diff --git a/weight_initialization.ipynb b/weight_initialization.ipynb
new file mode 100644
index 0000000..155a008
--- /dev/null
+++ b/weight_initialization.ipynb
@@ -0,0 +1,2659 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": 23,
+   "id": "5a8f46e4-e7a3-4014-ba70-6a9cf844c87c",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import numpy as np\n",
+    "import matplotlib.pyplot as plt\n",
+    "import tensorflow as tf\n",
+    "from tensorflow import keras\n",
+    "from sklearn.metrics import mean_squared_error, mean_absolute_error"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 17,
+   "id": "b2b3a27f-daaa-4a78-8cfd-3d16a3cb1754",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "image/png": "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",
+      "text/plain": [
+       "<Figure size 640x480 with 1 Axes>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    }
+   ],
+   "source": [
+    "#Glorot initialization: range of values for the weights as a function of n\n",
+    "values=[i for i in range(1,101)]\n",
+    "m=10 #number of output units\n",
+    "results_glorot=[np.sqrt(6/(i+m)) for i in values]\n",
+    "plt.errorbar(values,[0.0 for _ in values], yerr=results_glorot)\n",
+    "plt.title(\"Glorot initialization\")\n",
+    "plt.xlabel(\"n\") # number of input units\n",
+    "plt.show()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 18,
+   "id": "3b757f89-ced1-476e-97c8-1d46a4bdf255",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAiIAAAHFCAYAAADPHZKNAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjguMCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy81sbWrAAAACXBIWXMAAA9hAAAPYQGoP6dpAAAgt0lEQVR4nO3de5BU5Z344W/DMM1FmIgjl5HhZiyVRbyAJlEjo2aNBtzKXqyYjQGMxKBCUHYTRV25VHRI4lrJGiXiGnDXZLVSQddVKwlJgGhpFBASTFwTVhDCwLLKhsFLBmHO748t+uc4QIZhet7pmeep6ir69Ol+33mLYj6cc7o7l2VZFgAACXRLPQEAoOsSIgBAMkIEAEhGiAAAyQgRACAZIQIAJCNEAIBkhAgAkIwQAQCSESKQ0JIlSyKXy8Xq1asP+PjEiRNj+PDhbT7epk2bDvu5mzZtilwuF0uWLClse/bZZ2Pu3Lnxhz/8odn+NTU1UVNT06p5Hui5uVwu5s6d26rX+1Pq6upi7ty5sW7dumaPzZ07N3K5XFHGBSLKUk8AaD8TJkyI5557LgYPHnzYzx08eHA899xzcfzxxxe2PfvsszFv3ryYMmVKfOADH2iy/7333nuk023iueeeiyFDhrTpa+5XV1cX8+bNi+HDh8dpp53W5LGpU6fGxRdfXJRxASECXcqxxx4bxx57bKuem8/n48Mf/nCL9x81alSrxjmYwxm7LQ0ZMqRoAQQ4NQMlJ8uyuPfee+O0006LXr16xdFHHx1/8zd/E6+++uqffO6BTs3U1NTE6NGjY9WqVfHRj340evfuHSNHjowFCxZEY2NjYb/3n5qZO3dufOlLX4qIiBEjRkQul4tcLhcrVqwovO77T6/MmzcvPvShD0X//v2jX79+ccYZZ8QDDzwQLfnuzfefmhk+fHhhzPff9s9hw4YNceWVV8YJJ5wQvXv3juOOOy4uvfTSWL9+feF1VqxYEWeeeWZERFx55ZWF19g/1oFOzTQ2NsbXvva1OOmkkyKfz8eAAQNi0qRJ8fvf/77Jfi1dW+jKHBGBDmDfvn2xd+/eZtsP9Av6C1/4QixZsiS++MUvxle/+tXYuXNnzJ8/P84+++z45S9/GQMHDjzs8bdv3x6f+cxn4u/+7u9izpw58eijj8bs2bOjqqoqJk2adMDnTJ06NXbu3Bl33313LF26tHC651BHQjZt2hRf+MIXYujQoRER8Ytf/CJmzJgRW7dujdtuu+2w5vzoo49GQ0ND4X5jY2NMmzYtXn311cLr19XVxTHHHBMLFiyIY489Nnbu3BkPPvhgfOhDH4q1a9fGiSeeGGeccUYsXrw4rrzyyrj11ltjwoQJERGHPApyzTXXxKJFi2L69OkxceLE2LRpU/zDP/xDrFixIl588cWorKws7NuatYUuJQOSWbx4cRYRh7wNGzassP9zzz2XRUT2j//4j01eZ8uWLVmvXr2yL3/5yy0ab+PGjYVt48ePzyIie/7555vsO2rUqOzjH/944f7GjRuziMgWL15c2Pb1r3+92eu993XHjx9/0Lns27cve/fdd7P58+dnxxxzTNbY2HjI50ZENmfOnIO+3vTp07OysrLsqaeeOug+e/fuzfbs2ZOdcMIJ2Q033FDYvmrVqmY/235z5szJ3vtP5csvv5xFRHbttdc22e/555/PIiK7+eabm/wcLVlb6MqcmoEO4F/+5V9i1apVzW7nnntuk/2eeOKJyOVyccUVV8TevXsLt0GDBsWpp55aOCVxuAYNGhRnnXVWk21jxoyJ1157rbU/0gH97Gc/i4997GNRUVER3bt3jx49esRtt90Wb7zxRuzYsaPVr7tgwYL41re+Fd/+9rfjkksuKWzfu3dv3HHHHTFq1KgoLy+PsrKyKC8vj9/97nfx8ssvt2qs5cuXR0TElClTmmw/66yz4uSTT46f/vSnTba319pCqXJqBjqAk08+OcaNG9dse0VFRWzZsqVw/7//+78jy7KDnn4ZOXJkq8Y/5phjmm3L5/PxzjvvtOr1DuSFF16Iiy66KGpqauL++++PIUOGRHl5eTz22GNx++23t3qshx56KG6++ea47bbb4qqrrmry2KxZs+Kee+6JG2+8McaPHx9HH310dOvWLaZOndrq8d54442IiAO+86iqqqpZYLTH2kIpEyJQQiorKyOXy8XTTz8d+Xy+2eMH2tZRPPzww9GjR4944oknomfPnoXtjz32WKtfc9myZfG5z30upkyZEvPmzWv2+EMPPRSTJk2KO+64o8n2119/vdnbjVtqf1hs27at2XUkdXV1Ta4PAf40p2aghEycODGyLIutW7fGuHHjmt1OOeWUdp3P/vBpyf/uc7lclJWVRffu3Qvb3nnnnfjXf/3XVo29bt26+Ou//uu44IILYtGiRQcd8/1x9uSTT8bWrVubbDucn+OCCy6IiP+LnPdatWpVvPzyy3HhhRe2+GcAHBGBknLOOefE1VdfHVdeeWWsXr06zjvvvOjTp09s27YtnnnmmTjllFPimmuuabf57A+fb37zmzF58uTo0aNHnHjiidG3b99m+06YMCHuuuuu+Nu//du4+uqr44033og777yzVUdx6uvr4xOf+ET06tUr/v7v/77ZJ9OOGjUq+vXrFxMnTowlS5bESSedFGPGjIk1a9bE17/+9WZHMo4//vjo1atXfPe7342TTz45jjrqqKiqqoqqqqpmY5944olx9dVXx9133x3dunWLSy65pPCumerq6rjhhhsO++eBrkyIQIm577774sMf/nDcd999ce+990ZjY2NUVVXFOeec0+yiyGKrqamJ2bNnx4MPPhj3339/NDY2xvLlyw/40e4XXHBBfOc734mvfvWrcemll8Zxxx0Xn//852PAgAHNru34U3bu3Bnbtm2LiIg///M/b/b4/jl885vfjB49ekRtbW28+eabccYZZ8TSpUvj1ltvbbJ/79694zvf+U7MmzcvLrroonj33Xdjzpw5B/1I+YULF8bxxx8fDzzwQNxzzz1RUVERF198cdTW1h7wmhDg4HJZ1oJPEgIAKALXiAAAyQgRACAZIQIAJCNEAIBkhAgAkIwQAQCS6dCfI9LY2Bh1dXXRt2/fyOVyqacDALRAlmWxe/fuqKqqim7dDn3Mo0OHSF1dXVRXV6eeBgDQClu2bGn2Scbv16FDZP/HRG/ZsiX69euXeDYAQEvU19dHdXX1Ab/u4f06dIjsPx3Tr18/IQIAJaYll1W4WBUASEaIAADJCBEAIBkhAgAkI0QAgGSECACQjBABAJIRIgBAMkIEAEhGiAAAyQgRACAZIQIAJCNEAIBkhAgAkEyXDJG39+yN4Tc9GcNvejLe3rM39XQAoMvqkiECAHQMQgQASEaIAADJCBEAIBkhAgAkI0QAgGSECACQjBABAJIRIgBAMkIEAEhGiAAAyQgRACAZIQIAJCNEAIBkhAgAkIwQAQCSESIAQDJCBABIRogAAMkIEQAgGSECACQjRACAZIQIAJCMEAEAkhEiAEAyRQ2R2traOPPMM6Nv374xYMCA+OQnPxmvvPJKMYcEAEpIUUNk5cqVcd1118UvfvGLWLZsWezduzcuuuiieOutt4o5LABQIsqK+eI//OEPm9xfvHhxDBgwINasWRPnnXdeMYcGAEpAu14jsmvXroiI6N+/f3sOCwB0UEU9IvJeWZbFrFmz4txzz43Ro0cfcJ+GhoZoaGgo3K+vr2+v6QEACbTbEZHp06fHr371q/i3f/u3g+5TW1sbFRUVhVt1dXV7TQ8ASKBdQmTGjBnx+OOPx/Lly2PIkCEH3W/27Nmxa9euwm3Lli3tMT0AIJGinprJsixmzJgRjz76aKxYsSJGjBhxyP3z+Xzk8/liTgkA6ECKGiLXXXddfO9734t///d/j759+8b27dsjIqKioiJ69epVzKEBgBJQ1FMzCxcujF27dkVNTU0MHjy4cHvkkUeKOSwAUCKKfmoGAOBgfNcMAJCMEAEAkhEiAEAyQgQASEaIAADJCBEAIBkhAgAkI0QAgGSECACQjBABAJIRIgBAMkIEAEhGiETE23v2xvCbnozhNz0Zb+/Zm3o6ANBlCBEAIBkhAgAkI0QAgGSECACQjBABAJIRIgBAMkIEAEhGiAAAyQgRACAZIQIAJCNEAIBkhAgAkIwQAQCSESIAQDJCBABIRogAAMkIEQAgGSECACQjRACAZIQIAJCMEAEAkhEiB/H2nr0x/KYnY/hNT8bbe/amng4AdEpCBABIRogAAMkIEQAgGSECACQjRACAZIQIAJCMEAEAkhEiAEAyQgQASEaIAADJCBEAIBkhAgAkI0RayJfgAUDbEyIAQDJCBABIRogAAMkIEQAgGSECACQjRACAZIQIAJCMEAEAkhEiR8CHnAHAkREiAEAyRQ2Rn//853HppZdGVVVV5HK5eOyxx4o5HABQYooaIm+99Vaceuqp8a1vfauYwwAAJaqsmC9+ySWXxCWXXFLMIQCAElbUEDlcDQ0N0dDQULhfX1+fcDYAQLF1qItVa2tro6KionCrrq5OPaXD5p00ANByHSpEZs+eHbt27SrctmzZknpKAEARdahTM/l8PvL5fOppAADtpEMdEQEAupaiHhF58803Y8OGDYX7GzdujHXr1kX//v1j6NChxRwaACgBRQ2R1atXx/nnn1+4P2vWrIiImDx5cixZsqSYQ3cYb+/ZG6Nu+1FERPxm/sejd3mHOhsGAEkV9bdiTU1NZFlWzCEAgBLmGhEAIBkhAgAkI0QAgGSESAI+fRUA/o8QAQCSESIAQDJCpANwqgaArkqIAADJCBEAIBkh0kE5XQNAVyBEAIBkhEiJcIQEgM5IiAAAyQiREuYoCQClTogAAMkIkU7EERIASo0QAQCSESKdnKMkAHRkQgQASEaIdEGOkgDQUQgRhAkAyQgRDkicANAehAgtIkwAKAYhQquJEwCOlBChzQgTAA6XEKGoxAkAhyJEaFfCBID3EiIkJ04Aui4hQof0/jgRKwCdkxChZIkTgNInROg0DhQmYgWgYxMidCliBaBjESLwPmIFoP0IEWglF9QCHDkhAkXUklgRMEBXJkSgg2lJrIgXoLMQItBJCBigFAkR6OJac/pI0ABtRYgAbULQAK0hRIAOpbUBI3KgNAkRoMtoq6AROdB2hAhAGyhW5IgeOjshAlBiWnvURgjREQkRAA5LMU9xiaWuR4gA0OmIpdIhRAAggY4WS6kIEQAgGSECACQjRACAZIQIAJCMEAEAkhEiAEAyQgQASEaIAADJCBEAIBkhAgAkI0QAgGSECACQTLuEyL333hsjRoyInj17xtixY+Ppp59uj2EBgA6u6CHyyCOPxPXXXx+33HJLrF27Nj760Y/GJZdcEps3by720ABAB1dW7AHuuuuuuOqqq2Lq1KkREfGNb3wjfvSjH8XChQujtra22MMfUJZlhT+//6uP3/v1yK3Zp7XPa899Uo9vjubYGX+O1OObozkeyfPe+3uxveWyIo6+Z8+e6N27d3z/+9+Pv/zLvyxsnzlzZqxbty5WrlzZZP+GhoZoaGgo3K+vr4/q6urYtWtX9OvXr83m9fqbf4xxX/lpm70eAJSy1bdeGJVH9Wyz16uvr4+KiooW/f4u6qmZ119/Pfbt2xcDBw5ssn3gwIGxffv2ZvvX1tZGRUVF4VZdXV3M6QEAiRX91ExERC6Xa3I/y7Jm2yIiZs+eHbNmzSrc339EpK316tG98OfVt14YEVE4QrL61gujd3lZvL1nb5NtLdmntc9rz31Sj2+O5tgZf47U45ujOR7p8977e7G9FTVEKisro3v37s2OfuzYsaPZUZKIiHw+H/l8vphTioimYdS7vOkS9C4vO+C2luzT2ue15z6pxzdHc+yMP0fq8c3RHI/0eQc6ONBeinpqpry8PMaOHRvLli1rsn3ZsmVx9tlnF3NoAKAEFP3UzKxZs+Kzn/1sjBs3Lj7ykY/EokWLYvPmzTFt2rRiDw0AdHBFD5FPfepT8cYbb8T8+fNj27ZtMXr06Hjqqadi2LBhxR4aAOjg2uVi1WuvvTauvfba9hgKACghvmsGAEhGiAAAyQgRACAZIQIAJCNEAIBkhAgAkIwQAQCSESIAQDJCBABIRogAAMm0y0e8A0BX1ru8LDYtmPAnt7XVPq19XgpCBICSUsxf4J3tl3wpECIAHFRH/J88nYsQAejA2jIE/E+ejkiIABym9j5KAJ2ZEAE6pWLGAtB2hAiQTEtjwZEE6LyECHDYxALQVoQIdHECAkhJiEAJastTGgApCRHoYFoaFACdgRCBInLaA+DQhAi0ktMeAEdOiMD7ODUC0H6ECF2KyADoWIQInYbIACg9QoSSJTIASp8QoUNyIShA1yBESM6RDYCuS4jQrkQHAO8lRGgzLhYF4HAJEVpNZABwpLqlngAA0HU5IkKLOPoBQDEIEQ5IeADQHoQIogOAZIRIFyQ8AOgoXKwKACTjiEgn5+gHAB2ZIyIAQDKOiHQijn4AUGqESAkTHgCUOqdmAIBkhAgAkIxTMyXCaRgAOiNHRACAZBwR6aAcAQGgK3BEBABIRogAAMk4NdMBOA0DQFfliAgAkIwQAQCScWomAadiAOD/OCICACQjRACAZIQIAJCMa0SKzPUgAHBwRT0icvvtt8fZZ58dvXv3jg984APFHAoAKEFFDZE9e/bEZZddFtdcc00xhwEASlRRT83MmzcvIiKWLFlSzGEAgBLlYlUAIJkOdbFqQ0NDNDQ0FO7X19cnnA0AUGyHfURk7ty5kcvlDnlbvXp1qyZTW1sbFRUVhVt1dXWrXiel/e+S2bRgQvQu71CdBwAdzmH/ppw+fXpcfvnlh9xn+PDhrZrM7NmzY9asWYX79fX1JRkjAEDLHHaIVFZWRmVlZTHmEvl8PvL5fFFeGwDoeIp67mDz5s2xc+fO2Lx5c+zbty/WrVsXEREf/OAH46ijjirm0ABACShqiNx2223x4IMPFu6ffvrpERGxfPnyqKmpKebQAEAJKOrbd5csWRJZljW7iRAAIKKDvX231PgeGQA4Mj7QDABIRogAAMkIEQAgGSECACQjRACAZIQIAJCMEAEAkhEiAEAyPtCshXx4GQC0PUdEAIBkhAgAkIwQAQCSESIAQDJCBABIRogAAMkIEQAgGSECACQjRACAZIQIAJCMEAEAkhEiAEAyvvTuIHzJHQAUnyMiAEAyQgQASEaIAADJCBEAIBkhAgAkI0QAgGSECACQjBABAJIRIgBAMkIEAEhGiAAAyQgRACAZIQIAJCNEAIBkhAgAkIwQAQCSESIAQDJCBABIRogAAMkIEQAgGSECACRTlnoCHUHv8rLYtGBC6mkAQJfjiAgAkIwQAQCSESIAQDJCBABIRogAAMkIEQAgGSECACQjRACAZIQIAJCMEAEAkhEiAEAyQgQASEaIAADJFC1ENm3aFFdddVWMGDEievXqFccff3zMmTMn9uzZU6whAYASU1asF/7P//zPaGxsjPvuuy8++MEPxksvvRSf//zn46233oo777yzWMMCACWkaCFy8cUXx8UXX1y4P3LkyHjllVdi4cKFQgQAiIgihsiB7Nq1K/r373/QxxsaGqKhoaFwv76+vj2mBQAk0m4Xq/7Xf/1X3H333TFt2rSD7lNbWxsVFRWFW3V1dXtNDwBI4LBDZO7cuZHL5Q55W716dZPn1NXVxcUXXxyXXXZZTJ069aCvPXv27Ni1a1fhtmXLlsP/iQCAknHYp2amT58el19++SH3GT58eOHPdXV1cf7558dHPvKRWLRo0SGfl8/nI5/PH+6UAIASddghUllZGZWVlS3ad+vWrXH++efH2LFjY/HixdGtm48tAQD+v6JdrFpXVxc1NTUxdOjQuPPOO+N//ud/Co8NGjSoWMMCACWkaCHy4x//ODZs2BAbNmyIIUOGNHksy7JiDQsAlJCinSuZMmVKZFl2wBsAQITvmgEAEhIiAEAyQgQASEaIAADJCBEAIBkhAgAkI0QAgGSECACQjBABAJIRIgBAMkIEAEhGiAAAyQgRACAZIQIAJCNEAIBkhAgAkIwQAQCSESIAQDJCBABIRogAAMkIEQAgGSECACQjRACAZIQIAJBMWeoJpNC7vCw2LZiQehoA0OU5IgIAJCNEAIBkhAgAkIwQAQCSESIAQDJCBABIRogAAMkIEQAgGSECACQjRACAZIQIAJCMEAEAkhEiAEAyQgQASEaIAADJlKWewKFkWRYREfX19YlnAgC01P7f2/t/jx9Khw6R3bt3R0REdXV14pkAAIdr9+7dUVFRcch9cllLciWRxsbGqKuri759+0Yul2v169TX10d1dXVs2bIl+vXr14Yz5P2sdfux1u3HWrcv691+irXWWZbF7t27o6qqKrp1O/RVIB36iEi3bt1iyJAhbfZ6/fr185e6nVjr9mOt24+1bl/Wu/0UY63/1JGQ/VysCgAkI0QAgGS6RIjk8/mYM2dO5PP51FPp9Kx1+7HW7cdaty/r3X46wlp36ItVAYDOrUscEQEAOiYhAgAkI0QAgGSECACQTJcIkXvvvTdGjBgRPXv2jLFjx8bTTz+dekolrba2Ns4888zo27dvDBgwID75yU/GK6+80mSfLMti7ty5UVVVFb169Yqampr49a9/nWjGnUdtbW3kcrm4/vrrC9usddvaunVrXHHFFXHMMcdE796947TTTos1a9YUHrfebWPv3r1x6623xogRI6JXr14xcuTImD9/fjQ2Nhb2sdat8/Of/zwuvfTSqKqqilwuF4899liTx1uyrg0NDTFjxoyorKyMPn36xF/8xV/E73//++JMOOvkHn744axHjx7Z/fffn/3mN7/JZs6cmfXp0yd77bXXUk+tZH384x/PFi9enL300kvZunXrsgkTJmRDhw7N3nzzzcI+CxYsyPr27Zv94Ac/yNavX5996lOfygYPHpzV19cnnHlpe+GFF7Lhw4dnY8aMyWbOnFnYbq3bzs6dO7Nhw4ZlU6ZMyZ5//vls48aN2U9+8pNsw4YNhX2sd9v4yle+kh1zzDHZE088kW3cuDH7/ve/nx111FHZN77xjcI+1rp1nnrqqeyWW27JfvCDH2QRkT366KNNHm/Juk6bNi077rjjsmXLlmUvvvhidv7552ennnpqtnfv3jafb6cPkbPOOiubNm1ak20nnXRSdtNNNyWaUeezY8eOLCKylStXZlmWZY2NjdmgQYOyBQsWFPb54x//mFVUVGTf/va3U02zpO3evTs74YQTsmXLlmXjx48vhIi1bls33nhjdu655x70cevddiZMmJB97nOfa7Ltr/7qr7IrrrgiyzJr3VbeHyItWdc//OEPWY8ePbKHH364sM/WrVuzbt26ZT/84Q/bfI6d+tTMnj17Ys2aNXHRRRc12X7RRRfFs88+m2hWnc+uXbsiIqJ///4REbFx48bYvn17k3XP5/Mxfvx4695K1113XUyYMCE+9rGPNdlurdvW448/HuPGjYvLLrssBgwYEKeffnrcf//9hcetd9s599xz46c//Wn89re/jYiIX/7yl/HMM8/EJz7xiYiw1sXSknVds2ZNvPvuu032qaqqitGjRxdl7Tv0l94dqddffz327dsXAwcObLJ94MCBsX379kSz6lyyLItZs2bFueeeG6NHj46IKKztgdb9tddea/c5lrqHH344XnzxxVi1alWzx6x123r11Vdj4cKFMWvWrLj55pvjhRdeiC9+8YuRz+dj0qRJ1rsN3XjjjbFr16446aSTonv37rFv3764/fbb49Of/nRE+LtdLC1Z1+3bt0d5eXkcffTRzfYpxu/OTh0i++VyuSb3syxrto3WmT59evzqV7+KZ555ptlj1v3IbdmyJWbOnBk//vGPo2fPngfdz1q3jcbGxhg3blzccccdERFx+umnx69//etYuHBhTJo0qbCf9T5yjzzySDz00EPxve99L/7sz/4s1q1bF9dff31UVVXF5MmTC/tZ6+JozboWa+079amZysrK6N69e7OC27FjR7Ma5PDNmDEjHn/88Vi+fHkMGTKksH3QoEEREda9DaxZsyZ27NgRY8eOjbKysigrK4uVK1fGP/3TP0VZWVlhPa112xg8eHCMGjWqybaTTz45Nm/eHBH+brelL33pS3HTTTfF5ZdfHqecckp89rOfjRtuuCFqa2sjwloXS0vWddCgQbFnz5743//934Pu05Y6dYiUl5fH2LFjY9myZU22L1u2LM4+++xEsyp9WZbF9OnTY+nSpfGzn/0sRowY0eTxESNGxKBBg5qs+549e2LlypXW/TBdeOGFsX79+li3bl3hNm7cuPjMZz4T69ati5EjR1rrNnTOOec0eyv6b3/72xg2bFhE+Lvdlt5+++3o1q3pr6Du3bsX3r5rrYujJes6duzY6NGjR5N9tm3bFi+99FJx1r7NL3/tYPa/ffeBBx7IfvOb32TXX3991qdPn2zTpk2pp1ayrrnmmqyioiJbsWJFtm3btsLt7bffLuyzYMGCrKKiIlu6dGm2fv367NOf/rS33bWR975rJsusdVt64YUXsrKysuz222/Pfve732Xf/e53s969e2cPPfRQYR/r3TYmT56cHXfccYW37y5dujSrrKzMvvzlLxf2sdats3v37mzt2rXZ2rVrs4jI7rrrrmzt2rWFj61oybpOmzYtGzJkSPaTn/wke/HFF7MLLrjA23ePxD333JMNGzYsKy8vz84444zC20xpnYg44G3x4sWFfRobG7M5c+ZkgwYNyvL5fHbeeedl69evTzfpTuT9IWKt29Z//Md/ZKNHj87y+Xx20kknZYsWLWryuPVuG/X19dnMmTOzoUOHZj179sxGjhyZ3XLLLVlDQ0NhH2vdOsuXLz/gv9GTJ0/Osqxl6/rOO+9k06dPz/r375/16tUrmzhxYrZ58+aizDeXZVnW9sdZAAD+tE59jQgA0LEJEQAgGSECACQjRACAZIQIAJCMEAEAkhEiAEAyQgQASEaIAADJCBEAIJmy1BMAupaampoYM2ZM9OzZM/75n/85ysvLY9q0aTF37tzUUwMScEQEaHcPPvhg9OnTJ55//vn42te+FvPnz2/yleNA1+FL74B2VVNTE/v27Yunn366sO2ss86KCy64IBYsWJBwZkAKjogA7W7MmDFN7g8ePDh27NiRaDZASkIEaHc9evRocj+Xy0VjY2Oi2QApCREAIBkhAgAkI0QAgGS8awYASMYREQAgGSECACQjRACAZIQIAJCMEAEAkhEiAEAyQgQASEaIAADJCBEAIBkhAgAkI0QAgGSECACQzP8DWhGLR4T7/mIAAAAASUVORK5CYII=",
+      "text/plain": [
+       "<Figure size 640x480 with 1 Axes>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    }
+   ],
+   "source": [
+    "#He initialization\n",
+    "results_he=[np.sqrt(6.0/i) for i in values]\n",
+    "plt.errorbar(values,[0.0 for _ in values], yerr=results_he)\n",
+    "plt.title(\"He initialization\")\n",
+    "plt.xlabel(\"n\")\n",
+    "plt.show()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 24,
+   "id": "f7b0e872-947d-4fd3-9ab8-4b26fa72fe1a",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "folder='/home/unipi/v.vichi3/Desktop/'\n",
+    "X_train, X_val, X_test, y_train, y_val, y_test=np.load(folder+'X_train.npy'), np.load(folder+'X_val.npy'), np.load(folder+'X_test.npy'), np.load(folder+'y_train.npy'), np.load(folder+'y_val.npy'), np.load(folder+'y_test.npy')"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 25,
+   "id": "b6c28a1e-50b0-47ef-bb57-d3821542e8a4",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "#Choose the most appropriate standard deviation for the Random Normal initializer\n",
+    "rand_norm=[keras.initializers.RandomNormal(mean=0.0,stddev=1),keras.initializers.RandomNormal(mean=0.0,stddev=0.1),keras.initializers.RandomNormal(mean=0.0,stddev=0.01),keras.initializers.RandomNormal(mean=0.0,stddev=0.001)]\n",
+    "models=np.zeros_like(rand_norm)\n",
+    "for i in range(len(models)):\n",
+    "  models[i]=keras.models.Sequential()\n",
+    "  models[i].add(keras.layers.Dense(units=32, activation='relu', input_dim=X_train.shape[1], kernel_initializer=rand_norm[i]))\n",
+    "  models[i].add(keras.layers.Dense(units=32, activation='sigmoid', kernel_initializer=rand_norm[i]))\n",
+    "  models[i].add(keras.layers.Dense(units=64, activation='sigmoid', kernel_initializer=rand_norm[i]))\n",
+    "  models[i].add(keras.layers.Dense(units=1, activation='relu', kernel_initializer=rand_norm[i]))\n",
+    "  models[i].compile(optimizer='adam',\n",
+    "                    loss='mean_squared_error',\n",
+    "                    metrics=['mean_absolute_error'])"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 4,
+   "id": "0478e1aa-8975-4ed4-97e2-f8e7d65f09ec",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "2024-05-06 20:06:06.530048: I tensorflow/compiler/mlir/mlir_graph_optimization_pass.cc:116] None of the MLIR optimization passes are enabled (registered 2)\n",
+      "2024-05-06 20:06:06.559170: I tensorflow/core/platform/profile_utils/cpu_utils.cc:112] CPU Frequency: 3892860000 Hz\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Epoch 1/50\n",
+      "18750/18750 [==============================] - 10s 534us/step - loss: 0.2688 - mean_absolute_error: 0.2058 - val_loss: 0.0168 - val_mean_absolute_error: 0.1058\n",
+      "Epoch 2/50\n",
+      "18750/18750 [==============================] - 6s 347us/step - loss: 0.0170 - mean_absolute_error: 0.1065 - val_loss: 0.0148 - val_mean_absolute_error: 0.0984\n",
+      "Epoch 3/50\n",
+      "18750/18750 [==============================] - 6s 346us/step - loss: 0.0150 - mean_absolute_error: 0.0994 - val_loss: 0.0149 - val_mean_absolute_error: 0.0999\n",
+      "Epoch 4/50\n",
+      "18750/18750 [==============================] - 6s 343us/step - loss: 0.0146 - mean_absolute_error: 0.0979 - val_loss: 0.0152 - val_mean_absolute_error: 0.1009\n",
+      "Epoch 5/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 0.0145 - mean_absolute_error: 0.0971 - val_loss: 0.0144 - val_mean_absolute_error: 0.0962\n",
+      "Epoch 6/50\n",
+      "18750/18750 [==============================] - 6s 346us/step - loss: 0.0137 - mean_absolute_error: 0.0930 - val_loss: 0.0116 - val_mean_absolute_error: 0.0837\n",
+      "Epoch 7/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 0.0113 - mean_absolute_error: 0.0825 - val_loss: 0.0113 - val_mean_absolute_error: 0.0814\n",
+      "Epoch 8/50\n",
+      "18750/18750 [==============================] - 6s 345us/step - loss: 0.0107 - mean_absolute_error: 0.0794 - val_loss: 0.0102 - val_mean_absolute_error: 0.0763\n",
+      "Epoch 9/50\n",
+      "18750/18750 [==============================] - 6s 345us/step - loss: 0.0104 - mean_absolute_error: 0.0777 - val_loss: 0.0100 - val_mean_absolute_error: 0.0758\n",
+      "Epoch 10/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 0.0101 - mean_absolute_error: 0.0762 - val_loss: 0.0097 - val_mean_absolute_error: 0.0740\n",
+      "Epoch 11/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 0.0098 - mean_absolute_error: 0.0749 - val_loss: 0.0096 - val_mean_absolute_error: 0.0750\n",
+      "Epoch 12/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 0.0096 - mean_absolute_error: 0.0740 - val_loss: 0.0095 - val_mean_absolute_error: 0.0734\n",
+      "Epoch 13/50\n",
+      "18750/18750 [==============================] - 6s 345us/step - loss: 0.0089 - mean_absolute_error: 0.0707 - val_loss: 0.0084 - val_mean_absolute_error: 0.0677\n",
+      "Epoch 14/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 0.0083 - mean_absolute_error: 0.0678 - val_loss: 0.0077 - val_mean_absolute_error: 0.0643\n",
+      "Epoch 15/50\n",
+      "18750/18750 [==============================] - 6s 346us/step - loss: 0.0080 - mean_absolute_error: 0.0661 - val_loss: 0.0080 - val_mean_absolute_error: 0.0667\n",
+      "Epoch 16/50\n",
+      "18750/18750 [==============================] - 6s 345us/step - loss: 0.0077 - mean_absolute_error: 0.0650 - val_loss: 0.0077 - val_mean_absolute_error: 0.0652\n",
+      "Epoch 17/50\n",
+      "18750/18750 [==============================] - 6s 345us/step - loss: 0.0077 - mean_absolute_error: 0.0646 - val_loss: 0.0074 - val_mean_absolute_error: 0.0629\n",
+      "Epoch 18/50\n",
+      "18750/18750 [==============================] - 6s 345us/step - loss: 0.0075 - mean_absolute_error: 0.0639 - val_loss: 0.0086 - val_mean_absolute_error: 0.0706\n",
+      "Epoch 19/50\n",
+      "18750/18750 [==============================] - 6s 343us/step - loss: 0.0074 - mean_absolute_error: 0.0633 - val_loss: 0.0070 - val_mean_absolute_error: 0.0607\n",
+      "Epoch 20/50\n",
+      "18750/18750 [==============================] - 6s 343us/step - loss: 0.0071 - mean_absolute_error: 0.0619 - val_loss: 0.0066 - val_mean_absolute_error: 0.0590\n",
+      "Epoch 21/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 0.0068 - mean_absolute_error: 0.0604 - val_loss: 0.0067 - val_mean_absolute_error: 0.0599\n",
+      "Epoch 22/50\n",
+      "18750/18750 [==============================] - 6s 342us/step - loss: 0.0066 - mean_absolute_error: 0.0593 - val_loss: 0.0064 - val_mean_absolute_error: 0.0574\n",
+      "Epoch 23/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 0.0064 - mean_absolute_error: 0.0583 - val_loss: 0.0066 - val_mean_absolute_error: 0.0611\n",
+      "Epoch 24/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 0.0063 - mean_absolute_error: 0.0581 - val_loss: 0.0060 - val_mean_absolute_error: 0.0559\n",
+      "Epoch 25/50\n",
+      "18750/18750 [==============================] - 6s 343us/step - loss: 0.0062 - mean_absolute_error: 0.0574 - val_loss: 0.0069 - val_mean_absolute_error: 0.0618\n",
+      "Epoch 26/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 0.0061 - mean_absolute_error: 0.0570 - val_loss: 0.0058 - val_mean_absolute_error: 0.0549\n",
+      "Epoch 27/50\n",
+      "18750/18750 [==============================] - 6s 346us/step - loss: 0.0060 - mean_absolute_error: 0.0564 - val_loss: 0.0058 - val_mean_absolute_error: 0.0549\n",
+      "Epoch 28/50\n",
+      "18750/18750 [==============================] - 6s 342us/step - loss: 0.0059 - mean_absolute_error: 0.0556 - val_loss: 0.0056 - val_mean_absolute_error: 0.0537\n",
+      "Epoch 29/50\n",
+      "18750/18750 [==============================] - 6s 342us/step - loss: 0.0058 - mean_absolute_error: 0.0550 - val_loss: 0.0059 - val_mean_absolute_error: 0.0561\n",
+      "Epoch 30/50\n",
+      "18750/18750 [==============================] - 6s 346us/step - loss: 0.0056 - mean_absolute_error: 0.0543 - val_loss: 0.0053 - val_mean_absolute_error: 0.0520\n",
+      "Epoch 31/50\n",
+      "18750/18750 [==============================] - 6s 345us/step - loss: 0.0054 - mean_absolute_error: 0.0530 - val_loss: 0.0052 - val_mean_absolute_error: 0.0526\n",
+      "Epoch 32/50\n",
+      "18750/18750 [==============================] - 6s 346us/step - loss: 0.0050 - mean_absolute_error: 0.0512 - val_loss: 0.0047 - val_mean_absolute_error: 0.0500\n",
+      "Epoch 33/50\n",
+      "18750/18750 [==============================] - 6s 346us/step - loss: 0.0047 - mean_absolute_error: 0.0495 - val_loss: 0.0045 - val_mean_absolute_error: 0.0475\n",
+      "Epoch 34/50\n",
+      "18750/18750 [==============================] - 6s 346us/step - loss: 0.0045 - mean_absolute_error: 0.0483 - val_loss: 0.0041 - val_mean_absolute_error: 0.0458\n",
+      "Epoch 35/50\n",
+      "18750/18750 [==============================] - 7s 349us/step - loss: 0.0044 - mean_absolute_error: 0.0478 - val_loss: 0.0047 - val_mean_absolute_error: 0.0497\n",
+      "Epoch 36/50\n",
+      "18750/18750 [==============================] - 6s 346us/step - loss: 0.0043 - mean_absolute_error: 0.0471 - val_loss: 0.0038 - val_mean_absolute_error: 0.0443\n",
+      "Epoch 37/50\n",
+      "18750/18750 [==============================] - 6s 346us/step - loss: 0.0041 - mean_absolute_error: 0.0465 - val_loss: 0.0039 - val_mean_absolute_error: 0.0447\n",
+      "Epoch 38/50\n",
+      "18750/18750 [==============================] - 7s 347us/step - loss: 0.0041 - mean_absolute_error: 0.0461 - val_loss: 0.0040 - val_mean_absolute_error: 0.0446\n",
+      "Epoch 39/50\n",
+      "18750/18750 [==============================] - 6s 345us/step - loss: 0.0040 - mean_absolute_error: 0.0457 - val_loss: 0.0036 - val_mean_absolute_error: 0.0430\n",
+      "Epoch 40/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 0.0039 - mean_absolute_error: 0.0449 - val_loss: 0.0038 - val_mean_absolute_error: 0.0442\n",
+      "Epoch 41/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 0.0039 - mean_absolute_error: 0.0446 - val_loss: 0.0037 - val_mean_absolute_error: 0.0433\n",
+      "Epoch 42/50\n",
+      "18750/18750 [==============================] - 7s 348us/step - loss: 0.0038 - mean_absolute_error: 0.0441 - val_loss: 0.0035 - val_mean_absolute_error: 0.0425\n",
+      "Epoch 43/50\n",
