1400 - val_loss: 2110.0515\n",
      "Epoch 620/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 661.9533 - val_loss: 2104.2168\n",
      "Epoch 621/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 673.4298 - val_loss: 2103.8118\n",
      "Epoch 622/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 671.0021 - val_loss: 2143.6038\n",
      "Epoch 623/1000\n",
      "115/115 [==============================] - 0s 269us/step - loss: 659.7861 - val_loss: 2105.2083\n",
      "Epoch 624/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 672.2143 - val_loss: 2109.0896\n",
      "Epoch 625/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 687.2887 - val_loss: 2133.9370\n",
      "Epoch 626/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 671.4627 - val_loss: 2105.6360\n",
      "Epoch 627/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 661.2829 - val_loss: 2110.3496\n",
      "Epoch 628/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 663.0354 - val_loss: 2154.0283\n",
      "Epoch 629/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 663.9824 - val_loss: 2112.8840\n",
      "Epoch 630/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 656.7920 - val_loss: 2119.7153\n",
      "Epoch 631/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 667.6598 - val_loss: 2108.3467\n",
      "Epoch 632/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 662.3873 - val_loss: 2120.3982\n",
      "Epoch 633/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 660.4002 - val_loss: 2107.5742\n",
      "Epoch 634/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 659.5384 - val_loss: 2107.9104\n",
      "Epoch 635/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 665.6125 - val_loss: 2116.1885\n",
      "Epoch 636/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 665.3424 - val_loss: 2124.2686\n",
      "Epoch 637/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 666.4227 - val_loss: 2105.7000\n",
      "Epoch 638/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 659.4814 - val_loss: 2107.5310\n",
      "Epoch 639/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 664.9668 - val_loss: 2116.1809\n",
      "Epoch 640/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 659.3454 - val_loss: 2107.6521\n",
      "Epoch 641/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 673.8831 - val_loss: 2115.0247\n",
      "Epoch 642/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 661.4764 - val_loss: 2108.5837\n",
      "Epoch 643/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 667.9399 - val_loss: 2110.5845\n",
      "Epoch 644/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 667.9840 - val_loss: 2134.4792\n",
      "Epoch 645/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 666.9927 - val_loss: 2112.8738\n",
      "Epoch 646/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 667.4821 - val_loss: 2108.8464\n",
      "Epoch 647/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 664.0214 - val_loss: 2128.5305\n",
      "Epoch 648/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 663.2624 - val_loss: 2117.0564\n",
      "Epoch 649/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 658.7584 - val_loss: 2107.2771\n",
      "Epoch 650/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 659.8897 - val_loss: 2106.7476\n",
      "Epoch 651/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 661.9266 - val_loss: 2118.8665\n",
      "Epoch 652/1000\n",
      "115/115 [==============================] - 0s 269us/step - loss: 663.0147 - val_loss: 2106.8987\n",
      "Epoch 653/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 659.2957 - val_loss: 2118.0276\n",
      "Epoch 654/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 659.1877 - val_loss: 2113.0134\n",
      "Epoch 655/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 659.6046 - val_loss: 2108.9941\n",
      "Epoch 656/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 659.6643 - val_loss: 2107.3508\n",
      "Epoch 657/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 682.9115 - val_loss: 2110.3704\n",
      "Epoch 658/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 668.6948 - val_loss: 2156.6956\n",
      "Epoch 659/1000\n",
      "115/115 [==============================] - 0s 269us/step - loss: 667.1169 - val_loss: 2116.0823\n",
      "Epoch 660/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 668.9035 - val_loss: 2107.4932\n",
      "Epoch 661/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 662.1922 - val_loss: 2126.5198\n",
      "Epoch 662/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 659.3286 - val_loss: 2112.3479\n",
      "Epoch 663/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 678.1776 - val_loss: 2111.3467\n",
      "Epoch 664/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 673.2446 - val_loss: 2136.9885\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 665/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 667.6014 - val_loss: 2113.2151\n",
      "Epoch 666/1000\n",
      "115/115 [==============================] - 0s 226us/step - loss: 675.6317 - val_loss: 2207.2104\n",
      "Epoch 667/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 671.3931 - val_loss: 2107.7957\n",
      "Epoch 668/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 674.5012 - val_loss: 2144.2625\n",
      "Epoch 669/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 668.4300 - val_loss: 2111.6240\n",
      "Epoch 670/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 673.1155 - val_loss: 2207.1172\n",
      "Epoch 671/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 672.5758 - val_loss: 2105.2410\n",
      "Epoch 672/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 659.3675 - val_loss: 2128.6687\n",
      "Epoch 673/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 676.8693 - val_loss: 2104.2922\n",
      "Epoch 674/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 658.3148 - val_loss: 2149.4763\n",
      "Epoch 675/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 667.9102 - val_loss: 2121.9824\n",
      "Epoch 676/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 661.1291 - val_loss: 2111.1799\n",
      "Epoch 677/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 668.0155 - val_loss: 2112.0771\n",
      "Epoch 678/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 666.7351 - val_loss: 2108.4568\n",
      "Epoch 679/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 679.1882 - val_loss: 2165.9031\n",
      "Epoch 680/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 674.9168 - val_loss: 2114.7126\n",