+      "18750/18750 [==============================] - 6s 346us/step - loss: 0.0037 - mean_absolute_error: 0.0434 - val_loss: 0.0034 - val_mean_absolute_error: 0.0412\n",
+      "Epoch 44/50\n",
+      "18750/18750 [==============================] - 6s 345us/step - loss: 0.0036 - mean_absolute_error: 0.0428 - val_loss: 0.0035 - val_mean_absolute_error: 0.0421\n",
+      "Epoch 45/50\n",
+      "18750/18750 [==============================] - 6s 346us/step - loss: 0.0035 - mean_absolute_error: 0.0423 - val_loss: 0.0030 - val_mean_absolute_error: 0.0383\n",
+      "Epoch 46/50\n",
+      "18750/18750 [==============================] - 6s 345us/step - loss: 0.0033 - mean_absolute_error: 0.0410 - val_loss: 0.0035 - val_mean_absolute_error: 0.0417\n",
+      "Epoch 47/50\n",
+      "18750/18750 [==============================] - 6s 345us/step - loss: 0.0032 - mean_absolute_error: 0.0403 - val_loss: 0.0039 - val_mean_absolute_error: 0.0448\n",
+      "Epoch 48/50\n",
+      "18750/18750 [==============================] - 6s 345us/step - loss: 0.0030 - mean_absolute_error: 0.0392 - val_loss: 0.0028 - val_mean_absolute_error: 0.0380\n",
+      "Epoch 49/50\n",
+      "18750/18750 [==============================] - 6s 345us/step - loss: 0.0029 - mean_absolute_error: 0.0379 - val_loss: 0.0025 - val_mean_absolute_error: 0.0346\n",
+      "Epoch 50/50\n",
+      "18750/18750 [==============================] - 6s 345us/step - loss: 0.0027 - mean_absolute_error: 0.0367 - val_loss: 0.0024 - val_mean_absolute_error: 0.0350\n",
+      "Epoch 1/50\n",
+      "18750/18750 [==============================] - 7s 349us/step - loss: 0.0489 - mean_absolute_error: 0.1801 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 2/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 0.0488 - mean_absolute_error: 0.1801 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 3/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 4/50\n",
+      "18750/18750 [==============================] - 7s 348us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 5/50\n",
+      "18750/18750 [==============================] - 6s 341us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 6/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 7/50\n",
+      "18750/18750 [==============================] - 6s 343us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 8/50\n",
+      "18750/18750 [==============================] - 6s 343us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 9/50\n",
+      "18750/18750 [==============================] - 6s 340us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 10/50\n",
+      "18750/18750 [==============================] - 6s 342us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 11/50\n",
+      "18750/18750 [==============================] - 6s 342us/step - loss: 0.0492 - mean_absolute_error: 0.1808 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 12/50\n",
+      "18750/18750 [==============================] - 6s 341us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 13/50\n",
+      "18750/18750 [==============================] - 6s 342us/step - loss: 0.0491 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 14/50\n",
+      "18750/18750 [==============================] - 6s 342us/step - loss: 0.0491 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 15/50\n",
+      "18750/18750 [==============================] - 6s 341us/step - loss: 0.0490 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 16/50\n",
+      "18750/18750 [==============================] - 6s 342us/step - loss: 0.0489 - mean_absolute_error: 0.1801 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 17/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 18/50\n",
+      "18750/18750 [==============================] - 6s 342us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 19/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 20/50\n",
+      "18750/18750 [==============================] - 6s 341us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 21/50\n",
+      "18750/18750 [==============================] - 6s 342us/step - loss: 0.0489 - mean_absolute_error: 0.1800 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 22/50\n",
+      "18750/18750 [==============================] - 6s 342us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 23/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 24/50\n",
+      "18750/18750 [==============================] - 6s 343us/step - loss: 0.0489 - mean_absolute_error: 0.1800 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 25/50\n",
+      "18750/18750 [==============================] - 6s 341us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 26/50\n",
+      "18750/18750 [==============================] - 6s 342us/step - loss: 0.0490 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 27/50\n",
+      "18750/18750 [==============================] - 6s 341us/step - loss: 0.0490 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 28/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 0.0491 - mean_absolute_error: 0.1806 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 29/50\n",
+      "18750/18750 [==============================] - 6s 343us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 30/50\n",
+      "18750/18750 [==============================] - 6s 343us/step - loss: 0.0489 - mean_absolute_error: 0.1800 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 31/50\n",
+      "18750/18750 [==============================] - 6s 341us/step - loss: 0.0492 - mean_absolute_error: 0.1807 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 32/50\n",
+      "18750/18750 [==============================] - 6s 345us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 33/50\n",
+      "18750/18750 [==============================] - 6s 343us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 34/50\n",
+      "18750/18750 [==============================] - 6s 342us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 35/50\n",
+      "18750/18750 [==============================] - 6s 343us/step - loss: 0.0489 - mean_absolute_error: 0.1801 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 36/50\n",
+      "18750/18750 [==============================] - 6s 343us/step - loss: 0.0490 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 37/50\n",
+      "18750/18750 [==============================] - 6s 345us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 38/50\n",
+      "18750/18750 [==============================] - 6s 346us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 39/50\n",
+      "18750/18750 [==============================] - 6s 343us/step - loss: 0.0488 - mean_absolute_error: 0.1798 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 40/50\n",
+      "18750/18750 [==============================] - 6s 342us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 41/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 0.0490 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 42/50\n",
+      "18750/18750 [==============================] - 6s 345us/step - loss: 0.0491 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 43/50\n",
+      "18750/18750 [==============================] - 6s 341us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 44/50\n",
+      "18750/18750 [==============================] - 6s 342us/step - loss: 0.0489 - mean_absolute_error: 0.1801 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 45/50\n",
+      "18750/18750 [==============================] - 6s 342us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 46/50\n",
+      "18750/18750 [==============================] - 6s 341us/step - loss: 0.0491 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 47/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 0.0491 - mean_absolute_error: 0.1806 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 48/50\n",
+      "18750/18750 [==============================] - 6s 342us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 49/50\n",
+      "18750/18750 [==============================] - 6s 343us/step - loss: 0.0491 - mean_absolute_error: 0.1806 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 50/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 1/50\n",
+      "18750/18750 [==============================] - 7s 344us/step - loss: 0.0150 - mean_absolute_error: 0.0998 - val_loss: 0.0129 - val_mean_absolute_error: 0.0897\n",
+      "Epoch 2/50\n",
+      "18750/18750 [==============================] - 6s 345us/step - loss: 0.0119 - mean_absolute_error: 0.0854 - val_loss: 0.0091 - val_mean_absolute_error: 0.0733\n",
+      "Epoch 3/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 0.0087 - mean_absolute_error: 0.0704 - val_loss: 0.0069 - val_mean_absolute_error: 0.0617\n",
+      "Epoch 4/50\n",
+      "18750/18750 [==============================] - 6s 343us/step - loss: 0.0064 - mean_absolute_error: 0.0590 - val_loss: 0.0055 - val_mean_absolute_error: 0.0534\n",
+      "Epoch 5/50\n",
+      "18750/18750 [==============================] - 6s 343us/step - loss: 0.0053 - mean_absolute_error: 0.0525 - val_loss: 0.0044 - val_mean_absolute_error: 0.0459\n",
+      "Epoch 6/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 0.0047 - mean_absolute_error: 0.0484 - val_loss: 0.0038 - val_mean_absolute_error: 0.0430\n",
+      "Epoch 7/50\n",
+      "18750/18750 [==============================] - 6s 343us/step - loss: 0.0039 - mean_absolute_error: 0.0437 - val_loss: 0.0033 - val_mean_absolute_error: 0.0415\n",
+      "Epoch 8/50\n",
+      "18750/18750 [==============================] - 6s 346us/step - loss: 0.0031 - mean_absolute_error: 0.0394 - val_loss: 0.0058 - val_mean_absolute_error: 0.0613\n",
+      "Epoch 9/50\n",
+      "18750/18750 [==============================] - 7s 347us/step - loss: 0.0027 - mean_absolute_error: 0.0367 - val_loss: 0.0025 - val_mean_absolute_error: 0.0353\n",
+      "Epoch 10/50\n",
+      "18750/18750 [==============================] - 6s 345us/step - loss: 0.0025 - mean_absolute_error: 0.0348 - val_loss: 0.0024 - val_mean_absolute_error: 0.0352\n",
+      "Epoch 11/50\n",
+      "18750/18750 [==============================] - 6s 342us/step - loss: 0.0023 - mean_absolute_error: 0.0333 - val_loss: 0.0021 - val_mean_absolute_error: 0.0313\n",
+      "Epoch 12/50\n",
+      "18750/18750 [==============================] - 6s 342us/step - loss: 0.0020 - mean_absolute_error: 0.0316 - val_loss: 0.0030 - val_mean_absolute_error: 0.0397\n",
+      "Epoch 13/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 0.0019 - mean_absolute_error: 0.0304 - val_loss: 0.0015 - val_mean_absolute_error: 0.0275\n",
+      "Epoch 14/50\n",
+      "18750/18750 [==============================] - 6s 342us/step - loss: 0.0017 - mean_absolute_error: 0.0292 - val_loss: 0.0017 - val_mean_absolute_error: 0.0285\n",
+      "Epoch 15/50\n",
+      "18750/18750 [==============================] - 6s 345us/step - loss: 0.0016 - mean_absolute_error: 0.0281 - val_loss: 0.0012 - val_mean_absolute_error: 0.0232\n",
+      "Epoch 16/50\n",
+      "18750/18750 [==============================] - 6s 343us/step - loss: 0.0015 - mean_absolute_error: 0.0274 - val_loss: 0.0014 - val_mean_absolute_error: 0.0271\n",
+      "Epoch 17/50\n",
+      "18750/18750 [==============================] - 6s 342us/step - loss: 0.0014 - mean_absolute_error: 0.0267 - val_loss: 0.0011 - val_mean_absolute_error: 0.0239\n",
+      "Epoch 18/50\n",
+      "18750/18750 [==============================] - 6s 343us/step - loss: 0.0013 - mean_absolute_error: 0.0260 - val_loss: 0.0014 - val_mean_absolute_error: 0.0241\n",
+      "Epoch 19/50\n",
+      "18750/18750 [==============================] - 6s 343us/step - loss: 0.0013 - mean_absolute_error: 0.0253 - val_loss: 0.0017 - val_mean_absolute_error: 0.0319\n",
+      "Epoch 20/50\n",
+      "18750/18750 [==============================] - 7s 347us/step - loss: 0.0012 - mean_absolute_error: 0.0250 - val_loss: 9.8871e-04 - val_mean_absolute_error: 0.0216\n",
+      "Epoch 21/50\n",
+      "18750/18750 [==============================] - 6s 343us/step - loss: 0.0012 - mean_absolute_error: 0.0244 - val_loss: 0.0011 - val_mean_absolute_error: 0.0236\n",
+      "Epoch 22/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 0.0011 - mean_absolute_error: 0.0239 - val_loss: 0.0011 - val_mean_absolute_error: 0.0236\n",
+      "Epoch 23/50\n",
+      "18750/18750 [==============================] - 6s 343us/step - loss: 0.0011 - mean_absolute_error: 0.0236 - val_loss: 9.0454e-04 - val_mean_absolute_error: 0.0214\n",
+      "Epoch 24/50\n",
+      "18750/18750 [==============================] - 6s 343us/step - loss: 0.0011 - mean_absolute_error: 0.0231 - val_loss: 0.0014 - val_mean_absolute_error: 0.0249\n",
+      "Epoch 25/50\n",
+      "18750/18750 [==============================] - 7s 347us/step - loss: 0.0011 - mean_absolute_error: 0.0230 - val_loss: 0.0011 - val_mean_absolute_error: 0.0261\n",
+      "Epoch 26/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 9.8683e-04 - mean_absolute_error: 0.0224 - val_loss: 0.0011 - val_mean_absolute_error: 0.0246\n",
+      "Epoch 27/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 9.7194e-04 - mean_absolute_error: 0.0221 - val_loss: 7.0447e-04 - val_mean_absolute_error: 0.0185\n",
+      "Epoch 28/50\n",
+      "18750/18750 [==============================] - 6s 346us/step - loss: 9.3784e-04 - mean_absolute_error: 0.0217 - val_loss: 7.0741e-04 - val_mean_absolute_error: 0.0189\n",
+      "Epoch 29/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 9.2258e-04 - mean_absolute_error: 0.0215 - val_loss: 9.2435e-04 - val_mean_absolute_error: 0.0206\n",
+      "Epoch 30/50\n",
+      "18750/18750 [==============================] - 6s 345us/step - loss: 8.9570e-04 - mean_absolute_error: 0.0212 - val_loss: 8.0241e-04 - val_mean_absolute_error: 0.0210\n",
+      "Epoch 31/50\n",
+      "18750/18750 [==============================] - 6s 341us/step - loss: 8.8669e-04 - mean_absolute_error: 0.0210 - val_loss: 6.6892e-04 - val_mean_absolute_error: 0.0188\n",
+      "Epoch 32/50\n",
+      "18750/18750 [==============================] - 6s 345us/step - loss: 0.0011 - mean_absolute_error: 0.0228 - val_loss: 0.0011 - val_mean_absolute_error: 0.0238\n",
+      "Epoch 33/50\n",
+      "18750/18750 [==============================] - 6s 345us/step - loss: 0.0011 - mean_absolute_error: 0.0230 - val_loss: 0.0011 - val_mean_absolute_error: 0.0243\n",
+      "Epoch 34/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 0.0010 - mean_absolute_error: 0.0222 - val_loss: 8.0839e-04 - val_mean_absolute_error: 0.0197\n",
+      "Epoch 35/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 9.2916e-04 - mean_absolute_error: 0.0215 - val_loss: 9.6287e-04 - val_mean_absolute_error: 0.0211\n",
+      "Epoch 36/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 9.0594e-04 - mean_absolute_error: 0.0213 - val_loss: 8.0006e-04 - val_mean_absolute_error: 0.0201\n",
+      "Epoch 37/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 8.8600e-04 - mean_absolute_error: 0.0210 - val_loss: 7.0352e-04 - val_mean_absolute_error: 0.0189\n",
+      "Epoch 38/50\n",
+      "18750/18750 [==============================] - 6s 346us/step - loss: 8.5247e-04 - mean_absolute_error: 0.0206 - val_loss: 8.1930e-04 - val_mean_absolute_error: 0.0202\n",
+      "Epoch 39/50\n",
+      "18750/18750 [==============================] - 6s 345us/step - loss: 8.4065e-04 - mean_absolute_error: 0.0205 - val_loss: 0.0012 - val_mean_absolute_error: 0.0235\n",
+      "Epoch 40/50\n",
+      "18750/18750 [==============================] - 6s 345us/step - loss: 8.3871e-04 - mean_absolute_error: 0.0204 - val_loss: 0.0011 - val_mean_absolute_error: 0.0244\n",
+      "Epoch 41/50\n",
+      "18750/18750 [==============================] - 6s 343us/step - loss: 8.2854e-04 - mean_absolute_error: 0.0203 - val_loss: 0.0013 - val_mean_absolute_error: 0.0254\n",
+      "Epoch 42/50\n",
+      "18750/18750 [==============================] - 6s 343us/step - loss: 8.2941e-04 - mean_absolute_error: 0.0203 - val_loss: 8.3724e-04 - val_mean_absolute_error: 0.0222\n",
+      "Epoch 43/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 8.0728e-04 - mean_absolute_error: 0.0201 - val_loss: 0.0011 - val_mean_absolute_error: 0.0241\n",
+      "Epoch 44/50\n",
+      "18750/18750 [==============================] - 7s 348us/step - loss: 8.1050e-04 - mean_absolute_error: 0.0200 - val_loss: 5.8376e-04 - val_mean_absolute_error: 0.0179\n",
+      "Epoch 45/50\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 8.0727e-04 - mean_absolute_error: 0.0200 - val_loss: 7.3209e-04 - val_mean_absolute_error: 0.0189\n",
+      "Epoch 46/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 7.8446e-04 - mean_absolute_error: 0.0197 - val_loss: 7.8105e-04 - val_mean_absolute_error: 0.0206\n",
+      "Epoch 47/50\n",
+      "18750/18750 [==============================] - 7s 347us/step - loss: 7.7875e-04 - mean_absolute_error: 0.0196 - val_loss: 9.5163e-04 - val_mean_absolute_error: 0.0221\n",
+      "Epoch 48/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 7.9367e-04 - mean_absolute_error: 0.0198 - val_loss: 5.8918e-04 - val_mean_absolute_error: 0.0176\n",
+      "Epoch 49/50\n",
+      "18750/18750 [==============================] - 7s 349us/step - loss: 7.8587e-04 - mean_absolute_error: 0.0198 - val_loss: 8.9809e-04 - val_mean_absolute_error: 0.0213\n",
+      "Epoch 50/50\n",
+      "18750/18750 [==============================] - 6s 345us/step - loss: 7.6797e-04 - mean_absolute_error: 0.0195 - val_loss: 6.6490e-04 - val_mean_absolute_error: 0.0171\n",
+      "Epoch 1/50\n",
+      "18750/18750 [==============================] - 7s 345us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 2/50\n",
+      "18750/18750 [==============================] - 6s 345us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 3/50\n",
+      "18750/18750 [==============================] - 6s 343us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 4/50\n",
+      "18750/18750 [==============================] - 6s 346us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 5/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 0.0492 - mean_absolute_error: 0.1806 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 6/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 0.0491 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 7/50\n",
+      "18750/18750 [==============================] - 6s 343us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 8/50\n",
+      "18750/18750 [==============================] - 7s 348us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 9/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 10/50\n",
+      "18750/18750 [==============================] - 6s 346us/step - loss: 0.0490 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 11/50\n",
+      "18750/18750 [==============================] - 6s 345us/step - loss: 0.0490 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 12/50\n",
+      "18750/18750 [==============================] - 6s 345us/step - loss: 0.0491 - mean_absolute_error: 0.1806 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 13/50\n",
+      "18750/18750 [==============================] - 6s 343us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 14/50\n",
+      "18750/18750 [==============================] - 6s 346us/step - loss: 0.0488 - mean_absolute_error: 0.1800 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 15/50\n",
+      "18750/18750 [==============================] - 6s 345us/step - loss: 0.0491 - mean_absolute_error: 0.1807 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 16/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 0.0489 - mean_absolute_error: 0.1801 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 17/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 0.0490 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 18/50\n",
+      "18750/18750 [==============================] - 6s 341us/step - loss: 0.0491 - mean_absolute_error: 0.1806 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 19/50\n",
+      "18750/18750 [==============================] - 6s 342us/step - loss: 0.0490 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 20/50\n",
+      "18750/18750 [==============================] - 6s 338us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 21/50\n",
+      "18750/18750 [==============================] - 6s 340us/step - loss: 0.0491 - mean_absolute_error: 0.1806 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 22/50\n",
+      "18750/18750 [==============================] - 6s 341us/step - loss: 0.0492 - mean_absolute_error: 0.1806 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 23/50\n",
+      "18750/18750 [==============================] - 6s 345us/step - loss: 0.0489 - mean_absolute_error: 0.1801 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 24/50\n",
+      "18750/18750 [==============================] - 7s 365us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 25/50\n",
+      "18750/18750 [==============================] - 7s 348us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 26/50\n",
+      "18750/18750 [==============================] - 6s 346us/step - loss: 0.0490 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 27/50\n",
+      "18750/18750 [==============================] - 7s 353us/step - loss: 0.0490 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 28/50\n",
+      "18750/18750 [==============================] - 7s 350us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 29/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 30/50\n",
+      "18750/18750 [==============================] - 6s 341us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 31/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 0.0489 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 32/50\n",
+      "18750/18750 [==============================] - 6s 342us/step - loss: 0.0490 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 33/50\n",
+      "18750/18750 [==============================] - 6s 343us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 34/50\n",
+      "18750/18750 [==============================] - 6s 342us/step - loss: 0.0490 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 35/50\n",
+      "18750/18750 [==============================] - 6s 345us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 36/50\n",
+      "18750/18750 [==============================] - 6s 345us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 37/50\n",
+      "18750/18750 [==============================] - 6s 343us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 38/50\n",
+      "18750/18750 [==============================] - 6s 343us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 39/50\n",
+      "18750/18750 [==============================] - 6s 342us/step - loss: 0.0488 - mean_absolute_error: 0.1801 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 40/50\n",
+      "18750/18750 [==============================] - 6s 342us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 41/50\n",
+      "18750/18750 [==============================] - 6s 341us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 42/50\n",
+      "18750/18750 [==============================] - 6s 343us/step - loss: 0.0489 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 43/50\n",
+      "18750/18750 [==============================] - 6s 345us/step - loss: 0.0491 - mean_absolute_error: 0.1806 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 44/50\n",
+      "18750/18750 [==============================] - 6s 342us/step - loss: 0.0491 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 45/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 0.0492 - mean_absolute_error: 0.1807 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 46/50\n",
+      "18750/18750 [==============================] - 6s 343us/step - loss: 0.0492 - mean_absolute_error: 0.1807 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 47/50\n",
+      "18750/18750 [==============================] - 7s 350us/step - loss: 0.0491 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 48/50\n",
+      "18750/18750 [==============================] - 6s 343us/step - loss: 0.0491 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 49/50\n",
+      "18750/18750 [==============================] - 6s 343us/step - loss: 0.0490 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 50/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 0.0491 - mean_absolute_error: 0.1806 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n"
+     ]
+    }
+   ],
+   "source": [
+    "histories=np.zeros_like(models)\n",
+    "for i in range(len(models)):\n",
+    "    histories[i]=models[i].fit(X_train,y_train,\n",
+    "                               validation_data=(X_val,y_val),\n",
+    "                               batch_size=32,\n",
+    "                               epochs=50)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 5,
+   "id": "05f6e494-47dd-4e9d-b152-750f5a5b6467",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "image/png": "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",
+      "text/plain": [
+       "<Figure size 640x480 with 1 Axes>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    }
+   ],
+   "source": [
+    "colors=['orange','green','red','blue']\n",
+    "legend=['stddev=1.0','stddev=0.1','stddev=0.01','stddev=0.001']\n",
+    "for i in range(len(models)):\n",
+    "    plt.plot(histories[i].history['loss'],color=colors[i])\n",
+    "    plt.yscale('log')\n",
+    "plt.title('Model loss on the training set \\n for different values of the standard deviation \\n of the Random Normal initializer')\n",
+    "plt.xlabel('epoch')\n",
+    "plt.ylabel('mean squared error')\n",
+    "plt.legend(legend,loc='upper right')\n",
+    "plt.show()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 6,
+   "id": "f60e6de6-5947-4b60-849f-39a7583eecf8",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "image/png": "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",
+      "text/plain": [
+       "<Figure size 640x480 with 1 Axes>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    }
+   ],
+   "source": [
+    "colors=['orange','green','red','blue']\n",
+    "legend=['stddev=1.0','stddev=0.1','stddev=0.01','stddev=0.001']\n",
+    "for i in range(len(models)):\n",
+    "  plt.plot(histories[i].history['val_loss'],color=colors[i])\n",
+    "plt.title('Model loss on the validation set \\n for different values of the standard deviation  \\n of the Random Normal initializer')\n",
+    "plt.xlabel('epoch')\n",
+    "plt.ylabel('mean squared error')\n",
+    "plt.legend(legend,loc='upper right')\n",
+    "plt.show()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 26,
+   "id": "1bbc4a9c-5d06-493d-afdd-83d344551eaf",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "#Choose the most appropriate range for the Random Uniform initializer\n",
+    "rand_unif=[keras.initializers.RandomUniform(minval=-1,maxval=1),keras.initializers.RandomUniform(minval=-0.1,maxval=0.1),keras.initializers.RandomUniform(minval=-0.01,maxval=0.01),keras.initializers.RandomUniform(minval=-0.001,maxval=0.001)]\n",
+    "models=np.zeros_like(rand_unif)\n",
+    "for i in range(len(models)):\n",
+    "  models[i]=keras.models.Sequential()\n",
+    "  models[i].add(keras.layers.Dense(units=32, activation='relu', input_dim=X_train.shape[1], kernel_initializer=rand_unif[i]))\n",
+    "  models[i].add(keras.layers.Dense(units=32, activation='sigmoid', kernel_initializer=rand_unif[i]))\n",
+    "  models[i].add(keras.layers.Dense(units=64, activation='sigmoid', kernel_initializer=rand_unif[i]))\n",
+    "  models[i].add(keras.layers.Dense(units=1, activation='relu', kernel_initializer=rand_unif[i]))\n",
+    "  models[i].compile(optimizer='adam',\n",
+    "                    loss='mean_squared_error',\n",
+    "                    metrics=['mean_absolute_error'])"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 8,
+   "id": "fd398685-3b29-4270-822b-366b38024408",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Epoch 1/50\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0434 - mean_absolute_error: 0.1312 - val_loss: 0.0149 - val_mean_absolute_error: 0.0997\n",
+      "Epoch 2/50\n",
+      "18750/18750 [==============================] - 7s 351us/step - loss: 0.0145 - mean_absolute_error: 0.0974 - val_loss: 0.0139 - val_mean_absolute_error: 0.0945\n",
+      "Epoch 3/50\n",
+      "18750/18750 [==============================] - 7s 349us/step - loss: 0.0133 - mean_absolute_error: 0.0914 - val_loss: 0.0115 - val_mean_absolute_error: 0.0835\n",
+      "Epoch 4/50\n",
+      "18750/18750 [==============================] - 7s 349us/step - loss: 0.0112 - mean_absolute_error: 0.0823 - val_loss: 0.0106 - val_mean_absolute_error: 0.0795\n",
+      "Epoch 5/50\n",
+      "18750/18750 [==============================] - 6s 346us/step - loss: 0.0104 - mean_absolute_error: 0.0787 - val_loss: 0.0113 - val_mean_absolute_error: 0.0832\n",
+      "Epoch 6/50\n",
+      "18750/18750 [==============================] - 7s 351us/step - loss: 0.0093 - mean_absolute_error: 0.0734 - val_loss: 0.0087 - val_mean_absolute_error: 0.0706\n",
+      "Epoch 7/50\n",
+      "18750/18750 [==============================] - 7s 349us/step - loss: 0.0085 - mean_absolute_error: 0.0692 - val_loss: 0.0078 - val_mean_absolute_error: 0.0643\n",
+      "Epoch 8/50\n",
+      "18750/18750 [==============================] - 7s 350us/step - loss: 0.0075 - mean_absolute_error: 0.0642 - val_loss: 0.0072 - val_mean_absolute_error: 0.0641\n",
+      "Epoch 9/50\n",
+      "18750/18750 [==============================] - 7s 348us/step - loss: 0.0068 - mean_absolute_error: 0.0604 - val_loss: 0.0071 - val_mean_absolute_error: 0.0637\n",
+      "Epoch 10/50\n",
+      "18750/18750 [==============================] - 7s 352us/step - loss: 0.0063 - mean_absolute_error: 0.0579 - val_loss: 0.0060 - val_mean_absolute_error: 0.0572\n",
+      "Epoch 11/50\n",
+      "18750/18750 [==============================] - 6s 346us/step - loss: 0.0059 - mean_absolute_error: 0.0559 - val_loss: 0.0062 - val_mean_absolute_error: 0.0591\n",
+      "Epoch 12/50\n",
+      "18750/18750 [==============================] - 6s 345us/step - loss: 0.0057 - mean_absolute_error: 0.0544 - val_loss: 0.0056 - val_mean_absolute_error: 0.0533\n",
+      "Epoch 13/50\n",
+      "18750/18750 [==============================] - 7s 348us/step - loss: 0.0054 - mean_absolute_error: 0.0530 - val_loss: 0.0055 - val_mean_absolute_error: 0.0544\n",
+      "Epoch 14/50\n",
+      "18750/18750 [==============================] - 7s 347us/step - loss: 0.0051 - mean_absolute_error: 0.0515 - val_loss: 0.0048 - val_mean_absolute_error: 0.0493\n",
+      "Epoch 15/50\n",
+      "18750/18750 [==============================] - 7s 349us/step - loss: 0.0050 - mean_absolute_error: 0.0506 - val_loss: 0.0045 - val_mean_absolute_error: 0.0468\n",
+      "Epoch 16/50\n",
+      "18750/18750 [==============================] - 7s 350us/step - loss: 0.0048 - mean_absolute_error: 0.0496 - val_loss: 0.0050 - val_mean_absolute_error: 0.0511\n",
+      "Epoch 17/50\n",
+      "18750/18750 [==============================] - 7s 347us/step - loss: 0.0047 - mean_absolute_error: 0.0490 - val_loss: 0.0041 - val_mean_absolute_error: 0.0449\n",
+      "Epoch 18/50\n",
+      "18750/18750 [==============================] - 7s 347us/step - loss: 0.0045 - mean_absolute_error: 0.0480 - val_loss: 0.0049 - val_mean_absolute_error: 0.0529\n",
+      "Epoch 19/50\n",
+      "18750/18750 [==============================] - 7s 350us/step - loss: 0.0042 - mean_absolute_error: 0.0468 - val_loss: 0.0039 - val_mean_absolute_error: 0.0458\n",
+      "Epoch 20/50\n",
+      "18750/18750 [==============================] - 7s 352us/step - loss: 0.0040 - mean_absolute_error: 0.0451 - val_loss: 0.0035 - val_mean_absolute_error: 0.0426\n",
+      "Epoch 21/50\n",
+      "18750/18750 [==============================] - 7s 350us/step - loss: 0.0037 - mean_absolute_error: 0.0439 - val_loss: 0.0036 - val_mean_absolute_error: 0.0437\n",
+      "Epoch 22/50\n",
+      "18750/18750 [==============================] - 7s 350us/step - loss: 0.0035 - mean_absolute_error: 0.0422 - val_loss: 0.0034 - val_mean_absolute_error: 0.0438\n",
+      "Epoch 23/50\n",
+      "18750/18750 [==============================] - 7s 349us/step - loss: 0.0032 - mean_absolute_error: 0.0402 - val_loss: 0.0035 - val_mean_absolute_error: 0.0438\n",
+      "Epoch 24/50\n",
+      "18750/18750 [==============================] - 7s 348us/step - loss: 0.0030 - mean_absolute_error: 0.0390 - val_loss: 0.0029 - val_mean_absolute_error: 0.0398\n",
+      "Epoch 25/50\n",
+      "18750/18750 [==============================] - 7s 349us/step - loss: 0.0029 - mean_absolute_error: 0.0384 - val_loss: 0.0029 - val_mean_absolute_error: 0.0374\n",
+      "Epoch 26/50\n",
+      "18750/18750 [==============================] - 7s 350us/step - loss: 0.0028 - mean_absolute_error: 0.0378 - val_loss: 0.0023 - val_mean_absolute_error: 0.0337\n",
+      "Epoch 27/50\n",
+      "18750/18750 [==============================] - 7s 353us/step - loss: 0.0027 - mean_absolute_error: 0.0374 - val_loss: 0.0027 - val_mean_absolute_error: 0.0382\n",
+      "Epoch 28/50\n",
+      "18750/18750 [==============================] - 7s 350us/step - loss: 0.0026 - mean_absolute_error: 0.0366 - val_loss: 0.0022 - val_mean_absolute_error: 0.0327\n",
+      "Epoch 29/50\n",
+      "18750/18750 [==============================] - 7s 348us/step - loss: 0.0025 - mean_absolute_error: 0.0359 - val_loss: 0.0021 - val_mean_absolute_error: 0.0323\n",
+      "Epoch 30/50\n",
+      "18750/18750 [==============================] - 7s 351us/step - loss: 0.0025 - mean_absolute_error: 0.0356 - val_loss: 0.0027 - val_mean_absolute_error: 0.0367\n",
+      "Epoch 31/50\n",
+      "18750/18750 [==============================] - 7s 352us/step - loss: 0.0024 - mean_absolute_error: 0.0351 - val_loss: 0.0023 - val_mean_absolute_error: 0.0333\n",
+      "Epoch 32/50\n",
+      "18750/18750 [==============================] - 7s 347us/step - loss: 0.0024 - mean_absolute_error: 0.0347 - val_loss: 0.0020 - val_mean_absolute_error: 0.0301\n",
+      "Epoch 33/50\n",
+      "18750/18750 [==============================] - 7s 349us/step - loss: 0.0022 - mean_absolute_error: 0.0338 - val_loss: 0.0019 - val_mean_absolute_error: 0.0300\n",
+      "Epoch 34/50\n",
+      "18750/18750 [==============================] - 7s 351us/step - loss: 0.0022 - mean_absolute_error: 0.0332 - val_loss: 0.0025 - val_mean_absolute_error: 0.0371\n",
+      "Epoch 35/50\n",
+      "18750/18750 [==============================] - 7s 349us/step - loss: 0.0021 - mean_absolute_error: 0.0328 - val_loss: 0.0020 - val_mean_absolute_error: 0.0318\n",
+      "Epoch 36/50\n",
+      "18750/18750 [==============================] - 7s 350us/step - loss: 0.0021 - mean_absolute_error: 0.0324 - val_loss: 0.0023 - val_mean_absolute_error: 0.0337\n",
+      "Epoch 37/50\n",
+      "18750/18750 [==============================] - 7s 351us/step - loss: 0.0021 - mean_absolute_error: 0.0324 - val_loss: 0.0017 - val_mean_absolute_error: 0.0283\n",
+      "Epoch 38/50\n",
+      "18750/18750 [==============================] - 7s 352us/step - loss: 0.0020 - mean_absolute_error: 0.0321 - val_loss: 0.0016 - val_mean_absolute_error: 0.0269\n",
+      "Epoch 39/50\n",
+      "18750/18750 [==============================] - 7s 351us/step - loss: 0.0020 - mean_absolute_error: 0.0317 - val_loss: 0.0022 - val_mean_absolute_error: 0.0339\n",
+      "Epoch 40/50\n",
+      "18750/18750 [==============================] - 7s 350us/step - loss: 0.0020 - mean_absolute_error: 0.0318 - val_loss: 0.0018 - val_mean_absolute_error: 0.0306\n",
+      "Epoch 41/50\n",
+      "18750/18750 [==============================] - 7s 351us/step - loss: 0.0019 - mean_absolute_error: 0.0312 - val_loss: 0.0020 - val_mean_absolute_error: 0.0321\n",
+      "Epoch 42/50\n",
+      "18750/18750 [==============================] - 6s 345us/step - loss: 0.0020 - mean_absolute_error: 0.0313 - val_loss: 0.0017 - val_mean_absolute_error: 0.0295\n",
+      "Epoch 43/50\n",
+      "18750/18750 [==============================] - 7s 348us/step - loss: 0.0019 - mean_absolute_error: 0.0311 - val_loss: 0.0020 - val_mean_absolute_error: 0.0318\n",
+      "Epoch 44/50\n",
+      "18750/18750 [==============================] - 7s 350us/step - loss: 0.0019 - mean_absolute_error: 0.0309 - val_loss: 0.0015 - val_mean_absolute_error: 0.0266\n",
+      "Epoch 45/50\n",
+      "18750/18750 [==============================] - 7s 351us/step - loss: 0.0019 - mean_absolute_error: 0.0308 - val_loss: 0.0017 - val_mean_absolute_error: 0.0277\n",
+      "Epoch 46/50\n",
+      "18750/18750 [==============================] - 7s 350us/step - loss: 0.0019 - mean_absolute_error: 0.0307 - val_loss: 0.0021 - val_mean_absolute_error: 0.0339\n",
+      "Epoch 47/50\n",
+      "18750/18750 [==============================] - 7s 350us/step - loss: 0.0019 - mean_absolute_error: 0.0306 - val_loss: 0.0019 - val_mean_absolute_error: 0.0312\n",
+      "Epoch 48/50\n",
+      "18750/18750 [==============================] - 7s 353us/step - loss: 0.0018 - mean_absolute_error: 0.0302 - val_loss: 0.0016 - val_mean_absolute_error: 0.0289\n",
+      "Epoch 49/50\n",
+      "18750/18750 [==============================] - 7s 355us/step - loss: 0.0018 - mean_absolute_error: 0.0299 - val_loss: 0.0018 - val_mean_absolute_error: 0.0299\n",
+      "Epoch 50/50\n",
+      "18750/18750 [==============================] - 7s 349us/step - loss: 0.0018 - mean_absolute_error: 0.0297 - val_loss: 0.0015 - val_mean_absolute_error: 0.0261\n",
+      "Epoch 1/50\n",
+      "18750/18750 [==============================] - 7s 349us/step - loss: 0.0491 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 2/50\n",
+      "18750/18750 [==============================] - 7s 350us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 3/50\n",
+      "18750/18750 [==============================] - 7s 347us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 4/50\n",
+      "18750/18750 [==============================] - 6s 346us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 5/50\n",
+      "18750/18750 [==============================] - 7s 349us/step - loss: 0.0491 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 6/50\n",
+      "18750/18750 [==============================] - 6s 346us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 7/50\n",
+      "18750/18750 [==============================] - 7s 349us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 8/50\n",
+      "18750/18750 [==============================] - 7s 348us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 9/50\n",
+      "18750/18750 [==============================] - 7s 348us/step - loss: 0.0488 - mean_absolute_error: 0.1801 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 10/50\n",
+      "18750/18750 [==============================] - 7s 350us/step - loss: 0.0491 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 11/50\n",
+      "18750/18750 [==============================] - 6s 346us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 12/50\n",
+      "18750/18750 [==============================] - 7s 348us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 13/50\n",
+      "18750/18750 [==============================] - 7s 352us/step - loss: 0.0492 - mean_absolute_error: 0.1808 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 14/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 15/50\n",
+      "18750/18750 [==============================] - 6s 345us/step - loss: 0.0489 - mean_absolute_error: 0.1801 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 16/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 0.0489 - mean_absolute_error: 0.1800 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 17/50\n",
+      "18750/18750 [==============================] - 7s 348us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 18/50\n",
+      "18750/18750 [==============================] - 7s 350us/step - loss: 0.0491 - mean_absolute_error: 0.1806 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 19/50\n",
+      "18750/18750 [==============================] - 7s 350us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 20/50\n",
+      "18750/18750 [==============================] - 7s 347us/step - loss: 0.0489 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 21/50\n",
+      "18750/18750 [==============================] - 6s 346us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 22/50\n",
+      "18750/18750 [==============================] - 7s 352us/step - loss: 0.0490 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 23/50\n",
+      "18750/18750 [==============================] - 7s 349us/step - loss: 0.0489 - mean_absolute_error: 0.1801 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 24/50\n",
+      "18750/18750 [==============================] - 7s 348us/step - loss: 0.0491 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 25/50\n",
+      "18750/18750 [==============================] - 6s 346us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 26/50\n",
+      "18750/18750 [==============================] - 7s 347us/step - loss: 0.0491 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 27/50\n",
+      "18750/18750 [==============================] - 7s 349us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 28/50\n",
+      "18750/18750 [==============================] - 6s 346us/step - loss: 0.0491 - mean_absolute_error: 0.1806 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 29/50\n",
+      "18750/18750 [==============================] - 7s 347us/step - loss: 0.0491 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 30/50\n",
+      "18750/18750 [==============================] - 6s 346us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 31/50\n",
+      "18750/18750 [==============================] - 6s 345us/step - loss: 0.0491 - mean_absolute_error: 0.1806 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 32/50\n",
+      "18750/18750 [==============================] - 7s 347us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 33/50\n",
+      "18750/18750 [==============================] - 7s 348us/step - loss: 0.0491 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 34/50\n",
+      "18750/18750 [==============================] - 7s 348us/step - loss: 0.0490 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 35/50\n",
+      "18750/18750 [==============================] - 6s 346us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 36/50\n",
+      "18750/18750 [==============================] - 7s 348us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 37/50\n",
+      "18750/18750 [==============================] - 7s 347us/step - loss: 0.0489 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 38/50\n",
+      "18750/18750 [==============================] - 7s 348us/step - loss: 0.0490 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 39/50\n",
+      "18750/18750 [==============================] - 7s 349us/step - loss: 0.0491 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 40/50\n",
+      "18750/18750 [==============================] - 6s 346us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 41/50\n",
+      "18750/18750 [==============================] - 6s 346us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 42/50\n",
+      "18750/18750 [==============================] - 7s 347us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 43/50\n",
+      "18750/18750 [==============================] - 6s 346us/step - loss: 0.0489 - mean_absolute_error: 0.1801 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 44/50\n",
+      "18750/18750 [==============================] - 7s 351us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 45/50\n",
+      "18750/18750 [==============================] - 7s 347us/step - loss: 0.0491 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 46/50\n",
+      "18750/18750 [==============================] - 7s 349us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 47/50\n",
+      "18750/18750 [==============================] - 7s 348us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 48/50\n",
+      "18750/18750 [==============================] - 7s 348us/step - loss: 0.0491 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 49/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 50/50\n",
+      "18750/18750 [==============================] - 7s 348us/step - loss: 0.0492 - mean_absolute_error: 0.1807 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 1/50\n",
+      "18750/18750 [==============================] - 7s 351us/step - loss: 0.0150 - mean_absolute_error: 0.1000 - val_loss: 0.0116 - val_mean_absolute_error: 0.0834\n",
+      "Epoch 2/50\n",
+      "18750/18750 [==============================] - 7s 347us/step - loss: 0.0109 - mean_absolute_error: 0.0814 - val_loss: 0.0084 - val_mean_absolute_error: 0.0684\n",
+      "Epoch 3/50\n",
+      "18750/18750 [==============================] - 7s 348us/step - loss: 0.0078 - mean_absolute_error: 0.0658 - val_loss: 0.0060 - val_mean_absolute_error: 0.0563\n",
+      "Epoch 4/50\n",
+      "18750/18750 [==============================] - 7s 348us/step - loss: 0.0062 - mean_absolute_error: 0.0572 - val_loss: 0.0056 - val_mean_absolute_error: 0.0530\n",
+      "Epoch 5/50\n",
+      "18750/18750 [==============================] - 6s 345us/step - loss: 0.0054 - mean_absolute_error: 0.0531 - val_loss: 0.0044 - val_mean_absolute_error: 0.0480\n",
+      "Epoch 6/50\n",
+      "18750/18750 [==============================] - 7s 348us/step - loss: 0.0046 - mean_absolute_error: 0.0486 - val_loss: 0.0044 - val_mean_absolute_error: 0.0471\n",
+      "Epoch 7/50\n",
+      "18750/18750 [==============================] - 6s 345us/step - loss: 0.0038 - mean_absolute_error: 0.0443 - val_loss: 0.0037 - val_mean_absolute_error: 0.0437\n",
+      "Epoch 8/50\n",
+      "18750/18750 [==============================] - 6s 347us/step - loss: 0.0034 - mean_absolute_error: 0.0418 - val_loss: 0.0041 - val_mean_absolute_error: 0.0494\n",