      "Epoch 681/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 695.9895 - val_loss: 2137.0879\n",
      "Epoch 682/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 661.2528 - val_loss: 2140.9797\n",
      "Epoch 683/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 672.3915 - val_loss: 2195.3835\n",
      "Epoch 684/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 666.2233 - val_loss: 2105.5554\n",
      "Epoch 685/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 660.4327 - val_loss: 2116.3030\n",
      "Epoch 686/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 672.1175 - val_loss: 2108.1091\n",
      "Epoch 687/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 660.0988 - val_loss: 2105.5149\n",
      "Epoch 688/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 662.3956 - val_loss: 2105.3801\n",
      "Epoch 689/1000\n",
      "115/115 [==============================] - 0s 217us/step - loss: 667.0061 - val_loss: 2104.8574\n",
      "Epoch 690/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 661.3725 - val_loss: 2115.2917\n",
      "Epoch 691/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 658.2026 - val_loss: 2131.5173\n",
      "Epoch 692/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 665.9454 - val_loss: 2116.1787\n",
      "Epoch 693/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 661.7834 - val_loss: 2110.2976\n",
      "Epoch 694/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 673.8340 - val_loss: 2115.2522\n",
      "Epoch 695/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 664.4991 - val_loss: 2113.0359\n",
      "Epoch 696/1000\n",
      "115/115 [==============================] - 0s 278us/step - loss: 658.3187 - val_loss: 2137.1021\n",
      "Epoch 697/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 664.8402 - val_loss: 2113.3030\n",
      "Epoch 698/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 671.0901 - val_loss: 2115.1418\n",
      "Epoch 699/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 654.2488 - val_loss: 2116.2092\n",
      "Epoch 700/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 666.6590 - val_loss: 2106.5549\n",
      "Epoch 701/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 655.9694 - val_loss: 2125.5862\n",
      "Epoch 702/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 679.7157 - val_loss: 2176.2300\n",
      "Epoch 703/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 673.3310 - val_loss: 2107.0310\n",
      "Epoch 704/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 670.8066 - val_loss: 2108.4490\n",
      "Epoch 705/1000\n",
      "115/115 [==============================] - 0s 269us/step - loss: 655.3156 - val_loss: 2156.9783\n",
      "Epoch 706/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 683.2773 - val_loss: 2183.2058\n",
      "Epoch 707/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 665.5350 - val_loss: 2132.8049\n",
      "Epoch 708/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 676.7337 - val_loss: 2112.8259\n",
      "Epoch 709/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 662.0274 - val_loss: 2123.5925\n",
      "Epoch 710/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 664.8344 - val_loss: 2137.4688\n",
      "Epoch 711/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 665.7496 - val_loss: 2116.5122\n",
      "Epoch 712/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 659.4859 - val_loss: 2125.3601\n",
      "Epoch 713/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 662.6448 - val_loss: 2114.4211\n",
      "Epoch 714/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 661.6832 - val_loss: 2109.8220\n",
      "Epoch 715/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 681.1698 - val_loss: 2121.1116\n",
      "Epoch 716/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 658.8114 - val_loss: 2112.6904\n",
      "Epoch 717/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 659.2543 - val_loss: 2104.2761\n",
      "Epoch 718/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 659.0453 - val_loss: 2106.1633\n",
      "Epoch 719/1000\n",
      "115/115 [==============================] - 0s 269us/step - loss: 661.4435 - val_loss: 2111.9641\n",
      "Epoch 720/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 688.8830 - val_loss: 2103.7561\n",
      "Epoch 721/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 665.5308 - val_loss: 2136.6465\n",
      "Epoch 722/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 664.5154 - val_loss: 2112.4470\n",
      "Epoch 723/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 668.0331 - val_loss: 2105.2131\n",
      "Epoch 724/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 663.1652 - val_loss: 2118.9397\n",
      "Epoch 725/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 661.0367 - val_loss: 2108.8516\n",
      "Epoch 726/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 688.3926 - val_loss: 2113.3613\n",
      "Epoch 727/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 681.8660 - val_loss: 2153.7971\n",
      "Epoch 728/1000\n",
      "115/115 [==============================] - 0s 269us/step - loss: 665.8538 - val_loss: 2106.7517\n",
      "Epoch 729/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 664.9992 - val_loss: 2106.3113\n",
      "Epoch 730/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 663.7571 - val_loss: 2104.4961\n",
      "Epoch 731/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 670.8929 - val_loss: 2110.6482\n",
      "Epoch 732/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 665.7761 - val_loss: 2110.2729\n",
      "Epoch 733/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 664.8092 - val_loss: 2130.2471\n",
      "Epoch 734/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 661.9287 - val_loss: 2104.5144\n",
      "Epoch 735/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 660.4359 - val_loss: 2108.0811\n",
      "Epoch 736/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 662.2335 - val_loss: 2104.7026\n",
      "Epoch 737/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 681.4117 - val_loss: 2122.9021\n",
      "Epoch 738/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 656.6282 - val_loss: 2114.3381\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 739/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 674.6660 - val_loss: 2105.1506\n",
      "Epoch 740/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 662.5278 - val_loss: 2122.3247\n",
      "Epoch 741/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 661.6521 - val_loss: 2114.7307\n",