+      "Epoch 9/50\n",
+      "18750/18750 [==============================] - 6s 347us/step - loss: 0.0031 - mean_absolute_error: 0.0397 - val_loss: 0.0035 - val_mean_absolute_error: 0.0442\n",
+      "Epoch 10/50\n",
+      "18750/18750 [==============================] - 7s 348us/step - loss: 0.0029 - mean_absolute_error: 0.0380 - val_loss: 0.0023 - val_mean_absolute_error: 0.0329\n",
+      "Epoch 11/50\n",
+      "18750/18750 [==============================] - 7s 349us/step - loss: 0.0027 - mean_absolute_error: 0.0366 - val_loss: 0.0024 - val_mean_absolute_error: 0.0336\n",
+      "Epoch 12/50\n",
+      "18750/18750 [==============================] - 7s 349us/step - loss: 0.0025 - mean_absolute_error: 0.0354 - val_loss: 0.0020 - val_mean_absolute_error: 0.0306\n",
+      "Epoch 13/50\n",
+      "18750/18750 [==============================] - 7s 349us/step - loss: 0.0024 - mean_absolute_error: 0.0343 - val_loss: 0.0020 - val_mean_absolute_error: 0.0306\n",
+      "Epoch 14/50\n",
+      "18750/18750 [==============================] - 6s 346us/step - loss: 0.0022 - mean_absolute_error: 0.0332 - val_loss: 0.0022 - val_mean_absolute_error: 0.0327\n",
+      "Epoch 15/50\n",
+      "18750/18750 [==============================] - 7s 353us/step - loss: 0.0021 - mean_absolute_error: 0.0324 - val_loss: 0.0023 - val_mean_absolute_error: 0.0333\n",
+      "Epoch 16/50\n",
+      "18750/18750 [==============================] - 7s 350us/step - loss: 0.0021 - mean_absolute_error: 0.0317 - val_loss: 0.0025 - val_mean_absolute_error: 0.0344\n",
+      "Epoch 17/50\n",
+      "18750/18750 [==============================] - 6s 346us/step - loss: 0.0020 - mean_absolute_error: 0.0310 - val_loss: 0.0019 - val_mean_absolute_error: 0.0309\n",
+      "Epoch 18/50\n",
+      "18750/18750 [==============================] - 7s 347us/step - loss: 0.0019 - mean_absolute_error: 0.0302 - val_loss: 0.0021 - val_mean_absolute_error: 0.0352\n",
+      "Epoch 19/50\n",
+      "18750/18750 [==============================] - 7s 352us/step - loss: 0.0018 - mean_absolute_error: 0.0297 - val_loss: 0.0016 - val_mean_absolute_error: 0.0269\n",
+      "Epoch 20/50\n",
+      "18750/18750 [==============================] - 7s 351us/step - loss: 0.0017 - mean_absolute_error: 0.0292 - val_loss: 0.0015 - val_mean_absolute_error: 0.0259\n",
+      "Epoch 21/50\n",
+      "18750/18750 [==============================] - 7s 351us/step - loss: 0.0017 - mean_absolute_error: 0.0288 - val_loss: 0.0015 - val_mean_absolute_error: 0.0279\n",
+      "Epoch 22/50\n",
+      "18750/18750 [==============================] - 7s 352us/step - loss: 0.0016 - mean_absolute_error: 0.0281 - val_loss: 0.0014 - val_mean_absolute_error: 0.0259\n",
+      "Epoch 23/50\n",
+      "18750/18750 [==============================] - 7s 347us/step - loss: 0.0016 - mean_absolute_error: 0.0276 - val_loss: 0.0015 - val_mean_absolute_error: 0.0279\n",
+      "Epoch 24/50\n",
+      "18750/18750 [==============================] - 7s 349us/step - loss: 0.0015 - mean_absolute_error: 0.0273 - val_loss: 0.0014 - val_mean_absolute_error: 0.0263\n",
+      "Epoch 25/50\n",
+      "18750/18750 [==============================] - 6s 347us/step - loss: 0.0015 - mean_absolute_error: 0.0271 - val_loss: 0.0011 - val_mean_absolute_error: 0.0228\n",
+      "Epoch 26/50\n",
+      "18750/18750 [==============================] - 7s 352us/step - loss: 0.0014 - mean_absolute_error: 0.0266 - val_loss: 0.0013 - val_mean_absolute_error: 0.0245\n",
+      "Epoch 27/50\n",
+      "18750/18750 [==============================] - 7s 348us/step - loss: 0.0014 - mean_absolute_error: 0.0263 - val_loss: 0.0010 - val_mean_absolute_error: 0.0223\n",
+      "Epoch 28/50\n",
+      "18750/18750 [==============================] - 6s 346us/step - loss: 0.0014 - mean_absolute_error: 0.0260 - val_loss: 0.0013 - val_mean_absolute_error: 0.0259\n",
+      "Epoch 29/50\n",
+      "18750/18750 [==============================] - 7s 351us/step - loss: 0.0013 - mean_absolute_error: 0.0258 - val_loss: 0.0013 - val_mean_absolute_error: 0.0257\n",
+      "Epoch 30/50\n",
+      "18750/18750 [==============================] - 6s 346us/step - loss: 0.0013 - mean_absolute_error: 0.0253 - val_loss: 0.0013 - val_mean_absolute_error: 0.0253\n",
+      "Epoch 31/50\n",
+      "18750/18750 [==============================] - 7s 347us/step - loss: 0.0013 - mean_absolute_error: 0.0252 - val_loss: 9.8709e-04 - val_mean_absolute_error: 0.0218\n",
+      "Epoch 32/50\n",
+      "18750/18750 [==============================] - 7s 348us/step - loss: 0.0013 - mean_absolute_error: 0.0251 - val_loss: 0.0012 - val_mean_absolute_error: 0.0232\n",
+      "Epoch 33/50\n",
+      "18750/18750 [==============================] - 7s 351us/step - loss: 0.0012 - mean_absolute_error: 0.0246 - val_loss: 0.0011 - val_mean_absolute_error: 0.0228\n",
+      "Epoch 34/50\n",
+      "18750/18750 [==============================] - 7s 349us/step - loss: 0.0012 - mean_absolute_error: 0.0244 - val_loss: 0.0027 - val_mean_absolute_error: 0.0416\n",
+      "Epoch 35/50\n",
+      "18750/18750 [==============================] - 7s 348us/step - loss: 0.0012 - mean_absolute_error: 0.0242 - val_loss: 0.0021 - val_mean_absolute_error: 0.0340\n",
+      "Epoch 36/50\n",
+      "18750/18750 [==============================] - 7s 347us/step - loss: 0.0011 - mean_absolute_error: 0.0239 - val_loss: 9.6532e-04 - val_mean_absolute_error: 0.0216\n",
+      "Epoch 37/50\n",
+      "18750/18750 [==============================] - 7s 350us/step - loss: 0.0011 - mean_absolute_error: 0.0235 - val_loss: 0.0011 - val_mean_absolute_error: 0.0226\n",
+      "Epoch 38/50\n",
+      "18750/18750 [==============================] - 7s 352us/step - loss: 0.0011 - mean_absolute_error: 0.0233 - val_loss: 0.0014 - val_mean_absolute_error: 0.0288\n",
+      "Epoch 39/50\n",
+      "18750/18750 [==============================] - 7s 349us/step - loss: 0.0011 - mean_absolute_error: 0.0232 - val_loss: 0.0012 - val_mean_absolute_error: 0.0265\n",
+      "Epoch 40/50\n",
+      "18750/18750 [==============================] - 7s 348us/step - loss: 0.0011 - mean_absolute_error: 0.0231 - val_loss: 9.5703e-04 - val_mean_absolute_error: 0.0211\n",
+      "Epoch 41/50\n",
+      "18750/18750 [==============================] - 7s 350us/step - loss: 0.0010 - mean_absolute_error: 0.0226 - val_loss: 0.0012 - val_mean_absolute_error: 0.0249\n",
+      "Epoch 42/50\n",
+      "18750/18750 [==============================] - 7s 349us/step - loss: 0.0010 - mean_absolute_error: 0.0227 - val_loss: 0.0011 - val_mean_absolute_error: 0.0229\n",
+      "Epoch 43/50\n",
+      "18750/18750 [==============================] - 7s 353us/step - loss: 0.0010 - mean_absolute_error: 0.0224 - val_loss: 0.0011 - val_mean_absolute_error: 0.0229\n",
+      "Epoch 44/50\n",
+      "18750/18750 [==============================] - 7s 353us/step - loss: 9.9677e-04 - mean_absolute_error: 0.0223 - val_loss: 9.2860e-04 - val_mean_absolute_error: 0.0218\n",
+      "Epoch 45/50\n",
+      "18750/18750 [==============================] - 7s 350us/step - loss: 9.8307e-04 - mean_absolute_error: 0.0221 - val_loss: 0.0014 - val_mean_absolute_error: 0.0275\n",
+      "Epoch 46/50\n",
+      "18750/18750 [==============================] - 7s 349us/step - loss: 9.7966e-04 - mean_absolute_error: 0.0221 - val_loss: 9.5059e-04 - val_mean_absolute_error: 0.0212\n",
+      "Epoch 47/50\n",
+      "18750/18750 [==============================] - 7s 349us/step - loss: 9.4433e-04 - mean_absolute_error: 0.0216 - val_loss: 8.4521e-04 - val_mean_absolute_error: 0.0213\n",
+      "Epoch 48/50\n",
+      "18750/18750 [==============================] - 7s 350us/step - loss: 9.5105e-04 - mean_absolute_error: 0.0218 - val_loss: 9.1135e-04 - val_mean_absolute_error: 0.0215\n",
+      "Epoch 49/50\n",
+      "18750/18750 [==============================] - 6s 346us/step - loss: 9.2081e-04 - mean_absolute_error: 0.0214 - val_loss: 9.3821e-04 - val_mean_absolute_error: 0.0227\n",
+      "Epoch 50/50\n",
+      "18750/18750 [==============================] - 7s 349us/step - loss: 9.3326e-04 - mean_absolute_error: 0.0216 - val_loss: 9.6320e-04 - val_mean_absolute_error: 0.0210\n",
+      "Epoch 1/50\n",
+      "18750/18750 [==============================] - 7s 347us/step - loss: 0.0489 - mean_absolute_error: 0.1801 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 2/50\n",
+      "18750/18750 [==============================] - 6s 341us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 3/50\n",
+      "18750/18750 [==============================] - 6s 347us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 4/50\n",
+      "18750/18750 [==============================] - 6s 341us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 5/50\n",
+      "18750/18750 [==============================] - 6s 341us/step - loss: 0.0489 - mean_absolute_error: 0.1801 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 6/50\n",
+      "18750/18750 [==============================] - 6s 340us/step - loss: 0.0489 - mean_absolute_error: 0.1801 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 7/50\n",
+      "18750/18750 [==============================] - 6s 342us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 8/50\n",
+      "18750/18750 [==============================] - 6s 340us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 9/50\n",
+      "18750/18750 [==============================] - 6s 339us/step - loss: 0.0492 - mean_absolute_error: 0.1807 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 10/50\n",
+      "18750/18750 [==============================] - 6s 339us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 11/50\n",
+      "18750/18750 [==============================] - 6s 342us/step - loss: 0.0491 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 12/50\n",
+      "18750/18750 [==============================] - 6s 342us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 13/50\n",
+      "18750/18750 [==============================] - 6s 341us/step - loss: 0.0489 - mean_absolute_error: 0.1800 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 14/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 15/50\n",
+      "18750/18750 [==============================] - 6s 338us/step - loss: 0.0491 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 16/50\n",
+      "18750/18750 [==============================] - 6s 338us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 17/50\n",
+      "18750/18750 [==============================] - 6s 343us/step - loss: 0.0489 - mean_absolute_error: 0.1801 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 18/50\n",
+      "18750/18750 [==============================] - 6s 341us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 19/50\n",
+      "18750/18750 [==============================] - 6s 340us/step - loss: 0.0489 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 20/50\n",
+      "18750/18750 [==============================] - 6s 338us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 21/50\n",
+      "18750/18750 [==============================] - 6s 343us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 22/50\n",
+      "18750/18750 [==============================] - 6s 341us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 23/50\n",
+      "18750/18750 [==============================] - 6s 340us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 24/50\n",
+      "18750/18750 [==============================] - 6s 341us/step - loss: 0.0489 - mean_absolute_error: 0.1801 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 25/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 0.0489 - mean_absolute_error: 0.1801 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 26/50\n",
+      "18750/18750 [==============================] - 6s 342us/step - loss: 0.0488 - mean_absolute_error: 0.1801 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 27/50\n",
+      "18750/18750 [==============================] - 6s 342us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 28/50\n",
+      "18750/18750 [==============================] - 6s 341us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 29/50\n",
+      "18750/18750 [==============================] - 6s 342us/step - loss: 0.0488 - mean_absolute_error: 0.1801 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 30/50\n",
+      "18750/18750 [==============================] - 6s 342us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 31/50\n",
+      "18750/18750 [==============================] - 6s 340us/step - loss: 0.0488 - mean_absolute_error: 0.1799 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 32/50\n",
+      "18750/18750 [==============================] - 6s 343us/step - loss: 0.0489 - mean_absolute_error: 0.1801 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 33/50\n",
+      "18750/18750 [==============================] - 6s 342us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 34/50\n",
+      "18750/18750 [==============================] - 6s 345us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 35/50\n",
+      "18750/18750 [==============================] - 6s 341us/step - loss: 0.0489 - mean_absolute_error: 0.1801 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 36/50\n",
+      "18750/18750 [==============================] - 6s 339us/step - loss: 0.0492 - mean_absolute_error: 0.1807 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 37/50\n",
+      "18750/18750 [==============================] - 6s 340us/step - loss: 0.0491 - mean_absolute_error: 0.1806 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 38/50\n",
+      "18750/18750 [==============================] - 6s 341us/step - loss: 0.0491 - mean_absolute_error: 0.1806 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 39/50\n",
+      "18750/18750 [==============================] - 6s 342us/step - loss: 0.0490 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 40/50\n",
+      "18750/18750 [==============================] - 6s 340us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 41/50\n",
+      "18750/18750 [==============================] - 6s 341us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 42/50\n",
+      "18750/18750 [==============================] - 6s 340us/step - loss: 0.0491 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 43/50\n",
+      "18750/18750 [==============================] - 6s 340us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 44/50\n",
+      "18750/18750 [==============================] - 6s 338us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 45/50\n",
+      "18750/18750 [==============================] - 6s 342us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 46/50\n",
+      "18750/18750 [==============================] - 6s 345us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 47/50\n",
+      "18750/18750 [==============================] - 6s 342us/step - loss: 0.0489 - mean_absolute_error: 0.1801 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 48/50\n",
+      "18750/18750 [==============================] - 6s 344us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 49/50\n",
+      "18750/18750 [==============================] - 6s 340us/step - loss: 0.0491 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 50/50\n",
+      "18750/18750 [==============================] - 6s 338us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n"
+     ]
+    }
+   ],
+   "source": [
+    "histories=np.zeros_like(models)\n",
+    "for i in range(len(models)):\n",
+    "    histories[i]=models[i].fit(X_train,y_train,\n",
+    "                               validation_data=(X_val,y_val),\n",
+    "                               batch_size=32,\n",
+    "                               epochs=50)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 9,
+   "id": "4e2a356e-29e0-4531-9f8f-2195af597476",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "image/png": "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",
+      "text/plain": [
+       "<Figure size 640x480 with 1 Axes>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    }
+   ],
+   "source": [
+    "colors=['orange','green','red','blue']\n",
+    "legend=['max_val=1.0','max_val=0.1','max_val=0.01','max_val=0.001']\n",
+    "for i in range(len(models)):\n",
+    "    plt.plot(histories[i].history['loss'],color=colors[i])\n",
+    "    plt.yscale('log')\n",
+    "plt.title('Model loss on the training set \\n for different values of the range of the \\n Random Uniform initializer')\n",
+    "plt.xlabel('epoch')\n",
+    "plt.ylabel('mean squared error')\n",
+    "plt.legend(legend,loc='upper right')\n",
+    "plt.show()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 10,
+   "id": "c7cd0ae8-fb4e-4a4a-86c1-87c5ee2638b5",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "image/png": "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",
+      "text/plain": [
+       "<Figure size 640x480 with 1 Axes>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    }
+   ],
+   "source": [
+    "colors=['orange','green','red','blue']\n",
+    "legend=['max_val=1.0','max_val=0.1','max_val=0.01','max_val=0.001']\n",
+    "for i in range(len(models)):\n",
+    "  plt.plot(histories[i].history['val_loss'],color=colors[i])\n",
+    "plt.title('Model loss on the validation set \\n for different values of the range of \\n the Random Uniform initializer')\n",
+    "plt.xlabel('epoch')\n",
+    "plt.ylabel('mean squared error')\n",
+    "plt.legend(legend,loc='upper right')\n",
+    "plt.show()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 31,
+   "id": "4e65e4c7-c4dc-4e53-8bac-a4b4a20d32a2",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "#Now choose the most appropriate initializer among: Random Normal (with standard deviation as selected above), \n",
+    "#Random Uniform (with range selected above), Glorot Normal, Glorot Uniform, He Normal, He Uniform, \n",
+    "#mixed (Glorot Uniform for sigmoid activation, He Uniform for ReLU activation)\n",
+    "my_initializers=[keras.initializers.RandomNormal(mean=0, stddev=0.1), keras.initializers.RandomUniform(minval=-1, maxval=1), keras.initializers.GlorotNormal, keras.initializers.GlorotUniform, keras.initializers.HeNormal, keras.initializers.HeUniform]\n",
+    "models=np.zeros_like(my_initializers)\n",
+    "for i in range(len(models)):\n",
+    "  models[i]=keras.models.Sequential()\n",
+    "  models[i].add(keras.layers.Dense(units=32, activation='relu', input_dim=X_train.shape[1], kernel_initializer=my_initializers[i]))\n",
+    "  models[i].add(keras.layers.Dense(units=32, activation='sigmoid', kernel_initializer=my_initializers[i]))\n",
+    "  models[i].add(keras.layers.Dense(units=64, activation='sigmoid', kernel_initializer=my_initializers[i]))\n",
+    "  models[i].add(keras.layers.Dense(units=1, activation='relu', kernel_initializer=my_initializers[i]))\n",
+    "  models[i].compile(optimizer='adam',\n",
+    "                    loss='mean_squared_error',\n",
+    "                    metrics=['mean_absolute_error'])"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 32,
+   "id": "9e2af9a5-fa82-4ed3-9143-8fc6513b501d",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Epoch 1/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 0.0150 - mean_absolute_error: 0.0996 - val_loss: 0.0106 - val_mean_absolute_error: 0.0802\n",
+      "Epoch 2/100\n",
+      "18750/18750 [==============================] - 7s 371us/step - loss: 0.0101 - mean_absolute_error: 0.0772 - val_loss: 0.0076 - val_mean_absolute_error: 0.0649\n",
+      "Epoch 3/100\n",
+      "18750/18750 [==============================] - 7s 365us/step - loss: 0.0077 - mean_absolute_error: 0.0655 - val_loss: 0.0063 - val_mean_absolute_error: 0.0588\n",
+      "Epoch 4/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 0.0062 - mean_absolute_error: 0.0573 - val_loss: 0.0047 - val_mean_absolute_error: 0.0485\n",
+      "Epoch 5/100\n",
+      "18750/18750 [==============================] - 7s 367us/step - loss: 0.0049 - mean_absolute_error: 0.0506 - val_loss: 0.0038 - val_mean_absolute_error: 0.0442\n",
+      "Epoch 6/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 0.0040 - mean_absolute_error: 0.0454 - val_loss: 0.0036 - val_mean_absolute_error: 0.0424\n",
+      "Epoch 7/100\n",
+      "18750/18750 [==============================] - 7s 355us/step - loss: 0.0035 - mean_absolute_error: 0.0424 - val_loss: 0.0035 - val_mean_absolute_error: 0.0429\n",
+      "Epoch 8/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 0.0031 - mean_absolute_error: 0.0396 - val_loss: 0.0028 - val_mean_absolute_error: 0.0390\n",
+      "Epoch 9/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 0.0027 - mean_absolute_error: 0.0369 - val_loss: 0.0021 - val_mean_absolute_error: 0.0313\n",
+      "Epoch 10/100\n",
+      "18750/18750 [==============================] - 7s 355us/step - loss: 0.0024 - mean_absolute_error: 0.0349 - val_loss: 0.0019 - val_mean_absolute_error: 0.0304\n",
+      "Epoch 11/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 0.0022 - mean_absolute_error: 0.0332 - val_loss: 0.0025 - val_mean_absolute_error: 0.0350\n",
+      "Epoch 12/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0020 - mean_absolute_error: 0.0319 - val_loss: 0.0017 - val_mean_absolute_error: 0.0289\n",
+      "Epoch 13/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 0.0019 - mean_absolute_error: 0.0303 - val_loss: 0.0016 - val_mean_absolute_error: 0.0278\n",
+      "Epoch 14/100\n",
+      "18750/18750 [==============================] - 7s 358us/step - loss: 0.0017 - mean_absolute_error: 0.0289 - val_loss: 0.0015 - val_mean_absolute_error: 0.0270\n",
+      "Epoch 15/100\n",
+      "18750/18750 [==============================] - 7s 355us/step - loss: 0.0015 - mean_absolute_error: 0.0276 - val_loss: 0.0017 - val_mean_absolute_error: 0.0299\n",
+      "Epoch 16/100\n",
+      "18750/18750 [==============================] - 7s 358us/step - loss: 0.0014 - mean_absolute_error: 0.0264 - val_loss: 0.0013 - val_mean_absolute_error: 0.0262\n",
+      "Epoch 17/100\n",
+      "18750/18750 [==============================] - 7s 358us/step - loss: 0.0013 - mean_absolute_error: 0.0255 - val_loss: 0.0012 - val_mean_absolute_error: 0.0253\n",
+      "Epoch 18/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 0.0012 - mean_absolute_error: 0.0247 - val_loss: 0.0011 - val_mean_absolute_error: 0.0245\n",
+      "Epoch 19/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 0.0011 - mean_absolute_error: 0.0241 - val_loss: 0.0011 - val_mean_absolute_error: 0.0248\n",
+      "Epoch 20/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 0.0011 - mean_absolute_error: 0.0236 - val_loss: 0.0011 - val_mean_absolute_error: 0.0230\n",
+      "Epoch 21/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 0.0010 - mean_absolute_error: 0.0229 - val_loss: 6.1882e-04 - val_mean_absolute_error: 0.0177\n",
+      "Epoch 22/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 9.8046e-04 - mean_absolute_error: 0.0224 - val_loss: 7.2949e-04 - val_mean_absolute_error: 0.0188\n",
+      "Epoch 23/100\n",
+      "18750/18750 [==============================] - 7s 358us/step - loss: 9.3654e-04 - mean_absolute_error: 0.0219 - val_loss: 8.9770e-04 - val_mean_absolute_error: 0.0230\n",
+      "Epoch 24/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 9.3012e-04 - mean_absolute_error: 0.0218 - val_loss: 7.3034e-04 - val_mean_absolute_error: 0.0194\n",
+      "Epoch 25/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 8.4976e-04 - mean_absolute_error: 0.0209 - val_loss: 0.0012 - val_mean_absolute_error: 0.0244\n",
+      "Epoch 26/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 8.5389e-04 - mean_absolute_error: 0.0209 - val_loss: 6.8117e-04 - val_mean_absolute_error: 0.0189\n",
+      "Epoch 27/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 8.1240e-04 - mean_absolute_error: 0.0205 - val_loss: 6.8420e-04 - val_mean_absolute_error: 0.0193\n",
+      "Epoch 28/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 7.8549e-04 - mean_absolute_error: 0.0201 - val_loss: 6.0740e-04 - val_mean_absolute_error: 0.0181\n",
+      "Epoch 29/100\n",
+      "18750/18750 [==============================] - 7s 358us/step - loss: 7.7322e-04 - mean_absolute_error: 0.0200 - val_loss: 8.3531e-04 - val_mean_absolute_error: 0.0205\n",
+      "Epoch 30/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 7.8297e-04 - mean_absolute_error: 0.0199 - val_loss: 8.0232e-04 - val_mean_absolute_error: 0.0216\n",
+      "Epoch 31/100\n",
+      "18750/18750 [==============================] - 7s 358us/step - loss: 7.1489e-04 - mean_absolute_error: 0.0193 - val_loss: 6.6555e-04 - val_mean_absolute_error: 0.0185\n",
+      "Epoch 32/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 7.1104e-04 - mean_absolute_error: 0.0192 - val_loss: 0.0010 - val_mean_absolute_error: 0.0246\n",
+      "Epoch 33/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 7.0358e-04 - mean_absolute_error: 0.0191 - val_loss: 4.5792e-04 - val_mean_absolute_error: 0.0155\n",
+      "Epoch 34/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 6.8518e-04 - mean_absolute_error: 0.0189 - val_loss: 5.4363e-04 - val_mean_absolute_error: 0.0165\n",
+      "Epoch 35/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 7.1538e-04 - mean_absolute_error: 0.0190 - val_loss: 0.0023 - val_mean_absolute_error: 0.0333\n",
+      "Epoch 36/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 8.9945e-04 - mean_absolute_error: 0.0215 - val_loss: 8.9341e-04 - val_mean_absolute_error: 0.0223\n",
+      "Epoch 37/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 7.5983e-04 - mean_absolute_error: 0.0199 - val_loss: 7.1836e-04 - val_mean_absolute_error: 0.0197\n",
+      "Epoch 38/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 7.1494e-04 - mean_absolute_error: 0.0193 - val_loss: 0.0011 - val_mean_absolute_error: 0.0246\n",
+      "Epoch 39/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 6.8852e-04 - mean_absolute_error: 0.0189 - val_loss: 6.4331e-04 - val_mean_absolute_error: 0.0180\n",
+      "Epoch 40/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 6.7312e-04 - mean_absolute_error: 0.0187 - val_loss: 7.5726e-04 - val_mean_absolute_error: 0.0206\n",
+      "Epoch 41/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 6.5881e-04 - mean_absolute_error: 0.0184 - val_loss: 5.8433e-04 - val_mean_absolute_error: 0.0185\n",
+      "Epoch 42/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 6.3876e-04 - mean_absolute_error: 0.0182 - val_loss: 5.7359e-04 - val_mean_absolute_error: 0.0176\n",
+      "Epoch 43/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 6.3324e-04 - mean_absolute_error: 0.0181 - val_loss: 4.5158e-04 - val_mean_absolute_error: 0.0156\n",
+      "Epoch 44/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 6.0175e-04 - mean_absolute_error: 0.0177 - val_loss: 5.5036e-04 - val_mean_absolute_error: 0.0172\n",
+      "Epoch 45/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 5.8963e-04 - mean_absolute_error: 0.0176 - val_loss: 5.6257e-04 - val_mean_absolute_error: 0.0172\n",
+      "Epoch 46/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 6.0319e-04 - mean_absolute_error: 0.0177 - val_loss: 4.0057e-04 - val_mean_absolute_error: 0.0146\n",
+      "Epoch 47/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 5.9196e-04 - mean_absolute_error: 0.0176 - val_loss: 6.5019e-04 - val_mean_absolute_error: 0.0191\n",
+      "Epoch 48/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 5.8436e-04 - mean_absolute_error: 0.0173 - val_loss: 4.2232e-04 - val_mean_absolute_error: 0.0149\n",
+      "Epoch 49/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 5.6281e-04 - mean_absolute_error: 0.0171 - val_loss: 6.6305e-04 - val_mean_absolute_error: 0.0189\n",
+      "Epoch 50/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 5.6831e-04 - mean_absolute_error: 0.0173 - val_loss: 5.7010e-04 - val_mean_absolute_error: 0.0168\n",
+      "Epoch 51/100\n",
+      "18750/18750 [==============================] - 7s 358us/step - loss: 5.4060e-04 - mean_absolute_error: 0.0169 - val_loss: 3.6157e-04 - val_mean_absolute_error: 0.0140\n",
+      "Epoch 52/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 5.4299e-04 - mean_absolute_error: 0.0170 - val_loss: 5.5452e-04 - val_mean_absolute_error: 0.0178\n",
+      "Epoch 53/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 5.3025e-04 - mean_absolute_error: 0.0167 - val_loss: 4.0140e-04 - val_mean_absolute_error: 0.0147\n",
+      "Epoch 54/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 5.3510e-04 - mean_absolute_error: 0.0168 - val_loss: 6.9715e-04 - val_mean_absolute_error: 0.0209\n",
+      "Epoch 55/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 5.2140e-04 - mean_absolute_error: 0.0166 - val_loss: 5.1850e-04 - val_mean_absolute_error: 0.0165\n",
+      "Epoch 56/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 5.1881e-04 - mean_absolute_error: 0.0165 - val_loss: 4.1124e-04 - val_mean_absolute_error: 0.0149\n",
+      "Epoch 57/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 5.2783e-04 - mean_absolute_error: 0.0166 - val_loss: 4.3714e-04 - val_mean_absolute_error: 0.0154\n",
+      "Epoch 58/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 5.0825e-04 - mean_absolute_error: 0.0164 - val_loss: 3.4175e-04 - val_mean_absolute_error: 0.0133\n",
+      "Epoch 59/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 4.9845e-04 - mean_absolute_error: 0.0162 - val_loss: 5.4887e-04 - val_mean_absolute_error: 0.0169\n",
+      "Epoch 60/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 4.9524e-04 - mean_absolute_error: 0.0161 - val_loss: 4.3515e-04 - val_mean_absolute_error: 0.0158\n",
+      "Epoch 61/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 4.9686e-04 - mean_absolute_error: 0.0162 - val_loss: 7.6235e-04 - val_mean_absolute_error: 0.0200\n",
+      "Epoch 62/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 4.8167e-04 - mean_absolute_error: 0.0160 - val_loss: 9.1578e-04 - val_mean_absolute_error: 0.0201\n",
+      "Epoch 63/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 4.8026e-04 - mean_absolute_error: 0.0159 - val_loss: 3.3618e-04 - val_mean_absolute_error: 0.0132\n",
+      "Epoch 64/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 4.7928e-04 - mean_absolute_error: 0.0159 - val_loss: 3.5298e-04 - val_mean_absolute_error: 0.0135\n",
+      "Epoch 65/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 4.7191e-04 - mean_absolute_error: 0.0158 - val_loss: 5.6708e-04 - val_mean_absolute_error: 0.0180\n",
+      "Epoch 66/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 4.7980e-04 - mean_absolute_error: 0.0158 - val_loss: 7.5094e-04 - val_mean_absolute_error: 0.0205\n",
+      "Epoch 67/100\n",
+      "18750/18750 [==============================] - 7s 365us/step - loss: 4.8079e-04 - mean_absolute_error: 0.0158 - val_loss: 4.5502e-04 - val_mean_absolute_error: 0.0156\n",
+      "Epoch 68/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 4.5753e-04 - mean_absolute_error: 0.0156 - val_loss: 4.1649e-04 - val_mean_absolute_error: 0.0149\n",
+      "Epoch 69/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 4.4907e-04 - mean_absolute_error: 0.0155 - val_loss: 5.5033e-04 - val_mean_absolute_error: 0.0170\n",
+      "Epoch 70/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 4.5128e-04 - mean_absolute_error: 0.0155 - val_loss: 3.6042e-04 - val_mean_absolute_error: 0.0139\n",
+      "Epoch 71/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 4.4466e-04 - mean_absolute_error: 0.0153 - val_loss: 3.8639e-04 - val_mean_absolute_error: 0.0143\n",
+      "Epoch 72/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 4.5375e-04 - mean_absolute_error: 0.0154 - val_loss: 3.4324e-04 - val_mean_absolute_error: 0.0135\n",
+      "Epoch 73/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 4.4767e-04 - mean_absolute_error: 0.0154 - val_loss: 6.6995e-04 - val_mean_absolute_error: 0.0198\n",
+      "Epoch 74/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 4.5291e-04 - mean_absolute_error: 0.0155 - val_loss: 2.7605e-04 - val_mean_absolute_error: 0.0119\n",
+      "Epoch 75/100\n",
+      "18750/18750 [==============================] - 7s 358us/step - loss: 4.7022e-04 - mean_absolute_error: 0.0155 - val_loss: 3.1130e-04 - val_mean_absolute_error: 0.0129\n",
+      "Epoch 76/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 4.3973e-04 - mean_absolute_error: 0.0152 - val_loss: 3.2427e-04 - val_mean_absolute_error: 0.0133\n",
+      "Epoch 77/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 4.3325e-04 - mean_absolute_error: 0.0151 - val_loss: 3.3348e-04 - val_mean_absolute_error: 0.0134\n",
+      "Epoch 78/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 4.1825e-04 - mean_absolute_error: 0.0149 - val_loss: 5.3084e-04 - val_mean_absolute_error: 0.0179\n",
+      "Epoch 79/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 4.2844e-04 - mean_absolute_error: 0.0150 - val_loss: 3.4043e-04 - val_mean_absolute_error: 0.0136\n",
+      "Epoch 80/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 4.3344e-04 - mean_absolute_error: 0.0151 - val_loss: 3.4693e-04 - val_mean_absolute_error: 0.0139\n",
+      "Epoch 81/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 4.2056e-04 - mean_absolute_error: 0.0149 - val_loss: 7.4632e-04 - val_mean_absolute_error: 0.0196\n",
+      "Epoch 82/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 4.1617e-04 - mean_absolute_error: 0.0149 - val_loss: 4.3403e-04 - val_mean_absolute_error: 0.0154\n",
+      "Epoch 83/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 4.2341e-04 - mean_absolute_error: 0.0150 - val_loss: 3.4780e-04 - val_mean_absolute_error: 0.0135\n",
+      "Epoch 84/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 4.1225e-04 - mean_absolute_error: 0.0148 - val_loss: 2.6244e-04 - val_mean_absolute_error: 0.0117\n",
+      "Epoch 85/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 4.2347e-04 - mean_absolute_error: 0.0150 - val_loss: 3.1248e-04 - val_mean_absolute_error: 0.0130\n",
+      "Epoch 86/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 4.0531e-04 - mean_absolute_error: 0.0147 - val_loss: 3.7431e-04 - val_mean_absolute_error: 0.0140\n",
+      "Epoch 87/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 4.1065e-04 - mean_absolute_error: 0.0146 - val_loss: 3.2736e-04 - val_mean_absolute_error: 0.0134\n",
+      "Epoch 88/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 4.0657e-04 - mean_absolute_error: 0.0147 - val_loss: 3.3716e-04 - val_mean_absolute_error: 0.0134\n",
+      "Epoch 89/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 3.9811e-04 - mean_absolute_error: 0.0146 - val_loss: 3.3978e-04 - val_mean_absolute_error: 0.0135\n",
+      "Epoch 90/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 3.9004e-04 - mean_absolute_error: 0.0145 - val_loss: 4.2286e-04 - val_mean_absolute_error: 0.0150\n",
+      "Epoch 91/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 3.9751e-04 - mean_absolute_error: 0.0146 - val_loss: 4.2561e-04 - val_mean_absolute_error: 0.0148\n",
+      "Epoch 92/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 4.0213e-04 - mean_absolute_error: 0.0146 - val_loss: 2.8386e-04 - val_mean_absolute_error: 0.0127\n",
+      "Epoch 93/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 3.9579e-04 - mean_absolute_error: 0.0145 - val_loss: 3.1730e-04 - val_mean_absolute_error: 0.0133\n",
+      "Epoch 94/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 3.8603e-04 - mean_absolute_error: 0.0144 - val_loss: 5.4730e-04 - val_mean_absolute_error: 0.0180\n",
+      "Epoch 95/100\n",
+      "18750/18750 [==============================] - 7s 358us/step - loss: 3.8787e-04 - mean_absolute_error: 0.0144 - val_loss: 3.3596e-04 - val_mean_absolute_error: 0.0134\n",
+      "Epoch 96/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 3.8487e-04 - mean_absolute_error: 0.0143 - val_loss: 3.6563e-04 - val_mean_absolute_error: 0.0146\n",
+      "Epoch 97/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 4.0596e-04 - mean_absolute_error: 0.0146 - val_loss: 3.0084e-04 - val_mean_absolute_error: 0.0127\n",
+      "Epoch 98/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 3.9142e-04 - mean_absolute_error: 0.0144 - val_loss: 5.0896e-04 - val_mean_absolute_error: 0.0171\n",
+      "Epoch 99/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 3.7833e-04 - mean_absolute_error: 0.0143 - val_loss: 3.2704e-04 - val_mean_absolute_error: 0.0136\n",
+      "Epoch 100/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 3.8635e-04 - mean_absolute_error: 0.0143 - val_loss: 2.7104e-04 - val_mean_absolute_error: 0.0119\n",
+      "Epoch 1/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 0.1270 - mean_absolute_error: 0.1602 - val_loss: 0.0169 - val_mean_absolute_error: 0.1058\n",
+      "Epoch 2/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 0.0170 - mean_absolute_error: 0.1068 - val_loss: 0.0174 - val_mean_absolute_error: 0.1100\n",
+      "Epoch 3/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 0.0168 - mean_absolute_error: 0.1062 - val_loss: 0.0157 - val_mean_absolute_error: 0.1032\n",
+      "Epoch 4/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 0.0160 - mean_absolute_error: 0.1041 - val_loss: 0.0164 - val_mean_absolute_error: 0.1033\n",
+      "Epoch 5/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 0.0159 - mean_absolute_error: 0.1039 - val_loss: 0.0157 - val_mean_absolute_error: 0.1039\n",
+      "Epoch 6/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 0.0158 - mean_absolute_error: 0.1035 - val_loss: 0.0159 - val_mean_absolute_error: 0.1032\n",
+      "Epoch 7/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 0.0157 - mean_absolute_error: 0.1034 - val_loss: 0.0155 - val_mean_absolute_error: 0.1033\n",
+      "Epoch 8/100\n",
+      "18750/18750 [==============================] - 7s 365us/step - loss: 0.0157 - mean_absolute_error: 0.1031 - val_loss: 0.0157 - val_mean_absolute_error: 0.1023\n",
+      "Epoch 9/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 0.0156 - mean_absolute_error: 0.1029 - val_loss: 0.0154 - val_mean_absolute_error: 0.1016\n",
+      "Epoch 10/100\n",
+      "18750/18750 [==============================] - 7s 366us/step - loss: 0.0156 - mean_absolute_error: 0.1026 - val_loss: 0.0154 - val_mean_absolute_error: 0.1012\n",
+      "Epoch 11/100\n",
+      "18750/18750 [==============================] - 7s 367us/step - loss: 0.0155 - mean_absolute_error: 0.1023 - val_loss: 0.0160 - val_mean_absolute_error: 0.1040\n",
+      "Epoch 12/100\n",
+      "18750/18750 [==============================] - 7s 367us/step - loss: 0.0157 - mean_absolute_error: 0.1030 - val_loss: 0.0159 - val_mean_absolute_error: 0.1053\n",
+      "Epoch 13/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 0.0154 - mean_absolute_error: 0.1019 - val_loss: 0.0159 - val_mean_absolute_error: 0.1049\n",
+      "Epoch 14/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 0.0152 - mean_absolute_error: 0.1009 - val_loss: 0.0144 - val_mean_absolute_error: 0.0973\n",
+      "Epoch 15/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 0.0147 - mean_absolute_error: 0.0981 - val_loss: 0.0129 - val_mean_absolute_error: 0.0897\n",
+      "Epoch 16/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 0.0129 - mean_absolute_error: 0.0885 - val_loss: 0.0125 - val_mean_absolute_error: 0.0879\n",
+      "Epoch 17/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 0.0124 - mean_absolute_error: 0.0861 - val_loss: 0.0162 - val_mean_absolute_error: 0.1015\n",
+      "Epoch 18/100\n",
+      "18750/18750 [==============================] - 7s 365us/step - loss: 0.0122 - mean_absolute_error: 0.0853 - val_loss: 0.0120 - val_mean_absolute_error: 0.0855\n",
+      "Epoch 19/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 0.0121 - mean_absolute_error: 0.0845 - val_loss: 0.0117 - val_mean_absolute_error: 0.0837\n",
+      "Epoch 20/100\n",
+      "18750/18750 [==============================] - 7s 365us/step - loss: 0.0119 - mean_absolute_error: 0.0839 - val_loss: 0.0118 - val_mean_absolute_error: 0.0841\n",
+      "Epoch 21/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 0.0113 - mean_absolute_error: 0.0818 - val_loss: 0.0117 - val_mean_absolute_error: 0.0847\n",
+      "Epoch 22/100\n",
+      "18750/18750 [==============================] - 7s 366us/step - loss: 0.0108 - mean_absolute_error: 0.0792 - val_loss: 0.0109 - val_mean_absolute_error: 0.0796\n",
+      "Epoch 23/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 0.0105 - mean_absolute_error: 0.0779 - val_loss: 0.0104 - val_mean_absolute_error: 0.0763\n",
+      "Epoch 24/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 0.0102 - mean_absolute_error: 0.0770 - val_loss: 0.0102 - val_mean_absolute_error: 0.0763\n",
+      "Epoch 25/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 0.0100 - mean_absolute_error: 0.0761 - val_loss: 0.0101 - val_mean_absolute_error: 0.0752\n",
+      "Epoch 26/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 0.0098 - mean_absolute_error: 0.0751 - val_loss: 0.0095 - val_mean_absolute_error: 0.0733\n",
+      "Epoch 27/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 0.0096 - mean_absolute_error: 0.0742 - val_loss: 0.0091 - val_mean_absolute_error: 0.0732\n",
+      "Epoch 28/100\n",
+      "18750/18750 [==============================] - 7s 366us/step - loss: 0.0094 - mean_absolute_error: 0.0734 - val_loss: 0.0095 - val_mean_absolute_error: 0.0729\n",
+      "Epoch 29/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 0.0091 - mean_absolute_error: 0.0723 - val_loss: 0.0099 - val_mean_absolute_error: 0.0780\n",
+      "Epoch 30/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 0.0088 - mean_absolute_error: 0.0708 - val_loss: 0.0084 - val_mean_absolute_error: 0.0701\n",
+      "Epoch 31/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 0.0084 - mean_absolute_error: 0.0693 - val_loss: 0.0079 - val_mean_absolute_error: 0.0660\n",
+      "Epoch 32/100\n",
+      "18750/18750 [==============================] - 7s 367us/step - loss: 0.0080 - mean_absolute_error: 0.0671 - val_loss: 0.0079 - val_mean_absolute_error: 0.0660\n",
+      "Epoch 33/100\n",
+      "18750/18750 [==============================] - 7s 368us/step - loss: 0.0077 - mean_absolute_error: 0.0660 - val_loss: 0.0086 - val_mean_absolute_error: 0.0701\n",
+      "Epoch 34/100\n",
+      "18750/18750 [==============================] - 7s 365us/step - loss: 0.0075 - mean_absolute_error: 0.0650 - val_loss: 0.0068 - val_mean_absolute_error: 0.0599\n",
+      "Epoch 35/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 0.0072 - mean_absolute_error: 0.0631 - val_loss: 0.0075 - val_mean_absolute_error: 0.0654\n",
+      "Epoch 36/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 0.0067 - mean_absolute_error: 0.0608 - val_loss: 0.0064 - val_mean_absolute_error: 0.0584\n",
+      "Epoch 37/100\n",
+      "18750/18750 [==============================] - 7s 366us/step - loss: 0.0064 - mean_absolute_error: 0.0589 - val_loss: 0.0067 - val_mean_absolute_error: 0.0636\n",
+      "Epoch 38/100\n",
+      "18750/18750 [==============================] - 7s 365us/step - loss: 0.0061 - mean_absolute_error: 0.0576 - val_loss: 0.0059 - val_mean_absolute_error: 0.0560\n",
+      "Epoch 39/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 0.0058 - mean_absolute_error: 0.0559 - val_loss: 0.0050 - val_mean_absolute_error: 0.0517\n",
+      "Epoch 40/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 0.0054 - mean_absolute_error: 0.0542 - val_loss: 0.0051 - val_mean_absolute_error: 0.0528\n",
+      "Epoch 41/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 0.0051 - mean_absolute_error: 0.0522 - val_loss: 0.0045 - val_mean_absolute_error: 0.0488\n",
+      "Epoch 42/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 0.0047 - mean_absolute_error: 0.0498 - val_loss: 0.0040 - val_mean_absolute_error: 0.0448\n",
+      "Epoch 43/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 0.0043 - mean_absolute_error: 0.0479 - val_loss: 0.0044 - val_mean_absolute_error: 0.0499\n",
+      "Epoch 44/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 0.0036 - mean_absolute_error: 0.0439 - val_loss: 0.0030 - val_mean_absolute_error: 0.0388\n",
+      "Epoch 45/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 0.0031 - mean_absolute_error: 0.0402 - val_loss: 0.0030 - val_mean_absolute_error: 0.0386\n",
+      "Epoch 46/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 0.0028 - mean_absolute_error: 0.0380 - val_loss: 0.0023 - val_mean_absolute_error: 0.0333\n",
+      "Epoch 47/100\n",
+      "18750/18750 [==============================] - 7s 366us/step - loss: 0.0026 - mean_absolute_error: 0.0360 - val_loss: 0.0023 - val_mean_absolute_error: 0.0348\n",
+      "Epoch 48/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 0.0024 - mean_absolute_error: 0.0349 - val_loss: 0.0022 - val_mean_absolute_error: 0.0334\n",
+      "Epoch 49/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 0.0023 - mean_absolute_error: 0.0338 - val_loss: 0.0021 - val_mean_absolute_error: 0.0318\n",
+      "Epoch 50/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 0.0021 - mean_absolute_error: 0.0327 - val_loss: 0.0019 - val_mean_absolute_error: 0.0305\n",
+      "Epoch 51/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 0.0020 - mean_absolute_error: 0.0318 - val_loss: 0.0020 - val_mean_absolute_error: 0.0314\n",
+      "Epoch 52/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 0.0018 - mean_absolute_error: 0.0307 - val_loss: 0.0018 - val_mean_absolute_error: 0.0305\n",
+      "Epoch 53/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 0.0017 - mean_absolute_error: 0.0299 - val_loss: 0.0014 - val_mean_absolute_error: 0.0262\n",
+      "Epoch 54/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 0.0016 - mean_absolute_error: 0.0288 - val_loss: 0.0015 - val_mean_absolute_error: 0.0276\n",
+      "Epoch 55/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 0.0015 - mean_absolute_error: 0.0274 - val_loss: 0.0015 - val_mean_absolute_error: 0.0270\n",
+      "Epoch 56/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 0.0013 - mean_absolute_error: 0.0263 - val_loss: 0.0013 - val_mean_absolute_error: 0.0261\n",