      "Epoch 742/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 658.1359 - val_loss: 2103.9363\n",
      "Epoch 743/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 657.8951 - val_loss: 2106.0337\n",
      "Epoch 744/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 663.2917 - val_loss: 2108.3391\n",
      "Epoch 745/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 659.4325 - val_loss: 2109.0037\n",
      "Epoch 746/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 673.5925 - val_loss: 2122.7070\n",
      "Epoch 747/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 656.9659 - val_loss: 2107.8245\n",
      "Epoch 748/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 666.2013 - val_loss: 2104.7219\n",
      "Epoch 749/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 683.6304 - val_loss: 2128.3396\n",
      "Epoch 750/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 676.2950 - val_loss: 2109.5566\n",
      "Epoch 751/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 664.0562 - val_loss: 2169.0188\n",
      "Epoch 752/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 676.5767 - val_loss: 2123.0056\n",
      "Epoch 753/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 658.4412 - val_loss: 2116.2600\n",
      "Epoch 754/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 659.8084 - val_loss: 2110.0891\n",
      "Epoch 755/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 664.5583 - val_loss: 2107.8694\n",
      "Epoch 756/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 664.5787 - val_loss: 2116.7590\n",
      "Epoch 757/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 681.0354 - val_loss: 2106.3584\n",
      "Epoch 758/1000\n",
      "115/115 [==============================] - 0s 269us/step - loss: 673.8159 - val_loss: 2210.1785\n",
      "Epoch 759/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 674.8354 - val_loss: 2125.2305\n",
      "Epoch 760/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 654.3932 - val_loss: 2112.0349\n",
      "Epoch 761/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 670.8720 - val_loss: 2106.7134\n",
      "Epoch 762/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 675.1766 - val_loss: 2108.0667\n",
      "Epoch 763/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 660.9292 - val_loss: 2193.5540\n",
      "Epoch 764/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 675.6950 - val_loss: 2171.8000\n",
      "Epoch 765/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 677.6391 - val_loss: 2104.8115\n",
      "Epoch 766/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 661.9735 - val_loss: 2107.0007\n",
      "Epoch 767/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 658.7200 - val_loss: 2115.4607\n",
      "Epoch 768/1000\n",
      "115/115 [==============================] - 0s 269us/step - loss: 660.7541 - val_loss: 2113.8809\n",
      "Epoch 769/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 660.2508 - val_loss: 2108.3381\n",
      "Epoch 770/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 668.9447 - val_loss: 2108.0300\n",
      "Epoch 771/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 658.2546 - val_loss: 2187.9746\n",
      "Epoch 772/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 689.4065 - val_loss: 2113.2246\n",
      "Epoch 773/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 659.6221 - val_loss: 2113.7361\n",
      "Epoch 774/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 662.2275 - val_loss: 2135.9041\n",
      "Epoch 775/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 660.1919 - val_loss: 2103.7556\n",
      "Epoch 776/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 682.4263 - val_loss: 2103.1360\n",
      "Epoch 777/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 665.1691 - val_loss: 2170.1755\n",
      "Epoch 778/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 668.1923 - val_loss: 2114.0012\n",
      "Epoch 779/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 679.7225 - val_loss: 2114.2737\n",
      "Epoch 780/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 664.2471 - val_loss: 2116.5632\n",
      "Epoch 781/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 657.7382 - val_loss: 2142.8086\n",
      "Epoch 782/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 667.5443 - val_loss: 2152.2217\n",
      "Epoch 783/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 666.3320 - val_loss: 2100.4883\n",
      "Epoch 784/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 680.3818 - val_loss: 2106.1169\n",
      "Epoch 785/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 664.0617 - val_loss: 2131.2532\n",
      "Epoch 786/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 676.7255 - val_loss: 2104.3396\n",
      "Epoch 787/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 646.2994 - val_loss: 2179.7344\n",
      "Epoch 788/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 675.6909 - val_loss: 2221.7820\n",
      "Epoch 789/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 680.8643 - val_loss: 2128.7996\n",
      "Epoch 790/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 657.3137 - val_loss: 2107.9001\n",
      "Epoch 791/1000\n",
      "115/115 [==============================] - 0s 269us/step - loss: 670.7289 - val_loss: 2103.6604\n",
      "Epoch 792/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 668.7186 - val_loss: 2127.3867\n",
      "Epoch 793/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 671.3222 - val_loss: 2105.0457\n",
      "Epoch 794/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 657.2617 - val_loss: 2123.2673\n",
      "Epoch 795/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 661.5775 - val_loss: 2142.1970\n",
      "Epoch 796/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 672.1706 - val_loss: 2165.1262\n",
      "Epoch 797/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 655.6581 - val_loss: 2109.2405\n",
      "Epoch 798/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 675.0286 - val_loss: 2129.1133\n",
      "Epoch 799/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 661.4140 - val_loss: 2137.1877\n",
      "Epoch 800/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 662.1682 - val_loss: 2172.5803\n",
      "Epoch 801/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 671.5564 - val_loss: 2134.7715\n",
      "Epoch 802/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 661.2969 - val_loss: 2111.6460\n",
      "Epoch 803/1000\n",