+      "Epoch 57/100\n",
+      "18750/18750 [==============================] - 7s 367us/step - loss: 0.0013 - mean_absolute_error: 0.0255 - val_loss: 0.0012 - val_mean_absolute_error: 0.0238\n",
+      "Epoch 58/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 0.0012 - mean_absolute_error: 0.0245 - val_loss: 0.0011 - val_mean_absolute_error: 0.0240\n",
+      "Epoch 59/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 0.0011 - mean_absolute_error: 0.0236 - val_loss: 0.0017 - val_mean_absolute_error: 0.0306\n",
+      "Epoch 60/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 0.0010 - mean_absolute_error: 0.0231 - val_loss: 0.0011 - val_mean_absolute_error: 0.0248\n",
+      "Epoch 61/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 9.5867e-04 - mean_absolute_error: 0.0224 - val_loss: 0.0011 - val_mean_absolute_error: 0.0248\n",
+      "Epoch 62/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 8.9303e-04 - mean_absolute_error: 0.0217 - val_loss: 7.3453e-04 - val_mean_absolute_error: 0.0198\n",
+      "Epoch 63/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 8.3587e-04 - mean_absolute_error: 0.0210 - val_loss: 6.4563e-04 - val_mean_absolute_error: 0.0184\n",
+      "Epoch 64/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 7.7967e-04 - mean_absolute_error: 0.0203 - val_loss: 7.6727e-04 - val_mean_absolute_error: 0.0207\n",
+      "Epoch 65/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 7.4728e-04 - mean_absolute_error: 0.0199 - val_loss: 6.2977e-04 - val_mean_absolute_error: 0.0187\n",
+      "Epoch 66/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 7.3397e-04 - mean_absolute_error: 0.0197 - val_loss: 9.9324e-04 - val_mean_absolute_error: 0.0237\n",
+      "Epoch 67/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 7.0124e-04 - mean_absolute_error: 0.0193 - val_loss: 5.8110e-04 - val_mean_absolute_error: 0.0175\n",
+      "Epoch 68/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 6.7863e-04 - mean_absolute_error: 0.0190 - val_loss: 5.2384e-04 - val_mean_absolute_error: 0.0167\n",
+      "Epoch 69/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 6.6019e-04 - mean_absolute_error: 0.0187 - val_loss: 7.8950e-04 - val_mean_absolute_error: 0.0202\n",
+      "Epoch 70/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 6.3529e-04 - mean_absolute_error: 0.0184 - val_loss: 7.3229e-04 - val_mean_absolute_error: 0.0203\n",
+      "Epoch 71/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 6.3725e-04 - mean_absolute_error: 0.0184 - val_loss: 9.7328e-04 - val_mean_absolute_error: 0.0220\n",
+      "Epoch 72/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 6.2533e-04 - mean_absolute_error: 0.0182 - val_loss: 4.3794e-04 - val_mean_absolute_error: 0.0153\n",
+      "Epoch 73/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 6.0310e-04 - mean_absolute_error: 0.0179 - val_loss: 5.5161e-04 - val_mean_absolute_error: 0.0171\n",
+      "Epoch 74/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 6.0297e-04 - mean_absolute_error: 0.0179 - val_loss: 6.9279e-04 - val_mean_absolute_error: 0.0199\n",
+      "Epoch 75/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 5.8374e-04 - mean_absolute_error: 0.0176 - val_loss: 7.9075e-04 - val_mean_absolute_error: 0.0200\n",
+      "Epoch 76/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 5.8004e-04 - mean_absolute_error: 0.0176 - val_loss: 5.0505e-04 - val_mean_absolute_error: 0.0163\n",
+      "Epoch 77/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 5.6256e-04 - mean_absolute_error: 0.0173 - val_loss: 5.4434e-04 - val_mean_absolute_error: 0.0174\n",
+      "Epoch 78/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 5.7255e-04 - mean_absolute_error: 0.0174 - val_loss: 7.4152e-04 - val_mean_absolute_error: 0.0207\n",
+      "Epoch 79/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 5.5975e-04 - mean_absolute_error: 0.0173 - val_loss: 5.0125e-04 - val_mean_absolute_error: 0.0171\n",
+      "Epoch 80/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 5.4982e-04 - mean_absolute_error: 0.0170 - val_loss: 4.4738e-04 - val_mean_absolute_error: 0.0152\n",
+      "Epoch 81/100\n",
+      "18750/18750 [==============================] - 7s 366us/step - loss: 5.6277e-04 - mean_absolute_error: 0.0173 - val_loss: 3.6504e-04 - val_mean_absolute_error: 0.0138\n",
+      "Epoch 82/100\n",
+      "18750/18750 [==============================] - 7s 365us/step - loss: 5.5491e-04 - mean_absolute_error: 0.0171 - val_loss: 8.6827e-04 - val_mean_absolute_error: 0.0235\n",
+      "Epoch 83/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 5.4222e-04 - mean_absolute_error: 0.0170 - val_loss: 6.4187e-04 - val_mean_absolute_error: 0.0188\n",
+      "Epoch 84/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 5.4358e-04 - mean_absolute_error: 0.0169 - val_loss: 4.7220e-04 - val_mean_absolute_error: 0.0165\n",
+      "Epoch 85/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 5.2508e-04 - mean_absolute_error: 0.0167 - val_loss: 5.9247e-04 - val_mean_absolute_error: 0.0171\n",
+      "Epoch 86/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 5.4289e-04 - mean_absolute_error: 0.0169 - val_loss: 9.7191e-04 - val_mean_absolute_error: 0.0243\n",
+      "Epoch 87/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 5.1589e-04 - mean_absolute_error: 0.0166 - val_loss: 4.6512e-04 - val_mean_absolute_error: 0.0158\n",
+      "Epoch 88/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 5.2248e-04 - mean_absolute_error: 0.0165 - val_loss: 8.3073e-04 - val_mean_absolute_error: 0.0231\n",
+      "Epoch 89/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 5.0159e-04 - mean_absolute_error: 0.0163 - val_loss: 4.1243e-04 - val_mean_absolute_error: 0.0144\n",
+      "Epoch 90/100\n",
+      "18750/18750 [==============================] - 7s 365us/step - loss: 5.0463e-04 - mean_absolute_error: 0.0164 - val_loss: 4.2216e-04 - val_mean_absolute_error: 0.0150\n",
+      "Epoch 91/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 5.0054e-04 - mean_absolute_error: 0.0163 - val_loss: 3.7637e-04 - val_mean_absolute_error: 0.0134\n",
+      "Epoch 92/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 4.9631e-04 - mean_absolute_error: 0.0162 - val_loss: 3.1038e-04 - val_mean_absolute_error: 0.0126\n",
+      "Epoch 93/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 4.9139e-04 - mean_absolute_error: 0.0160 - val_loss: 0.0011 - val_mean_absolute_error: 0.0246\n",
+      "Epoch 94/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 4.7813e-04 - mean_absolute_error: 0.0159 - val_loss: 4.2250e-04 - val_mean_absolute_error: 0.0147\n",
+      "Epoch 95/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 4.7804e-04 - mean_absolute_error: 0.0159 - val_loss: 3.2062e-04 - val_mean_absolute_error: 0.0129\n",
+      "Epoch 96/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 4.5880e-04 - mean_absolute_error: 0.0156 - val_loss: 4.4660e-04 - val_mean_absolute_error: 0.0159\n",
+      "Epoch 97/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 5.1498e-04 - mean_absolute_error: 0.0159 - val_loss: 6.7684e-04 - val_mean_absolute_error: 0.0198\n",
+      "Epoch 98/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 4.4076e-04 - mean_absolute_error: 0.0154 - val_loss: 4.7802e-04 - val_mean_absolute_error: 0.0157\n",
+      "Epoch 99/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 4.7131e-04 - mean_absolute_error: 0.0157 - val_loss: 4.9443e-04 - val_mean_absolute_error: 0.0172\n",
+      "Epoch 100/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 4.5786e-04 - mean_absolute_error: 0.0155 - val_loss: 5.8704e-04 - val_mean_absolute_error: 0.0173\n",
+      "Epoch 1/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 0.0492 - mean_absolute_error: 0.1807 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 2/100\n",
+      "18750/18750 [==============================] - 7s 355us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 3/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0491 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 4/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 5/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 6/100\n",
+      "18750/18750 [==============================] - 7s 355us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 7/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 8/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 0.0489 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 9/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 10/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 11/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0492 - mean_absolute_error: 0.1807 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 12/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 0.0489 - mean_absolute_error: 0.1801 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 13/100\n",
+      "18750/18750 [==============================] - 7s 355us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 14/100\n",
+      "18750/18750 [==============================] - 7s 354us/step - loss: 0.0490 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 15/100\n",
+      "18750/18750 [==============================] - 7s 355us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 16/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 0.0491 - mean_absolute_error: 0.1806 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 17/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 18/100\n",
+      "18750/18750 [==============================] - 7s 365us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 19/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 20/100\n",
+      "18750/18750 [==============================] - 7s 368us/step - loss: 0.0491 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 21/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 0.0491 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 22/100\n",
+      "18750/18750 [==============================] - 7s 355us/step - loss: 0.0489 - mean_absolute_error: 0.1801 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 23/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 0.0490 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 24/100\n",
+      "18750/18750 [==============================] - 7s 358us/step - loss: 0.0489 - mean_absolute_error: 0.1801 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 25/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 0.0491 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 26/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 27/100\n",
+      "18750/18750 [==============================] - 7s 355us/step - loss: 0.0490 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 28/100\n",
+      "18750/18750 [==============================] - 7s 354us/step - loss: 0.0491 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 29/100\n",
+      "18750/18750 [==============================] - 7s 354us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 30/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 31/100\n",
+      "18750/18750 [==============================] - 7s 353us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 32/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0491 - mean_absolute_error: 0.1807 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 33/100\n",
+      "18750/18750 [==============================] - 7s 353us/step - loss: 0.0491 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 34/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 35/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 36/100\n",
+      "18750/18750 [==============================] - 7s 355us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 37/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0490 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 38/100\n",
+      "18750/18750 [==============================] - 7s 354us/step - loss: 0.0491 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 39/100\n",
+      "18750/18750 [==============================] - 7s 355us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 40/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 41/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 42/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0489 - mean_absolute_error: 0.1801 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 43/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 0.0492 - mean_absolute_error: 0.1806 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 44/100\n",
+      "18750/18750 [==============================] - 7s 354us/step - loss: 0.0489 - mean_absolute_error: 0.1801 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 45/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 46/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 47/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 48/100\n",
+      "18750/18750 [==============================] - 7s 355us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 49/100\n",
+      "18750/18750 [==============================] - 7s 355us/step - loss: 0.0489 - mean_absolute_error: 0.1801 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 50/100\n",
+      "18750/18750 [==============================] - 7s 355us/step - loss: 0.0489 - mean_absolute_error: 0.1801 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 51/100\n",
+      "18750/18750 [==============================] - 7s 353us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 52/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 53/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 0.0491 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 54/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 0.0491 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 55/100\n",
+      "18750/18750 [==============================] - 7s 354us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 56/100\n",
+      "18750/18750 [==============================] - 7s 354us/step - loss: 0.0490 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 57/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 58/100\n",
+      "18750/18750 [==============================] - 7s 354us/step - loss: 0.0491 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 59/100\n",
+      "18750/18750 [==============================] - 7s 355us/step - loss: 0.0490 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 60/100\n",
+      "18750/18750 [==============================] - 7s 355us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 61/100\n",
+      "18750/18750 [==============================] - 7s 355us/step - loss: 0.0490 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 62/100\n",
+      "18750/18750 [==============================] - 7s 355us/step - loss: 0.0491 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 63/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 64/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0491 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 65/100\n",
+      "18750/18750 [==============================] - 7s 358us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 66/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0489 - mean_absolute_error: 0.1801 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 67/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 68/100\n",
+      "18750/18750 [==============================] - 7s 355us/step - loss: 0.0489 - mean_absolute_error: 0.1801 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 69/100\n",
+      "18750/18750 [==============================] - 7s 358us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 70/100\n",
+      "18750/18750 [==============================] - 7s 358us/step - loss: 0.0489 - mean_absolute_error: 0.1801 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 71/100\n",
+      "18750/18750 [==============================] - 7s 358us/step - loss: 0.0488 - mean_absolute_error: 0.1800 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 72/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 0.0490 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 73/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 74/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0488 - mean_absolute_error: 0.1800 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 75/100\n",
+      "18750/18750 [==============================] - 7s 355us/step - loss: 0.0490 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 76/100\n",
+      "18750/18750 [==============================] - 7s 354us/step - loss: 0.0491 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 77/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 0.0490 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 78/100\n",
+      "18750/18750 [==============================] - 7s 353us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 79/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 80/100\n",
+      "18750/18750 [==============================] - 7s 355us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 81/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 82/100\n",
+      "18750/18750 [==============================] - 7s 355us/step - loss: 0.0492 - mean_absolute_error: 0.1806 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 83/100\n",
+      "18750/18750 [==============================] - 7s 354us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 84/100\n",
+      "18750/18750 [==============================] - 7s 354us/step - loss: 0.0489 - mean_absolute_error: 0.1800 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 85/100\n",
+      "18750/18750 [==============================] - 7s 355us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 86/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 0.0489 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 87/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 88/100\n",
+      "18750/18750 [==============================] - 7s 355us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 89/100\n",
+      "18750/18750 [==============================] - 7s 355us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 90/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 0.0490 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 91/100\n",
+      "18750/18750 [==============================] - 7s 358us/step - loss: 0.0491 - mean_absolute_error: 0.1806 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 92/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 93/100\n",
+      "18750/18750 [==============================] - 7s 355us/step - loss: 0.0488 - mean_absolute_error: 0.1800 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 94/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 0.0489 - mean_absolute_error: 0.1801 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 95/100\n",
+      "18750/18750 [==============================] - 7s 354us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 96/100\n",
+      "18750/18750 [==============================] - 7s 355us/step - loss: 0.0492 - mean_absolute_error: 0.1808 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 97/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 0.0489 - mean_absolute_error: 0.1801 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 98/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 99/100\n",
+      "18750/18750 [==============================] - 7s 358us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 100/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 1/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 2/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 3/100\n",
+      "18750/18750 [==============================] - 7s 358us/step - loss: 0.0490 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 4/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 0.0489 - mean_absolute_error: 0.1801 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 5/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 6/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 7/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0490 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 8/100\n",
+      "18750/18750 [==============================] - 7s 354us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 9/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 10/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 0.0489 - mean_absolute_error: 0.1801 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 11/100\n",
+      "18750/18750 [==============================] - 7s 358us/step - loss: 0.0491 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 12/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 0.0489 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 13/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 0.0490 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 14/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 15/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 16/100\n",
+      "18750/18750 [==============================] - 7s 358us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 17/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0489 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 18/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 0.0491 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 19/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 0.0491 - mean_absolute_error: 0.1806 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 20/100\n",
+      "18750/18750 [==============================] - 7s 355us/step - loss: 0.0491 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 21/100\n",
+      "18750/18750 [==============================] - 7s 354us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 22/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 23/100\n",
+      "18750/18750 [==============================] - 7s 355us/step - loss: 0.0491 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 24/100\n",
+      "18750/18750 [==============================] - 7s 354us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 25/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 26/100\n",
+      "18750/18750 [==============================] - 7s 355us/step - loss: 0.0488 - mean_absolute_error: 0.1800 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 27/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 28/100\n",
+      "18750/18750 [==============================] - 7s 354us/step - loss: 0.0491 - mean_absolute_error: 0.1807 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 29/100\n",
+      "18750/18750 [==============================] - 7s 354us/step - loss: 0.0490 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 30/100\n",
+      "18750/18750 [==============================] - 7s 353us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 31/100\n",
+      "18750/18750 [==============================] - 7s 355us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 32/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0488 - mean_absolute_error: 0.1801 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 33/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 0.0490 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 34/100\n",
+      "18750/18750 [==============================] - 7s 358us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 35/100\n",
+      "18750/18750 [==============================] - 7s 358us/step - loss: 0.0491 - mean_absolute_error: 0.1807 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 36/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0491 - mean_absolute_error: 0.1806 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 37/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 38/100\n",
+      "18750/18750 [==============================] - 7s 355us/step - loss: 0.0489 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 39/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 40/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 41/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 0.0491 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 42/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 43/100\n",
+      "18750/18750 [==============================] - 7s 355us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 44/100\n",
+      "18750/18750 [==============================] - 7s 358us/step - loss: 0.0491 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 45/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 0.0491 - mean_absolute_error: 0.1806 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 46/100\n",
+      "18750/18750 [==============================] - 7s 358us/step - loss: 0.0491 - mean_absolute_error: 0.1807 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 47/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0489 - mean_absolute_error: 0.1801 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 48/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 49/100\n",
+      "18750/18750 [==============================] - 7s 355us/step - loss: 0.0490 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 50/100\n",
+      "18750/18750 [==============================] - 7s 358us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 51/100\n",
+      "18750/18750 [==============================] - 7s 354us/step - loss: 0.0489 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 52/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 0.0491 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 53/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0490 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 54/100\n",
+      "18750/18750 [==============================] - 7s 355us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 55/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0491 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 56/100\n",
+      "18750/18750 [==============================] - 7s 354us/step - loss: 0.0490 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 57/100\n",
+      "18750/18750 [==============================] - 7s 355us/step - loss: 0.0490 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 58/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 59/100\n",
+      "18750/18750 [==============================] - 7s 355us/step - loss: 0.0490 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 60/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 61/100\n",
+      "18750/18750 [==============================] - 7s 358us/step - loss: 0.0489 - mean_absolute_error: 0.1801 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 62/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 63/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0490 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 64/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 0.0491 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 65/100\n",
+      "18750/18750 [==============================] - 7s 354us/step - loss: 0.0491 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 66/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0488 - mean_absolute_error: 0.1800 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 67/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 68/100\n",
+      "18750/18750 [==============================] - 7s 355us/step - loss: 0.0489 - mean_absolute_error: 0.1801 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 69/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 70/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 0.0489 - mean_absolute_error: 0.1801 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 71/100\n",
+      "18750/18750 [==============================] - 7s 358us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 72/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 73/100\n",
+      "18750/18750 [==============================] - 7s 355us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 74/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 75/100\n",
+      "18750/18750 [==============================] - 7s 355us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 76/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0489 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 77/100\n",
+      "18750/18750 [==============================] - 7s 353us/step - loss: 0.0491 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 78/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0489 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 79/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 80/100\n",
+      "18750/18750 [==============================] - 7s 358us/step - loss: 0.0490 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 81/100\n",
+      "18750/18750 [==============================] - 7s 355us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 82/100\n",
+      "18750/18750 [==============================] - 7s 354us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 83/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 84/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 85/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 0.0491 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 86/100\n",
+      "18750/18750 [==============================] - 7s 354us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 87/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 88/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 89/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 90/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 0.0491 - mean_absolute_error: 0.1806 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 91/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 0.0491 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 92/100\n",
+      "18750/18750 [==============================] - 7s 358us/step - loss: 0.0488 - mean_absolute_error: 0.1800 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 93/100\n",
+      "18750/18750 [==============================] - 7s 355us/step - loss: 0.0490 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 94/100\n",
+      "18750/18750 [==============================] - 7s 355us/step - loss: 0.0491 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 95/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 0.0489 - mean_absolute_error: 0.1803 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 96/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0491 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 97/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0489 - mean_absolute_error: 0.1802 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 98/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0491 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 99/100\n",
+      "18750/18750 [==============================] - 7s 355us/step - loss: 0.0490 - mean_absolute_error: 0.1804 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 100/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 0.0490 - mean_absolute_error: 0.1805 - val_loss: 0.0491 - val_mean_absolute_error: 0.1804\n",
+      "Epoch 1/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 0.0165 - mean_absolute_error: 0.1054 - val_loss: 0.0157 - val_mean_absolute_error: 0.1023\n",
+      "Epoch 2/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 0.0159 - mean_absolute_error: 0.1038 - val_loss: 0.0155 - val_mean_absolute_error: 0.1025\n",
+      "Epoch 3/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 0.0156 - mean_absolute_error: 0.1031 - val_loss: 0.0154 - val_mean_absolute_error: 0.1014\n",
+      "Epoch 4/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 0.0155 - mean_absolute_error: 0.1023 - val_loss: 0.0150 - val_mean_absolute_error: 0.1001\n",
+      "Epoch 5/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 0.0149 - mean_absolute_error: 0.0995 - val_loss: 0.0122 - val_mean_absolute_error: 0.0861\n",
+      "Epoch 6/100\n",
+      "18750/18750 [==============================] - 7s 365us/step - loss: 0.0123 - mean_absolute_error: 0.0859 - val_loss: 0.0118 - val_mean_absolute_error: 0.0826\n",
+      "Epoch 7/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 0.0116 - mean_absolute_error: 0.0828 - val_loss: 0.0113 - val_mean_absolute_error: 0.0820\n",
+      "Epoch 8/100\n",
+      "18750/18750 [==============================] - 7s 365us/step - loss: 0.0113 - mean_absolute_error: 0.0816 - val_loss: 0.0108 - val_mean_absolute_error: 0.0792\n",
+      "Epoch 9/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 0.0109 - mean_absolute_error: 0.0799 - val_loss: 0.0107 - val_mean_absolute_error: 0.0781\n",
+      "Epoch 10/100\n",
+      "18750/18750 [==============================] - 7s 366us/step - loss: 0.0107 - mean_absolute_error: 0.0789 - val_loss: 0.0112 - val_mean_absolute_error: 0.0815\n",
+      "Epoch 11/100\n",
+      "18750/18750 [==============================] - 7s 366us/step - loss: 0.0104 - mean_absolute_error: 0.0778 - val_loss: 0.0102 - val_mean_absolute_error: 0.0760\n",
+      "Epoch 12/100\n",
+      "18750/18750 [==============================] - 7s 366us/step - loss: 0.0101 - mean_absolute_error: 0.0762 - val_loss: 0.0097 - val_mean_absolute_error: 0.0745\n",
+      "Epoch 13/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 0.0097 - mean_absolute_error: 0.0745 - val_loss: 0.0092 - val_mean_absolute_error: 0.0716\n",
+      "Epoch 14/100\n",
+      "18750/18750 [==============================] - 7s 365us/step - loss: 0.0096 - mean_absolute_error: 0.0737 - val_loss: 0.0090 - val_mean_absolute_error: 0.0712\n",
+      "Epoch 15/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 0.0092 - mean_absolute_error: 0.0723 - val_loss: 0.0089 - val_mean_absolute_error: 0.0701\n",
+      "Epoch 16/100\n",
+      "18750/18750 [==============================] - 7s 367us/step - loss: 0.0090 - mean_absolute_error: 0.0713 - val_loss: 0.0091 - val_mean_absolute_error: 0.0717\n",
+      "Epoch 17/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 0.0086 - mean_absolute_error: 0.0691 - val_loss: 0.0077 - val_mean_absolute_error: 0.0649\n",
+      "Epoch 18/100\n",
+      "18750/18750 [==============================] - 7s 365us/step - loss: 0.0075 - mean_absolute_error: 0.0642 - val_loss: 0.0071 - val_mean_absolute_error: 0.0615\n",
+      "Epoch 19/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 0.0069 - mean_absolute_error: 0.0610 - val_loss: 0.0072 - val_mean_absolute_error: 0.0625\n",
+      "Epoch 20/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 0.0063 - mean_absolute_error: 0.0574 - val_loss: 0.0056 - val_mean_absolute_error: 0.0531\n",
+      "Epoch 21/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 0.0058 - mean_absolute_error: 0.0550 - val_loss: 0.0051 - val_mean_absolute_error: 0.0504\n",
+      "Epoch 22/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 0.0053 - mean_absolute_error: 0.0520 - val_loss: 0.0045 - val_mean_absolute_error: 0.0475\n",
+      "Epoch 23/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 0.0047 - mean_absolute_error: 0.0485 - val_loss: 0.0045 - val_mean_absolute_error: 0.0475\n",
+      "Epoch 24/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 0.0040 - mean_absolute_error: 0.0450 - val_loss: 0.0040 - val_mean_absolute_error: 0.0469\n",
+      "Epoch 25/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 0.0036 - mean_absolute_error: 0.0426 - val_loss: 0.0044 - val_mean_absolute_error: 0.0475\n",
+      "Epoch 26/100\n",
+      "18750/18750 [==============================] - 7s 366us/step - loss: 0.0031 - mean_absolute_error: 0.0397 - val_loss: 0.0028 - val_mean_absolute_error: 0.0380\n",
+      "Epoch 27/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 0.0027 - mean_absolute_error: 0.0370 - val_loss: 0.0030 - val_mean_absolute_error: 0.0387\n",
+      "Epoch 28/100\n",
+      "18750/18750 [==============================] - 7s 367us/step - loss: 0.0023 - mean_absolute_error: 0.0344 - val_loss: 0.0024 - val_mean_absolute_error: 0.0341\n",
+      "Epoch 29/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 0.0021 - mean_absolute_error: 0.0326 - val_loss: 0.0028 - val_mean_absolute_error: 0.0381\n",
+      "Epoch 30/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 0.0020 - mean_absolute_error: 0.0316 - val_loss: 0.0019 - val_mean_absolute_error: 0.0317\n",
+      "Epoch 31/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 0.0019 - mean_absolute_error: 0.0308 - val_loss: 0.0015 - val_mean_absolute_error: 0.0278\n",
+      "Epoch 32/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 0.0017 - mean_absolute_error: 0.0293 - val_loss: 0.0013 - val_mean_absolute_error: 0.0263\n",
+      "Epoch 33/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 0.0015 - mean_absolute_error: 0.0281 - val_loss: 0.0011 - val_mean_absolute_error: 0.0233\n",
+      "Epoch 34/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 0.0014 - mean_absolute_error: 0.0268 - val_loss: 0.0011 - val_mean_absolute_error: 0.0242\n",
+      "Epoch 35/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 0.0013 - mean_absolute_error: 0.0260 - val_loss: 0.0011 - val_mean_absolute_error: 0.0245\n",
+      "Epoch 36/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 0.0012 - mean_absolute_error: 0.0250 - val_loss: 0.0019 - val_mean_absolute_error: 0.0345\n",
+      "Epoch 37/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 0.0012 - mean_absolute_error: 0.0246 - val_loss: 9.4716e-04 - val_mean_absolute_error: 0.0229\n",
+      "Epoch 38/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 0.0011 - mean_absolute_error: 0.0237 - val_loss: 9.3770e-04 - val_mean_absolute_error: 0.0218\n",
+      "Epoch 39/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 0.0010 - mean_absolute_error: 0.0233 - val_loss: 0.0014 - val_mean_absolute_error: 0.0276\n",
+      "Epoch 40/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 9.7522e-04 - mean_absolute_error: 0.0227 - val_loss: 0.0010 - val_mean_absolute_error: 0.0228\n",
+      "Epoch 41/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 9.5104e-04 - mean_absolute_error: 0.0223 - val_loss: 6.0226e-04 - val_mean_absolute_error: 0.0176\n",
+      "Epoch 42/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 8.9702e-04 - mean_absolute_error: 0.0218 - val_loss: 0.0015 - val_mean_absolute_error: 0.0268\n",
+      "Epoch 43/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 8.6049e-04 - mean_absolute_error: 0.0213 - val_loss: 5.9147e-04 - val_mean_absolute_error: 0.0175\n",
+      "Epoch 44/100\n",
+      "18750/18750 [==============================] - 7s 366us/step - loss: 8.4166e-04 - mean_absolute_error: 0.0210 - val_loss: 8.2422e-04 - val_mean_absolute_error: 0.0215\n",
+      "Epoch 45/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 7.6746e-04 - mean_absolute_error: 0.0202 - val_loss: 5.8282e-04 - val_mean_absolute_error: 0.0170\n",
+      "Epoch 46/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 7.7737e-04 - mean_absolute_error: 0.0202 - val_loss: 5.7449e-04 - val_mean_absolute_error: 0.0175\n",
+      "Epoch 47/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 7.5904e-04 - mean_absolute_error: 0.0199 - val_loss: 5.5269e-04 - val_mean_absolute_error: 0.0173\n",
+      "Epoch 48/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 7.3981e-04 - mean_absolute_error: 0.0197 - val_loss: 5.9635e-04 - val_mean_absolute_error: 0.0179\n",
+      "Epoch 49/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 7.4019e-04 - mean_absolute_error: 0.0197 - val_loss: 6.2830e-04 - val_mean_absolute_error: 0.0182\n",
+      "Epoch 50/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 7.0524e-04 - mean_absolute_error: 0.0192 - val_loss: 6.7813e-04 - val_mean_absolute_error: 0.0188\n",
+      "Epoch 51/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 6.9473e-04 - mean_absolute_error: 0.0191 - val_loss: 7.3266e-04 - val_mean_absolute_error: 0.0203\n",
+      "Epoch 52/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 6.9849e-04 - mean_absolute_error: 0.0191 - val_loss: 5.9904e-04 - val_mean_absolute_error: 0.0178\n",
+      "Epoch 53/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 6.8526e-04 - mean_absolute_error: 0.0190 - val_loss: 9.8613e-04 - val_mean_absolute_error: 0.0252\n",
+      "Epoch 54/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 7.0058e-04 - mean_absolute_error: 0.0191 - val_loss: 6.6902e-04 - val_mean_absolute_error: 0.0191\n",
+      "Epoch 55/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 6.7038e-04 - mean_absolute_error: 0.0187 - val_loss: 4.6837e-04 - val_mean_absolute_error: 0.0160\n",
+      "Epoch 56/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 6.6695e-04 - mean_absolute_error: 0.0188 - val_loss: 6.6198e-04 - val_mean_absolute_error: 0.0190\n",
+      "Epoch 57/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 6.5161e-04 - mean_absolute_error: 0.0185 - val_loss: 6.5184e-04 - val_mean_absolute_error: 0.0191\n",
+      "Epoch 58/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 6.4593e-04 - mean_absolute_error: 0.0184 - val_loss: 6.7298e-04 - val_mean_absolute_error: 0.0205\n",
+      "Epoch 59/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 6.5881e-04 - mean_absolute_error: 0.0186 - val_loss: 6.2784e-04 - val_mean_absolute_error: 0.0195\n",
+      "Epoch 60/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 6.4830e-04 - mean_absolute_error: 0.0184 - val_loss: 4.5519e-04 - val_mean_absolute_error: 0.0158\n",
+      "Epoch 61/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 6.3590e-04 - mean_absolute_error: 0.0183 - val_loss: 5.4012e-04 - val_mean_absolute_error: 0.0167\n",
+      "Epoch 62/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 6.3265e-04 - mean_absolute_error: 0.0183 - val_loss: 6.2915e-04 - val_mean_absolute_error: 0.0187\n",
+      "Epoch 63/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 6.2927e-04 - mean_absolute_error: 0.0182 - val_loss: 6.5929e-04 - val_mean_absolute_error: 0.0193\n",
+      "Epoch 64/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 6.1288e-04 - mean_absolute_error: 0.0180 - val_loss: 4.5137e-04 - val_mean_absolute_error: 0.0156\n",
+      "Epoch 65/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 6.0204e-04 - mean_absolute_error: 0.0178 - val_loss: 5.5783e-04 - val_mean_absolute_error: 0.0174\n",
+      "Epoch 66/100\n",
+      "18750/18750 [==============================] - 7s 365us/step - loss: 6.2032e-04 - mean_absolute_error: 0.0181 - val_loss: 5.6639e-04 - val_mean_absolute_error: 0.0173\n",
+      "Epoch 67/100\n",
+      "18750/18750 [==============================] - 7s 367us/step - loss: 6.1125e-04 - mean_absolute_error: 0.0178 - val_loss: 5.2804e-04 - val_mean_absolute_error: 0.0169\n",
+      "Epoch 68/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 5.9464e-04 - mean_absolute_error: 0.0177 - val_loss: 7.4432e-04 - val_mean_absolute_error: 0.0195\n",
+      "Epoch 69/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 6.0688e-04 - mean_absolute_error: 0.0177 - val_loss: 8.2405e-04 - val_mean_absolute_error: 0.0214\n",
+      "Epoch 70/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 5.9729e-04 - mean_absolute_error: 0.0177 - val_loss: 6.1181e-04 - val_mean_absolute_error: 0.0186\n",
+      "Epoch 71/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 5.9481e-04 - mean_absolute_error: 0.0176 - val_loss: 7.0983e-04 - val_mean_absolute_error: 0.0194\n",
+      "Epoch 72/100\n",
+      "18750/18750 [==============================] - 7s 367us/step - loss: 5.8441e-04 - mean_absolute_error: 0.0175 - val_loss: 4.4231e-04 - val_mean_absolute_error: 0.0153\n",
+      "Epoch 73/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 5.7132e-04 - mean_absolute_error: 0.0173 - val_loss: 5.0463e-04 - val_mean_absolute_error: 0.0166\n",
+      "Epoch 74/100\n",
+      "18750/18750 [==============================] - 7s 366us/step - loss: 5.7265e-04 - mean_absolute_error: 0.0174 - val_loss: 5.1096e-04 - val_mean_absolute_error: 0.0162\n",
+      "Epoch 75/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 5.7707e-04 - mean_absolute_error: 0.0174 - val_loss: 6.8017e-04 - val_mean_absolute_error: 0.0184\n",
+      "Epoch 76/100\n",
+      "18750/18750 [==============================] - 7s 365us/step - loss: 5.7570e-04 - mean_absolute_error: 0.0174 - val_loss: 4.6310e-04 - val_mean_absolute_error: 0.0154\n",
+      "Epoch 77/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 5.5860e-04 - mean_absolute_error: 0.0172 - val_loss: 3.8882e-04 - val_mean_absolute_error: 0.0142\n",
+      "Epoch 78/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 5.4337e-04 - mean_absolute_error: 0.0169 - val_loss: 4.0328e-04 - val_mean_absolute_error: 0.0146\n",
+      "Epoch 79/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 5.6962e-04 - mean_absolute_error: 0.0173 - val_loss: 6.2615e-04 - val_mean_absolute_error: 0.0179\n",
+      "Epoch 80/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 5.4498e-04 - mean_absolute_error: 0.0170 - val_loss: 6.2189e-04 - val_mean_absolute_error: 0.0185\n",
+      "Epoch 81/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 5.4304e-04 - mean_absolute_error: 0.0169 - val_loss: 4.3306e-04 - val_mean_absolute_error: 0.0152\n",
+      "Epoch 82/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 5.4121e-04 - mean_absolute_error: 0.0169 - val_loss: 3.6010e-04 - val_mean_absolute_error: 0.0137\n",
+      "Epoch 83/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 5.3484e-04 - mean_absolute_error: 0.0168 - val_loss: 4.1020e-04 - val_mean_absolute_error: 0.0149\n",
+      "Epoch 84/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 5.2267e-04 - mean_absolute_error: 0.0166 - val_loss: 4.2822e-04 - val_mean_absolute_error: 0.0152\n",
+      "Epoch 85/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 5.1975e-04 - mean_absolute_error: 0.0166 - val_loss: 3.5975e-04 - val_mean_absolute_error: 0.0136\n",
+      "Epoch 86/100\n",
+      "18750/18750 [==============================] - 7s 365us/step - loss: 5.2816e-04 - mean_absolute_error: 0.0167 - val_loss: 3.5296e-04 - val_mean_absolute_error: 0.0136\n",
+      "Epoch 87/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 5.2140e-04 - mean_absolute_error: 0.0166 - val_loss: 5.5337e-04 - val_mean_absolute_error: 0.0172\n",