      "115/115 [==============================] - 0s 269us/step - loss: 668.0732 - val_loss: 2116.6707\n",
      "Epoch 804/1000\n",
      "115/115 [==============================] - 0s 269us/step - loss: 664.2528 - val_loss: 2108.3313\n",
      "Epoch 805/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 664.0148 - val_loss: 2113.1133\n",
      "Epoch 806/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 660.7753 - val_loss: 2158.6396\n",
      "Epoch 807/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 663.8421 - val_loss: 2113.0969\n",
      "Epoch 808/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 675.2114 - val_loss: 2106.0576\n",
      "Epoch 809/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 672.7997 - val_loss: 2148.5994\n",
      "Epoch 810/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 667.7252 - val_loss: 2115.2141\n",
      "Epoch 811/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 663.5793 - val_loss: 2112.3113\n",
      "Epoch 812/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 667.1696 - val_loss: 2117.0898\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 813/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 666.1485 - val_loss: 2112.7886\n",
      "Epoch 814/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 676.7887 - val_loss: 2167.0647\n",
      "Epoch 815/1000\n",
      "115/115 [==============================] - 0s 217us/step - loss: 661.2802 - val_loss: 2106.7627\n",
      "Epoch 816/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 677.0306 - val_loss: 2113.3440\n",
      "Epoch 817/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 664.0124 - val_loss: 2128.4951\n",
      "Epoch 818/1000\n",
      "115/115 [==============================] - 0s 217us/step - loss: 672.9311 - val_loss: 2155.7185\n",
      "Epoch 819/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 666.3948 - val_loss: 2098.9612\n",
      "Epoch 820/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 662.4648 - val_loss: 2097.0417\n",
      "Epoch 821/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 661.5751 - val_loss: 2122.8945\n",
      "Epoch 822/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 663.7016 - val_loss: 2105.9578\n",
      "Epoch 823/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 661.9085 - val_loss: 2104.4104\n",
      "Epoch 824/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 682.0627 - val_loss: 2105.4583\n",
      "Epoch 825/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 659.8774 - val_loss: 2115.8811\n",
      "Epoch 826/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 662.1600 - val_loss: 2255.2405\n",
      "Epoch 827/1000\n",
      "115/115 [==============================] - 0s 295us/step - loss: 687.5612 - val_loss: 2122.2783\n",
      "Epoch 828/1000\n",
      "115/115 [==============================] - 0s 269us/step - loss: 670.0765 - val_loss: 2111.2527\n",
      "Epoch 829/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 665.8262 - val_loss: 2104.7332\n",
      "Epoch 830/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 657.2336 - val_loss: 2120.0388\n",
      "Epoch 831/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 659.2849 - val_loss: 2122.7991\n",
      "Epoch 832/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 660.8334 - val_loss: 2119.8772\n",
      "Epoch 833/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 661.8502 - val_loss: 2108.4512\n",
      "Epoch 834/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 663.8509 - val_loss: 2109.9062\n",
      "Epoch 835/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 668.4195 - val_loss: 2130.9109\n",
      "Epoch 836/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 660.2218 - val_loss: 2111.1943\n",
      "Epoch 837/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 663.7423 - val_loss: 2106.1096\n",
      "Epoch 838/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 678.1032 - val_loss: 2113.6580\n",
      "Epoch 839/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 678.8405 - val_loss: 2164.8059\n",
      "Epoch 840/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 665.5472 - val_loss: 2130.5073\n",
      "Epoch 841/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 658.3764 - val_loss: 2111.4026\n",
      "Epoch 842/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 672.6960 - val_loss: 2118.1536\n",
      "Epoch 843/1000\n",
      "115/115 [==============================] - 0s 269us/step - loss: 663.2793 - val_loss: 2125.9929\n",
      "Epoch 844/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 659.1076 - val_loss: 2133.8252\n",
      "Epoch 845/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 685.1111 - val_loss: 2135.0210\n",
      "Epoch 846/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 649.3386 - val_loss: 2112.7612\n",
      "Epoch 847/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 698.6468 - val_loss: 2147.1619\n",
      "Epoch 848/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 680.0403 - val_loss: 2190.6072\n",
      "Epoch 849/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 679.4447 - val_loss: 2180.4719\n",
      "Epoch 850/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 671.2407 - val_loss: 2118.6389\n",
      "Epoch 851/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 673.6806 - val_loss: 2109.0117\n",
      "Epoch 852/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 662.5108 - val_loss: 2111.7803\n",
      "Epoch 853/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 663.1372 - val_loss: 2111.6118\n",
      "Epoch 854/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 669.9158 - val_loss: 2159.4988\n",
      "Epoch 855/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 658.3295 - val_loss: 2103.5654\n",
      "Epoch 856/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 673.2040 - val_loss: 2110.0242\n",
      "Epoch 857/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 666.9565 - val_loss: 2106.2197\n",
      "Epoch 858/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 661.6546 - val_loss: 2119.6885\n",
      "Epoch 859/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 659.1648 - val_loss: 2103.0671\n",
      "Epoch 860/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 659.0655 - val_loss: 2103.6113\n",
      "Epoch 861/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 659.6937 - val_loss: 2105.6379\n",
      "Epoch 862/1000\n",
      "115/115 [==============================] - 0s 278us/step - loss: 662.6145 - val_loss: 2108.8181\n",
      "Epoch 863/1000\n",
      "115/115 [==============================] - 0s 269us/step - loss: 666.5143 - val_loss: 2105.2576\n",
      "Epoch 864/1000\n",
      "115/115 [==============================] - ETA: 0s - loss: 799.336 - 0s 269us/step - loss: 664.8394 - val_loss: 2105.5696\n",
      "Epoch 865/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 656.6492 - val_loss: 2142.7498\n",
      "Epoch 866/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 675.1071 - val_loss: 2142.8542\n",
      "Epoch 867/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 649.8871 - val_loss: 2120.9377\n",
      "Epoch 868/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 674.1688 - val_loss: 2113.5735\n",
      "Epoch 869/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 664.5278 - val_loss: 2109.4648\n",
      "Epoch 870/1000\n",
      "115/115 [==============================] - 0s 217us/step - loss: 667.5621 - val_loss: 2129.4285\n",
      "Epoch 871/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 662.2221 - val_loss: 2111.5986\n",
      "Epoch 872/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 663.0191 - val_loss: 2118.4292\n",
      "Epoch 873/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 658.0972 - val_loss: 2115.9099\n",
      "Epoch 874/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 662.7929 - val_loss: 2118.9895\n",
      "Epoch 875/1000\n",
      "115/115 [==============================] - 0s 295us/step - loss: 664.3581 - val_loss: 2116.2717\n",
      "Epoch 876/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 672.0749 - val_loss: 2122.3445\n",
      "Epoch 877/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 660.5461 - val_loss: 2134.3142\n",
      "Epoch 878/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 665.7696 - val_loss: 2165.6755\n",
      "Epoch 879/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 673.4714 - val_loss: 2112.6755\n",
      "Epoch 880/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 663.7998 - val_loss: 2121.0762\n",
      "Epoch 881/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 666.5770 - val_loss: 2126.7214\n",
      "Epoch 882/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 668.0923 - val_loss: 2109.8953\n",
      "Epoch 883/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 662.5616 - val_loss: 2118.0105\n",
      "Epoch 884/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 662.1860 - val_loss: 2126.6660\n",
      "Epoch 885/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 658.8443 - val_loss: 2104.9111\n",
      "Epoch 886/1000\n",
      "115/115 [==============================] - 0s 269us/step - loss: 658.6962 - val_loss: 2102.2361\n",
      "Epoch 887/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 666.2512 - val_loss: 2102.5273\n",
      "Epoch 888/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 657.1175 - val_loss: 2146.4231\n",
      "Epoch 889/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 672.4402 - val_loss: 2142.0518\n",
      "Epoch 890/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 671.0545 - val_loss: 2103.7288\n",
      "Epoch 891/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 658.2872 - val_loss: 2109.7107\n",
      "Epoch 892/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 659.4498 - val_loss: 2109.3135\n",
      "Epoch 893/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 663.5789 - val_loss: 2123.1172\n",
      "Epoch 894/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 677.8186 - val_loss: 2105.4946\n",
      "Epoch 895/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 663.7687 - val_loss: 2134.8445\n",
      "Epoch 896/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 664.6327 - val_loss: 2112.8450\n",
      "Epoch 897/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 662.4642 - val_loss: 2116.9543\n",
      "Epoch 898/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 661.4490 - val_loss: 2102.3008\n",
      "Epoch 899/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 662.6394 - val_loss: 2105.7346\n",
      "Epoch 900/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 660.9366 - val_loss: 2128.0488\n",
      "Epoch 901/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 665.5935 - val_loss: 2124.9075\n",
      "Epoch 902/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 667.7507 - val_loss: 2099.4373\n",
      "Epoch 903/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 662.3025 - val_loss: 2118.8823\n",
      "Epoch 904/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 658.3226 - val_loss: 2106.9885\n",
      "Epoch 905/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 659.4423 - val_loss: 2101.1846\n",
      "Epoch 906/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 662.0623 - val_loss: 2100.9163\n",
      "Epoch 907/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 665.4536 - val_loss: 2106.6580\n",
      "Epoch 908/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 670.0957 - val_loss: 2144.9524\n",
      "Epoch 909/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 668.0389 - val_loss: 2103.5781\n",
      "Epoch 910/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 671.4215 - val_loss: 2121.2668\n",
      "Epoch 911/1000\n",
      "115/115 [==============================] - 0s 278us/step - loss: 669.8567 - val_loss: 2102.9500\n",
      "Epoch 912/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 689.7542 - val_loss: 2220.2349\n",
      "Epoch 913/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 669.6154 - val_loss: 2099.3289\n",
      "Epoch 914/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 666.9869 - val_loss: 2112.7795\n",
      "Epoch 915/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 666.6228 - val_loss: 2118.8557\n",
      "Epoch 916/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 662.3278 - val_loss: 2142.8086\n",
      "Epoch 917/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 662.6203 - val_loss: 2107.7744\n",
      "Epoch 918/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 676.5985 - val_loss: 2100.2168\n",
      "Epoch 919/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 661.3082 - val_loss: 2146.2886\n",
      "Epoch 920/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 673.0294 - val_loss: 2140.7957\n",
      "Epoch 921/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 658.9366 - val_loss: 2106.3828\n",
      "Epoch 922/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 667.7098 - val_loss: 2106.6335\n",
      "Epoch 923/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 665.7583 - val_loss: 2108.8384\n",
      "Epoch 924/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 658.1900 - val_loss: 2148.8792\n",
      "Epoch 925/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 689.2915 - val_loss: 2135.8135\n",
      "Epoch 926/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 664.6561 - val_loss: 2123.8479\n",
      "Epoch 927/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 670.2936 - val_loss: 2113.4641\n",
      "Epoch 928/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 668.7962 - val_loss: 2151.7092\n",
      "Epoch 929/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 664.9291 - val_loss: 2121.0369\n",
      "Epoch 930/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 658.6844 - val_loss: 2112.1658\n",
      "Epoch 931/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 660.1523 - val_loss: 2118.4502\n",
      "Epoch 932/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 667.4626 - val_loss: 2131.3123\n",
      "Epoch 933/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 656.2394 - val_loss: 2109.0935\n",
      "Epoch 934/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 662.7071 - val_loss: 2108.5237\n",
      "Epoch 935/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 660.6798 - val_loss: 2108.5598\n",
      "Epoch 936/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 657.2251 - val_loss: 2114.0742\n",
      "Epoch 937/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 660.1462 - val_loss: 2117.4299\n",
      "Epoch 938/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 658.4893 - val_loss: 2111.0486\n",
      "Epoch 939/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 662.8165 - val_loss: 2112.1619\n",
      "Epoch 940/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 660.6889 - val_loss: 2127.7683\n",
      "Epoch 941/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 662.4883 - val_loss: 2119.0232\n",
      "Epoch 942/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 676.1271 - val_loss: 2116.9380\n",
      "Epoch 943/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 666.3255 - val_loss: 2115.8845\n",
      "Epoch 944/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 657.5281 - val_loss: 2131.5427\n",
      "Epoch 945/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 660.3583 - val_loss: 2141.7747\n",
      "Epoch 946/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 664.6576 - val_loss: 2120.1238\n",
      "Epoch 947/1000\n",
      "115/115 [==============================] - 0s 295us/step - loss: 652.6924 - val_loss: 2123.9373\n",
      "Epoch 948/1000\n",
      "115/115 [==============================] - 0s 269us/step - loss: 682.1800 - val_loss: 2124.6772\n",
      "Epoch 949/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 669.6382 - val_loss: 2156.9856\n",
      "Epoch 950/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 669.7364 - val_loss: 2172.3845\n",
      "Epoch 951/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 660.7552 - val_loss: 2108.5398\n",
      "Epoch 952/1000\n",
      "115/115 [==============================] - 0s 252us/step - loss: 679.7811 - val_loss: 2112.7334\n",
      "Epoch 953/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 650.0021 - val_loss: 2165.6829\n",
      "Epoch 954/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 670.2374 - val_loss: 2167.1355\n",
      "Epoch 955/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 667.2322 - val_loss: 2106.4705\n",
      "Epoch 956/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 660.8291 - val_loss: 2100.0466\n",
      "Epoch 957/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 664.1990 - val_loss: 2099.9331\n",
      "Epoch 958/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 657.2931 - val_loss: 2116.4084\n",
      "Epoch 959/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 660.9267 - val_loss: 2124.3496\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 960/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 666.0048 - val_loss: 2108.2092\n",
      "Epoch 961/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 657.9240 - val_loss: 2108.5605\n",
      "Epoch 962/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 659.2235 - val_loss: 2106.2322\n",
      "Epoch 963/1000\n",
      "115/115 [==============================] - 0s 217us/step - loss: 674.3193 - val_loss: 2114.6677\n",
      "Epoch 964/1000\n",
      "115/115 [==============================] - 0s 226us/step - loss: 661.6887 - val_loss: 2211.2043\n",
      "Epoch 965/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 675.0400 - val_loss: 2138.5879\n",
      "Epoch 966/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 658.5402 - val_loss: 2107.6650\n",
      "Epoch 967/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 661.2069 - val_loss: 2107.2734\n",
      "Epoch 968/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 663.6721 - val_loss: 2106.0479\n",
      "Epoch 969/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 653.2895 - val_loss: 2149.4365\n",
      "Epoch 970/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 665.0422 - val_loss: 2148.5271\n",
      "Epoch 971/1000\n",
      "115/115 [==============================] - 0s 304us/step - loss: 657.8701 - val_loss: 2104.3191\n",
      "Epoch 972/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 659.7341 - val_loss: 2105.3604\n",
      "Epoch 973/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 660.9922 - val_loss: 2106.6191\n",
      "Epoch 974/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 658.2769 - val_loss: 2112.6565\n",
      "Epoch 975/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 659.6182 - val_loss: 2105.8513\n",
      "Epoch 976/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 660.6928 - val_loss: 2125.7117\n",
      "Epoch 977/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 679.6643 - val_loss: 2147.3357\n",
      "Epoch 978/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 664.2549 - val_loss: 2098.8259\n",
      "Epoch 979/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 663.4194 - val_loss: 2098.7314\n",
      "Epoch 980/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 657.6150 - val_loss: 2112.7126\n",
      "Epoch 981/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 670.6008 - val_loss: 2189.4934\n",
      "Epoch 982/1000\n",
      "115/115 [==============================] - 0s 278us/step - loss: 688.0272 - val_loss: 2135.5713\n",
      "Epoch 983/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 677.4450 - val_loss: 2155.6042\n",
      "Epoch 984/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 696.9172 - val_loss: 2126.5605\n",
      "Epoch 985/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 662.9612 - val_loss: 2159.0664\n",
      "Epoch 986/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 666.4201 - val_loss: 2127.0720\n",
      "Epoch 987/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 665.6559 - val_loss: 2106.7693\n",
      "Epoch 988/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 659.7702 - val_loss: 2124.0776\n",
      "Epoch 989/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 659.8136 - val_loss: 2111.4341\n",
      "Epoch 990/1000\n",
      "115/115 [==============================] - 0s 225us/step - loss: 658.5692 - val_loss: 2110.0964\n",
      "Epoch 991/1000\n",
      "115/115 [==============================] - 0s 243us/step - loss: 657.0888 - val_loss: 2105.4221\n",
      "Epoch 992/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 664.0849 - val_loss: 2104.1760\n",
      "Epoch 993/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 696.7739 - val_loss: 2165.1951\n",
      "Epoch 994/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 661.1655 - val_loss: 2102.0691\n",
      "Epoch 995/1000\n",
      "115/115 [==============================] - 0s 234us/step - loss: 664.4292 - val_loss: 2101.5867\n",
      "Epoch 996/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 656.4029 - val_loss: 2135.2271\n",
      "Epoch 997/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 661.2860 - val_loss: 2161.1555\n",
      "Epoch 998/1000\n",
      "115/115 [==============================] - 0s 251us/step - loss: 677.0454 - val_loss: 2156.9919\n",
      "Epoch 999/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 665.1407 - val_loss: 2107.1272\n",
      "Epoch 1000/1000\n",
      "115/115 [==============================] - 0s 260us/step - loss: 669.6464 - val_loss: 2108.3953\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x24bdf14bcf8>"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# fit model\n",
    "model.fit(X_train, y_train, epochs=1000, verbose=1, validation_data=(X_val, y_val))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "TRAIN DATA 0 : predicted=120.191406, ground_truth=135\n",
      "TRAIN DATA 1 : predicted=140.191986, ground_truth=148\n",
      "TRAIN DATA 2 : predicted=153.536377, ground_truth=148\n",
      "TRAIN DATA 3 : predicted=150.091507, ground_truth=136\n",
      "TRAIN DATA 4 : predicted=130.835434, ground_truth=119\n",
      "TRAIN DATA 5 : predicted=116.443466, ground_truth=104\n",
      "TRAIN DATA 6 : predicted=108.155396, ground_truth=118\n",
      "TRAIN DATA 7 : predicted=134.021271, ground_truth=115\n",
      "TRAIN DATA 8 : predicted=121.048599, ground_truth=126\n",
      "TRAIN DATA 9 : predicted=137.187851, ground_truth=141\n",
      "TRAIN DATA 10 : predicted=142.139648, ground_truth=135\n",
      "TRAIN DATA 11 : predicted=133.639771, ground_truth=125\n",
      "TRAIN DATA 12 : predicted=122.634842, ground_truth=149\n",
      "TRAIN DATA 13 : predicted=157.312012, ground_truth=170\n",
      "TRAIN DATA 14 : predicted=176.507477, ground_truth=170\n",
      "TRAIN DATA 15 : predicted=170.082062, ground_truth=158\n",
      "TRAIN DATA 16 : predicted=149.005722, ground_truth=133\n",
      "TRAIN DATA 17 : predicted=125.560593, ground_truth=114\n",
      "TRAIN DATA 18 : predicted=118.764290, ground_truth=140\n",
      "TRAIN DATA 19 : predicted=160.779266, ground_truth=145\n",
      "TRAIN DATA 20 : predicted=153.229080, ground_truth=150\n",
      "TRAIN DATA 21 : predicted=157.941895, ground_truth=178\n",
      "TRAIN DATA 22 : predicted=179.664001, ground_truth=163\n",
      "TRAIN DATA 23 : predicted=153.860825, ground_truth=172\n",
      "TRAIN DATA 24 : predicted=177.801193, ground_truth=178\n",
      "TRAIN DATA 25 : predicted=173.234879, ground_truth=199\n",
      "TRAIN DATA 26 : predicted=210.917404, ground_truth=199\n",
      "TRAIN DATA 27 : predicted=194.189117, ground_truth=184\n",
      "TRAIN DATA 28 : predicted=179.649139, ground_truth=162\n",
      "TRAIN DATA 29 : predicted=155.923370, ground_truth=146\n",
      "TRAIN DATA 30 : predicted=150.959061, ground_truth=166\n",
      "TRAIN DATA 31 : predicted=184.964615, ground_truth=171\n",
      "TRAIN DATA 32 : predicted=179.280960, ground_truth=180\n",
      "TRAIN DATA 33 : predicted=188.379333, ground_truth=193\n",
      "TRAIN DATA 34 : predicted=191.268845, ground_truth=181\n",
      "TRAIN DATA 35 : predicted=175.728592, ground_truth=183\n",
      "TRAIN DATA 36 : predicted=185.841843, ground_truth=218\n",
      "TRAIN DATA 37 : predicted=227.369797, ground_truth=230\n",
      "TRAIN DATA 38 : predicted=232.138290, ground_truth=242\n",
      "TRAIN DATA 39 : predicted=242.582169, ground_truth=209\n",
      "TRAIN DATA 40 : predicted=188.153397, ground_truth=191\n",
      "TRAIN DATA 41 : predicted=191.962387, ground_truth=172\n",
      "TRAIN DATA 42 : predicted=172.520874, ground_truth=194\n",
      "TRAIN DATA 43 : predicted=219.425751, ground_truth=196\n",
      "TRAIN DATA 44 : predicted=201.773117, ground_truth=196\n",
      "TRAIN DATA 45 : predicted=203.195587, ground_truth=236\n",
      "TRAIN DATA 46 : predicted=243.342514, ground_truth=235\n",
      "TRAIN DATA 47 : predicted=228.002426, ground_truth=229\n",
      "TRAIN DATA 48 : predicted=227.742432, ground_truth=243\n",
      "TRAIN DATA 49 : predicted=238.784286, ground_truth=264\n",
      "TRAIN DATA 50 : predicted=270.360321, ground_truth=272\n",
      "TRAIN DATA 51 : predicted=273.195160, ground_truth=237\n",
      "TRAIN DATA 52 : predicted=222.293518, ground_truth=211\n",
      "TRAIN DATA 53 : predicted=207.460175, ground_truth=180\n",
      "TRAIN DATA 54 : predicted=180.252274, ground_truth=201\n",
      "TRAIN DATA 55 : predicted=229.999649, ground_truth=204\n",
      "TRAIN DATA 56 : predicted=213.584961, ground_truth=188\n",
      "TRAIN DATA 57 : predicted=194.286453, ground_truth=235\n",
      "TRAIN DATA 58 : predicted=248.215698, ground_truth=227\n",
      "TRAIN DATA 59 : predicted=217.263184, ground_truth=234\n",
      "TRAIN DATA 60 : predicted=242.552368, ground_truth=264\n",
      "TRAIN DATA 61 : predicted=259.572662, ground_truth=302\n",
      "TRAIN DATA 62 : predicted=310.662720, ground_truth=293\n",
      "TRAIN DATA 63 : predicted=280.167236, ground_truth=259\n",
      "TRAIN DATA 64 : predicted=242.087967, ground_truth=229\n",
      "TRAIN DATA 65 : predicted=219.822906, ground_truth=203\n",
      "TRAIN DATA 66 : predicted=210.720993, ground_truth=229\n",
      "TRAIN DATA 67 : predicted=258.092102, ground_truth=242\n",
      "TRAIN DATA 68 : predicted=253.991623, ground_truth=233\n",
      "TRAIN DATA 69 : predicted=236.834106, ground_truth=267\n",
      "TRAIN DATA 70 : predicted=272.459625, ground_truth=269\n",
      "TRAIN DATA 71 : predicted=261.183533, ground_truth=270\n",
      "TRAIN DATA 72 : predicted=273.062500, ground_truth=315\n",
      "TRAIN DATA 73 : predicted=319.301117, ground_truth=364\n",
      "TRAIN DATA 74 : predicted=369.312469, ground_truth=347\n",
      "TRAIN DATA 75 : predicted=329.911163, ground_truth=312\n",
      "TRAIN DATA 76 : predicted=290.818970, ground_truth=274\n",
      "TRAIN DATA 77 : predicted=259.594238, ground_truth=237\n",
      "TRAIN DATA 78 : predicted=245.769928, ground_truth=278\n",
      "TRAIN DATA 79 : predicted=314.444763, ground_truth=284\n",
      "TRAIN DATA 80 : predicted=293.622345, ground_truth=277\n",
      "TRAIN DATA 81 : predicted=285.141327, ground_truth=317\n",
      "TRAIN DATA 82 : predicted=319.194763, ground_truth=313\n",
      "TRAIN DATA 83 : predicted=303.908325, ground_truth=318\n",
      "TRAIN DATA 84 : predicted=322.231171, ground_truth=374\n",
      "TRAIN DATA 85 : predicted=378.716400, ground_truth=413\n",
      "TRAIN DATA 86 : predicted=414.279358, ground_truth=405\n",
      "TRAIN DATA 87 : predicted=390.378235, ground_truth=355\n",
      "TRAIN DATA 88 : predicted=324.477264, ground_truth=306\n",
      "TRAIN DATA 89 : predicted=294.173676, ground_truth=271\n",
      "TRAIN DATA 90 : predicted=280.842834, ground_truth=306\n",
      "TRAIN DATA 91 : predicted=345.588074, ground_truth=315\n",
      "TRAIN DATA 92 : predicted=329.970001, ground_truth=301\n",
      "TRAIN DATA 93 : predicted=306.107269, ground_truth=356\n",
      "TRAIN DATA 94 : predicted=365.508362, ground_truth=348\n",
      "TRAIN DATA 95 : predicted=333.703583, ground_truth=355\n",
      "TRAIN DATA 96 : predicted=361.644165, ground_truth=422\n",
      "TRAIN DATA 97 : predicted=424.683350, ground_truth=465\n",
      "TRAIN DATA 98 : predicted=465.738159, ground_truth=467\n",
      "TRAIN DATA 99 : predicted=452.565491, ground_truth=404\n",
      "TRAIN DATA 100 : predicted=363.471832, ground_truth=347\n",
      "TRAIN DATA 101 : predicted=333.634827, ground_truth=305\n",
      "TRAIN DATA 102 : predicted=312.212280, ground_truth=336\n",
      "TRAIN DATA 103 : predicted=380.491699, ground_truth=340\n",
      "TRAIN DATA 104 : predicted=357.297913, ground_truth=318\n",
      "TRAIN DATA 105 : predicted=324.329803, ground_truth=362\n",
      "TRAIN DATA 106 : predicted=371.745819, ground_truth=348\n",
      "TRAIN DATA 107 : predicted=336.607239, ground_truth=363\n",
      "TRAIN DATA 108 : predicted=376.518280, ground_truth=435\n",
      "TRAIN DATA 109 : predicted=441.095856, ground_truth=491\n",
      "TRAIN DATA 110 : predicted=495.493561, ground_truth=505\n",
      "TRAIN DATA 111 : predicted=487.961395, ground_truth=404\n",
      "TRAIN DATA 112 : predicted=347.424835, ground_truth=359\n",
      "TRAIN DATA 113 : predicted=352.676575, ground_truth=310\n",
      "TRAIN DATA 114 : predicted=311.335419, ground_truth=337\n"
     ]
    }
   ],
   "source": [
    "# check how exactly fit train data\n",
    "\n",
    "y_hats = []\n",
    "cur_y_hat = 0\n",
    "for i in range(115):\n",
    "    x_input = X_train[i]\n",
    "    x_input_reshape = x_input.reshape(1,5,1)\n",
    "    cur_y_hat = model.predict(x_input_reshape, verbose=0)\n",
    "    cur_y_hat = np.squeeze(cur_y_hat)\n",
    "    print('TRAIN DATA %d : predicted=%f, ground_truth=%d' % (i, cur_y_hat, y_train[i]))\n",
    "    y_hats.append(cur_y_hat)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 1.0, 'RMSE: 25.6209')"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(y_train)\n",
    "plt.plot(y_hats)\n",
    "plt.title('RMSE: %.4f'% np.sqrt(sum((y_hats-y_train)**2)/len(y_train)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "TEST DATA 0 : predicted=393.034576, ground_truth=360\n",
      "TEST DATA 1 : predicted=422.501343, ground_truth=342\n",
      "TEST DATA 2 : predicted=433.603912, ground_truth=406\n",
      "TEST DATA 3 : predicted=419.511322, ground_truth=396\n",
      "TEST DATA 4 : predicted=395.186554, ground_truth=420\n",
      "TEST DATA 5 : predicted=383.413300, ground_truth=472\n",
      "TEST DATA 6 : predicted=385.134094, ground_truth=548\n",
      "TEST DATA 7 : predicted=396.485321, ground_truth=559\n",
      "TEST DATA 8 : predicted=409.807343, ground_truth=463\n",
      "TEST DATA 9 : predicted=416.550446, ground_truth=407\n",
      "TEST DATA 10 : predicted=416.018311, ground_truth=362\n",
      "TEST DATA 11 : predicted=411.737640, ground_truth=405\n",
      "TEST DATA 12 : predicted=407.896912, ground_truth=417\n",
      "TEST DATA 13 : predicted=407.626373, ground_truth=391\n",
      "TEST DATA 14 : predicted=410.824188, ground_truth=419\n",
      "TEST DATA 15 : predicted=415.332642, ground_truth=461\n",
      "TEST DATA 16 : predicted=419.016876, ground_truth=472\n",
      "TEST DATA 17 : predicted=420.783997, ground_truth=535\n",
      "TEST DATA 18 : predicted=420.998627, ground_truth=622\n",
      "TEST DATA 19 : predicted=420.848022, ground_truth=606\n",
      "TEST DATA 20 : predicted=421.337677, ground_truth=508\n",
      "TEST DATA 21 : predicted=422.800293, ground_truth=461\n",
      "TEST DATA 22 : predicted=424.900330, ground_truth=390\n",
      "TEST DATA 23 : predicted=427.019836, ground_truth=432\n"
     ]
    }
   ],
   "source": [
    "# predict with trained model\n",
    "\n",
    "y_hats = []\n",
    "cur_y_hat = 0\n",
    "for i in range(24):\n",
    "    if i == 0:  # test data 의 맨 첫번째 row에 대해서만 X 값을 그대로 활용\n",
    "        x_input = X_test[0]\n",
    "    else:       # 두번째 row부터는 이전 row의 prediction을 반영하여 X를 재구성\n",
    "        x_input = x_input.tolist()\n",
    "        del x_input[0]\n",
    "        x_input.append(cur_y_hat)\n",
    "        x_input = np.array(x_input)\n",
    "    x_input_reshape = x_input.reshape(1,5,1)\n",
    "    cur_y_hat = model.predict(x_input_reshape, verbose=0)\n",
    "    cur_y_hat = np.squeeze(cur_y_hat)\n",
    "    print('TEST DATA %d : predicted=%f, ground_truth=%d' % (i, cur_y_hat, y_test[i]))\n",
    "    y_hats.append(cur_y_hat)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 1.0, 'RMSE: 86.4652')"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(y_test)\n",
    "plt.plot(y_hats)\n",
    "plt.title('RMSE: %.4f'% np.sqrt(sum((y_hats-y_test)**2)/len(y_test)))"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "tf13",
   "language": "python",
   "name": "tf13"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}