+      "Epoch 88/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 5.0899e-04 - mean_absolute_error: 0.0164 - val_loss: 4.6756e-04 - val_mean_absolute_error: 0.0158\n",
+      "Epoch 89/100\n",
+      "18750/18750 [==============================] - 7s 366us/step - loss: 5.1143e-04 - mean_absolute_error: 0.0164 - val_loss: 5.7264e-04 - val_mean_absolute_error: 0.0183\n",
+      "Epoch 90/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 5.0084e-04 - mean_absolute_error: 0.0163 - val_loss: 5.3241e-04 - val_mean_absolute_error: 0.0171\n",
+      "Epoch 91/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 5.1482e-04 - mean_absolute_error: 0.0165 - val_loss: 3.6940e-04 - val_mean_absolute_error: 0.0138\n",
+      "Epoch 92/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 5.0302e-04 - mean_absolute_error: 0.0163 - val_loss: 5.4232e-04 - val_mean_absolute_error: 0.0172\n",
+      "Epoch 93/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 5.0533e-04 - mean_absolute_error: 0.0164 - val_loss: 4.6021e-04 - val_mean_absolute_error: 0.0155\n",
+      "Epoch 94/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 5.1851e-04 - mean_absolute_error: 0.0165 - val_loss: 3.7934e-04 - val_mean_absolute_error: 0.0140\n",
+      "Epoch 95/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 4.9983e-04 - mean_absolute_error: 0.0162 - val_loss: 5.0210e-04 - val_mean_absolute_error: 0.0166\n",
+      "Epoch 96/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 4.9377e-04 - mean_absolute_error: 0.0161 - val_loss: 5.1563e-04 - val_mean_absolute_error: 0.0168\n",
+      "Epoch 97/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 4.9085e-04 - mean_absolute_error: 0.0161 - val_loss: 4.1783e-04 - val_mean_absolute_error: 0.0149\n",
+      "Epoch 98/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 4.8465e-04 - mean_absolute_error: 0.0160 - val_loss: 3.9469e-04 - val_mean_absolute_error: 0.0150\n",
+      "Epoch 99/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 4.7694e-04 - mean_absolute_error: 0.0159 - val_loss: 4.5126e-04 - val_mean_absolute_error: 0.0158\n",
+      "Epoch 100/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 4.6634e-04 - mean_absolute_error: 0.0157 - val_loss: 3.4697e-04 - val_mean_absolute_error: 0.0134\n",
+      "Epoch 1/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 0.0187 - mean_absolute_error: 0.1104 - val_loss: 0.0158 - val_mean_absolute_error: 0.1039\n",
+      "Epoch 2/100\n",
+      "18750/18750 [==============================] - 7s 354us/step - loss: 0.0159 - mean_absolute_error: 0.1039 - val_loss: 0.0157 - val_mean_absolute_error: 0.1037\n",
+      "Epoch 3/100\n",
+      "18750/18750 [==============================] - 7s 358us/step - loss: 0.0156 - mean_absolute_error: 0.1029 - val_loss: 0.0153 - val_mean_absolute_error: 0.1019\n",
+      "Epoch 4/100\n",
+      "18750/18750 [==============================] - 7s 358us/step - loss: 0.0153 - mean_absolute_error: 0.1017 - val_loss: 0.0152 - val_mean_absolute_error: 0.0996\n",
+      "Epoch 5/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 0.0138 - mean_absolute_error: 0.0933 - val_loss: 0.0121 - val_mean_absolute_error: 0.0833\n",
+      "Epoch 6/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 0.0119 - mean_absolute_error: 0.0840 - val_loss: 0.0106 - val_mean_absolute_error: 0.0779\n",
+      "Epoch 7/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 0.0112 - mean_absolute_error: 0.0810 - val_loss: 0.0120 - val_mean_absolute_error: 0.0845\n",
+      "Epoch 8/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 0.0105 - mean_absolute_error: 0.0781 - val_loss: 0.0098 - val_mean_absolute_error: 0.0754\n",
+      "Epoch 9/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 0.0102 - mean_absolute_error: 0.0769 - val_loss: 0.0100 - val_mean_absolute_error: 0.0737\n",
+      "Epoch 10/100\n",
+      "18750/18750 [==============================] - 7s 358us/step - loss: 0.0095 - mean_absolute_error: 0.0733 - val_loss: 0.0091 - val_mean_absolute_error: 0.0699\n",
+      "Epoch 11/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 0.0089 - mean_absolute_error: 0.0708 - val_loss: 0.0082 - val_mean_absolute_error: 0.0671\n",
+      "Epoch 12/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 0.0083 - mean_absolute_error: 0.0678 - val_loss: 0.0073 - val_mean_absolute_error: 0.0629\n",
+      "Epoch 13/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 0.0073 - mean_absolute_error: 0.0632 - val_loss: 0.0069 - val_mean_absolute_error: 0.0609\n",
+      "Epoch 14/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 0.0064 - mean_absolute_error: 0.0587 - val_loss: 0.0055 - val_mean_absolute_error: 0.0549\n",
+      "Epoch 15/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 0.0055 - mean_absolute_error: 0.0543 - val_loss: 0.0048 - val_mean_absolute_error: 0.0490\n",
+      "Epoch 16/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 0.0048 - mean_absolute_error: 0.0503 - val_loss: 0.0043 - val_mean_absolute_error: 0.0472\n",
+      "Epoch 17/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 0.0042 - mean_absolute_error: 0.0470 - val_loss: 0.0059 - val_mean_absolute_error: 0.0581\n",
+      "Epoch 18/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 0.0037 - mean_absolute_error: 0.0436 - val_loss: 0.0032 - val_mean_absolute_error: 0.0410\n",
+      "Epoch 19/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 0.0033 - mean_absolute_error: 0.0412 - val_loss: 0.0026 - val_mean_absolute_error: 0.0361\n",
+      "Epoch 20/100\n",
+      "18750/18750 [==============================] - 7s 370us/step - loss: 0.0030 - mean_absolute_error: 0.0392 - val_loss: 0.0023 - val_mean_absolute_error: 0.0338\n",
+      "Epoch 21/100\n",
+      "18750/18750 [==============================] - 7s 371us/step - loss: 0.0026 - mean_absolute_error: 0.0365 - val_loss: 0.0022 - val_mean_absolute_error: 0.0336\n",
+      "Epoch 22/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 0.0022 - mean_absolute_error: 0.0330 - val_loss: 0.0027 - val_mean_absolute_error: 0.0358\n",
+      "Epoch 23/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 0.0019 - mean_absolute_error: 0.0305 - val_loss: 0.0014 - val_mean_absolute_error: 0.0263\n",
+      "Epoch 24/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 0.0017 - mean_absolute_error: 0.0289 - val_loss: 0.0014 - val_mean_absolute_error: 0.0266\n",
+      "Epoch 25/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 0.0015 - mean_absolute_error: 0.0275 - val_loss: 0.0012 - val_mean_absolute_error: 0.0239\n",
+      "Epoch 26/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 0.0013 - mean_absolute_error: 0.0258 - val_loss: 0.0012 - val_mean_absolute_error: 0.0245\n",
+      "Epoch 27/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 0.0012 - mean_absolute_error: 0.0248 - val_loss: 0.0011 - val_mean_absolute_error: 0.0231\n",
+      "Epoch 28/100\n",
+      "18750/18750 [==============================] - 7s 358us/step - loss: 0.0011 - mean_absolute_error: 0.0242 - val_loss: 9.8805e-04 - val_mean_absolute_error: 0.0228\n",
+      "Epoch 29/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 0.0011 - mean_absolute_error: 0.0235 - val_loss: 0.0011 - val_mean_absolute_error: 0.0248\n",
+      "Epoch 30/100\n",
+      "18750/18750 [==============================] - 7s 367us/step - loss: 0.0010 - mean_absolute_error: 0.0230 - val_loss: 9.6145e-04 - val_mean_absolute_error: 0.0224\n",
+      "Epoch 31/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 9.8571e-04 - mean_absolute_error: 0.0225 - val_loss: 7.5892e-04 - val_mean_absolute_error: 0.0197\n",
+      "Epoch 32/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 9.3009e-04 - mean_absolute_error: 0.0219 - val_loss: 7.9565e-04 - val_mean_absolute_error: 0.0201\n",
+      "Epoch 33/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 8.9681e-04 - mean_absolute_error: 0.0214 - val_loss: 5.8689e-04 - val_mean_absolute_error: 0.0170\n",
+      "Epoch 34/100\n",
+      "18750/18750 [==============================] - 7s 365us/step - loss: 8.6196e-04 - mean_absolute_error: 0.0210 - val_loss: 0.0012 - val_mean_absolute_error: 0.0256\n",
+      "Epoch 35/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 8.3124e-04 - mean_absolute_error: 0.0206 - val_loss: 7.9152e-04 - val_mean_absolute_error: 0.0203\n",
+      "Epoch 36/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 7.8601e-04 - mean_absolute_error: 0.0200 - val_loss: 0.0010 - val_mean_absolute_error: 0.0224\n",
+      "Epoch 37/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 7.7720e-04 - mean_absolute_error: 0.0198 - val_loss: 7.0561e-04 - val_mean_absolute_error: 0.0186\n",
+      "Epoch 38/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 7.6506e-04 - mean_absolute_error: 0.0196 - val_loss: 7.0692e-04 - val_mean_absolute_error: 0.0192\n",
+      "Epoch 39/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 7.4454e-04 - mean_absolute_error: 0.0194 - val_loss: 9.0127e-04 - val_mean_absolute_error: 0.0218\n",
+      "Epoch 40/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 7.2479e-04 - mean_absolute_error: 0.0192 - val_loss: 6.1556e-04 - val_mean_absolute_error: 0.0173\n",
+      "Epoch 41/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 7.1619e-04 - mean_absolute_error: 0.0191 - val_loss: 7.6910e-04 - val_mean_absolute_error: 0.0195\n",
+      "Epoch 42/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 7.0075e-04 - mean_absolute_error: 0.0188 - val_loss: 6.8010e-04 - val_mean_absolute_error: 0.0183\n",
+      "Epoch 43/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 6.8589e-04 - mean_absolute_error: 0.0186 - val_loss: 5.3766e-04 - val_mean_absolute_error: 0.0160\n",
+      "Epoch 44/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 6.5634e-04 - mean_absolute_error: 0.0183 - val_loss: 7.2940e-04 - val_mean_absolute_error: 0.0201\n",
+      "Epoch 45/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 6.5187e-04 - mean_absolute_error: 0.0182 - val_loss: 9.1681e-04 - val_mean_absolute_error: 0.0227\n",
+      "Epoch 46/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 6.2411e-04 - mean_absolute_error: 0.0179 - val_loss: 4.9152e-04 - val_mean_absolute_error: 0.0154\n",
+      "Epoch 47/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 6.2071e-04 - mean_absolute_error: 0.0178 - val_loss: 5.1954e-04 - val_mean_absolute_error: 0.0161\n",
+      "Epoch 48/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 6.1338e-04 - mean_absolute_error: 0.0177 - val_loss: 4.8405e-04 - val_mean_absolute_error: 0.0152\n",
+      "Epoch 49/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 6.2495e-04 - mean_absolute_error: 0.0178 - val_loss: 4.5350e-04 - val_mean_absolute_error: 0.0148\n",
+      "Epoch 50/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 6.1242e-04 - mean_absolute_error: 0.0176 - val_loss: 5.4714e-04 - val_mean_absolute_error: 0.0166\n",
+      "Epoch 51/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 6.1050e-04 - mean_absolute_error: 0.0177 - val_loss: 4.5383e-04 - val_mean_absolute_error: 0.0157\n",
+      "Epoch 52/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 5.9639e-04 - mean_absolute_error: 0.0175 - val_loss: 6.8893e-04 - val_mean_absolute_error: 0.0196\n",
+      "Epoch 53/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 5.9083e-04 - mean_absolute_error: 0.0174 - val_loss: 5.0113e-04 - val_mean_absolute_error: 0.0158\n",
+      "Epoch 54/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 5.6995e-04 - mean_absolute_error: 0.0171 - val_loss: 9.8139e-04 - val_mean_absolute_error: 0.0233\n",
+      "Epoch 55/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 5.9813e-04 - mean_absolute_error: 0.0173 - val_loss: 8.9144e-04 - val_mean_absolute_error: 0.0222\n",
+      "Epoch 56/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 5.8197e-04 - mean_absolute_error: 0.0172 - val_loss: 4.6464e-04 - val_mean_absolute_error: 0.0154\n",
+      "Epoch 57/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 5.7610e-04 - mean_absolute_error: 0.0171 - val_loss: 4.4228e-04 - val_mean_absolute_error: 0.0150\n",
+      "Epoch 58/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 5.5040e-04 - mean_absolute_error: 0.0168 - val_loss: 6.0541e-04 - val_mean_absolute_error: 0.0172\n",
+      "Epoch 59/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 5.5517e-04 - mean_absolute_error: 0.0169 - val_loss: 5.9842e-04 - val_mean_absolute_error: 0.0175\n",
+      "Epoch 60/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 5.4632e-04 - mean_absolute_error: 0.0168 - val_loss: 4.8324e-04 - val_mean_absolute_error: 0.0156\n",
+      "Epoch 61/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 5.5144e-04 - mean_absolute_error: 0.0168 - val_loss: 3.5748e-04 - val_mean_absolute_error: 0.0132\n",
+      "Epoch 62/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 5.5050e-04 - mean_absolute_error: 0.0167 - val_loss: 5.3119e-04 - val_mean_absolute_error: 0.0169\n",
+      "Epoch 63/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 5.3908e-04 - mean_absolute_error: 0.0167 - val_loss: 3.9886e-04 - val_mean_absolute_error: 0.0142\n",
+      "Epoch 64/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 5.3556e-04 - mean_absolute_error: 0.0166 - val_loss: 4.8528e-04 - val_mean_absolute_error: 0.0165\n",
+      "Epoch 65/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 5.3341e-04 - mean_absolute_error: 0.0165 - val_loss: 4.7665e-04 - val_mean_absolute_error: 0.0154\n",
+      "Epoch 66/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 5.3228e-04 - mean_absolute_error: 0.0165 - val_loss: 3.9131e-04 - val_mean_absolute_error: 0.0139\n",
+      "Epoch 67/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 5.2299e-04 - mean_absolute_error: 0.0164 - val_loss: 4.5984e-04 - val_mean_absolute_error: 0.0156\n",
+      "Epoch 68/100\n",
+      "18750/18750 [==============================] - 7s 367us/step - loss: 5.2837e-04 - mean_absolute_error: 0.0165 - val_loss: 6.8560e-04 - val_mean_absolute_error: 0.0192\n",
+      "Epoch 69/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 5.1798e-04 - mean_absolute_error: 0.0163 - val_loss: 4.3355e-04 - val_mean_absolute_error: 0.0153\n",
+      "Epoch 70/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 5.0453e-04 - mean_absolute_error: 0.0161 - val_loss: 5.4180e-04 - val_mean_absolute_error: 0.0166\n",
+      "Epoch 71/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 5.1300e-04 - mean_absolute_error: 0.0162 - val_loss: 5.4801e-04 - val_mean_absolute_error: 0.0166\n",
+      "Epoch 72/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 5.0951e-04 - mean_absolute_error: 0.0162 - val_loss: 4.7352e-04 - val_mean_absolute_error: 0.0154\n",
+      "Epoch 73/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 5.0311e-04 - mean_absolute_error: 0.0161 - val_loss: 4.2083e-04 - val_mean_absolute_error: 0.0144\n",
+      "Epoch 74/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 4.8638e-04 - mean_absolute_error: 0.0158 - val_loss: 4.3866e-04 - val_mean_absolute_error: 0.0149\n",
+      "Epoch 75/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 4.9500e-04 - mean_absolute_error: 0.0159 - val_loss: 3.5620e-04 - val_mean_absolute_error: 0.0135\n",
+      "Epoch 76/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 4.8254e-04 - mean_absolute_error: 0.0158 - val_loss: 5.0160e-04 - val_mean_absolute_error: 0.0162\n",
+      "Epoch 77/100\n",
+      "18750/18750 [==============================] - 7s 365us/step - loss: 4.7849e-04 - mean_absolute_error: 0.0157 - val_loss: 5.9712e-04 - val_mean_absolute_error: 0.0184\n",
+      "Epoch 78/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 4.8515e-04 - mean_absolute_error: 0.0157 - val_loss: 4.1748e-04 - val_mean_absolute_error: 0.0146\n",
+      "Epoch 79/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 4.7240e-04 - mean_absolute_error: 0.0156 - val_loss: 4.0813e-04 - val_mean_absolute_error: 0.0144\n",
+      "Epoch 80/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 4.6968e-04 - mean_absolute_error: 0.0155 - val_loss: 4.9191e-04 - val_mean_absolute_error: 0.0157\n",
+      "Epoch 81/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 4.8081e-04 - mean_absolute_error: 0.0157 - val_loss: 4.2521e-04 - val_mean_absolute_error: 0.0149\n",
+      "Epoch 82/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 4.7546e-04 - mean_absolute_error: 0.0156 - val_loss: 5.1604e-04 - val_mean_absolute_error: 0.0169\n",
+      "Epoch 83/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 4.7661e-04 - mean_absolute_error: 0.0156 - val_loss: 4.4765e-04 - val_mean_absolute_error: 0.0154\n",
+      "Epoch 84/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 4.7375e-04 - mean_absolute_error: 0.0156 - val_loss: 4.6632e-04 - val_mean_absolute_error: 0.0155\n",
+      "Epoch 85/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 4.6642e-04 - mean_absolute_error: 0.0155 - val_loss: 4.2406e-04 - val_mean_absolute_error: 0.0146\n",
+      "Epoch 86/100\n",
+      "18750/18750 [==============================] - 7s 365us/step - loss: 4.6717e-04 - mean_absolute_error: 0.0154 - val_loss: 3.9780e-04 - val_mean_absolute_error: 0.0143\n",
+      "Epoch 87/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 4.5896e-04 - mean_absolute_error: 0.0153 - val_loss: 4.2838e-04 - val_mean_absolute_error: 0.0155\n",
+      "Epoch 88/100\n",
+      "18750/18750 [==============================] - 7s 366us/step - loss: 4.5054e-04 - mean_absolute_error: 0.0152 - val_loss: 3.7960e-04 - val_mean_absolute_error: 0.0146\n",
+      "Epoch 89/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 4.5130e-04 - mean_absolute_error: 0.0152 - val_loss: 3.3870e-04 - val_mean_absolute_error: 0.0132\n",
+      "Epoch 90/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 4.4078e-04 - mean_absolute_error: 0.0150 - val_loss: 3.2388e-04 - val_mean_absolute_error: 0.0128\n",
+      "Epoch 91/100\n",
+      "18750/18750 [==============================] - 7s 365us/step - loss: 4.5106e-04 - mean_absolute_error: 0.0152 - val_loss: 4.2711e-04 - val_mean_absolute_error: 0.0150\n",
+      "Epoch 92/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 4.4019e-04 - mean_absolute_error: 0.0150 - val_loss: 3.7699e-04 - val_mean_absolute_error: 0.0133\n",
+      "Epoch 93/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 4.4261e-04 - mean_absolute_error: 0.0150 - val_loss: 3.7417e-04 - val_mean_absolute_error: 0.0136\n",
+      "Epoch 94/100\n",
+      "18750/18750 [==============================] - 7s 365us/step - loss: 4.3228e-04 - mean_absolute_error: 0.0149 - val_loss: 4.2668e-04 - val_mean_absolute_error: 0.0152\n",
+      "Epoch 95/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 4.3701e-04 - mean_absolute_error: 0.0149 - val_loss: 3.0990e-04 - val_mean_absolute_error: 0.0121\n",
+      "Epoch 96/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 4.2944e-04 - mean_absolute_error: 0.0148 - val_loss: 3.0912e-04 - val_mean_absolute_error: 0.0123\n",
+      "Epoch 97/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 4.2298e-04 - mean_absolute_error: 0.0147 - val_loss: 4.6810e-04 - val_mean_absolute_error: 0.0155\n",
+      "Epoch 98/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 4.2599e-04 - mean_absolute_error: 0.0148 - val_loss: 3.9567e-04 - val_mean_absolute_error: 0.0143\n",
+      "Epoch 99/100\n",
+      "18750/18750 [==============================] - 7s 366us/step - loss: 4.3458e-04 - mean_absolute_error: 0.0148 - val_loss: 7.5533e-04 - val_mean_absolute_error: 0.0217\n",
+      "Epoch 100/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 4.0902e-04 - mean_absolute_error: 0.0145 - val_loss: 3.5114e-04 - val_mean_absolute_error: 0.0136\n"
+     ]
+    }
+   ],
+   "source": [
+    "histories=np.zeros_like(models)\n",
+    "for i in range(len(models)):\n",
+    "    histories[i]=models[i].fit(X_train,y_train,\n",
+    "                               validation_data=(X_val,y_val),\n",
+    "                               batch_size=32,\n",
+    "                               epochs=100)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 35,
+   "id": "41f3f861-6d2a-4424-8cb5-94ace18364ea",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "model_mixed=keras.models.Sequential()\n",
+    "model_mixed.add(keras.layers.Dense(units=32, activation='relu', input_dim=X_train.shape[1], kernel_initializer=keras.initializers.HeUniform))\n",
+    "model_mixed.add(keras.layers.Dense(units=32, activation='sigmoid', kernel_initializer=keras.initializers.GlorotUniform))\n",
+    "model_mixed.add(keras.layers.Dense(units=64, activation='sigmoid', kernel_initializer=keras.initializers.GlorotUniform))\n",
+    "model_mixed.add(keras.layers.Dense(units=1, activation='relu', kernel_initializer=keras.initializers.HeUniform))\n",
+    "model_mixed.compile(optimizer='adam',\n",
+    "                    loss='mean_squared_error',\n",
+    "                    metrics=['mean_absolute_error'])"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 36,
+   "id": "0318ddd8-dfa4-4bfc-9d77-7c8b129a87e3",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Epoch 1/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 0.0163 - mean_absolute_error: 0.1050 - val_loss: 0.0160 - val_mean_absolute_error: 0.1058\n",
+      "Epoch 2/100\n",
+      "18750/18750 [==============================] - 7s 356us/step - loss: 0.0158 - mean_absolute_error: 0.1035 - val_loss: 0.0158 - val_mean_absolute_error: 0.1044\n",
+      "Epoch 3/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 0.0155 - mean_absolute_error: 0.1024 - val_loss: 0.0151 - val_mean_absolute_error: 0.1016\n",
+      "Epoch 4/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 0.0146 - mean_absolute_error: 0.0981 - val_loss: 0.0143 - val_mean_absolute_error: 0.0964\n",
+      "Epoch 5/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 0.0125 - mean_absolute_error: 0.0870 - val_loss: 0.0112 - val_mean_absolute_error: 0.0804\n",
+      "Epoch 6/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 0.0113 - mean_absolute_error: 0.0811 - val_loss: 0.0108 - val_mean_absolute_error: 0.0792\n",
+      "Epoch 7/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 0.0109 - mean_absolute_error: 0.0792 - val_loss: 0.0112 - val_mean_absolute_error: 0.0810\n",
+      "Epoch 8/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 0.0105 - mean_absolute_error: 0.0775 - val_loss: 0.0096 - val_mean_absolute_error: 0.0746\n",
+      "Epoch 9/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 0.0093 - mean_absolute_error: 0.0727 - val_loss: 0.0091 - val_mean_absolute_error: 0.0732\n",
+      "Epoch 10/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 0.0085 - mean_absolute_error: 0.0688 - val_loss: 0.0073 - val_mean_absolute_error: 0.0631\n",
+      "Epoch 11/100\n",
+      "18750/18750 [==============================] - 7s 371us/step - loss: 0.0074 - mean_absolute_error: 0.0638 - val_loss: 0.0066 - val_mean_absolute_error: 0.0608\n",
+      "Epoch 12/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 0.0062 - mean_absolute_error: 0.0574 - val_loss: 0.0060 - val_mean_absolute_error: 0.0559\n",
+      "Epoch 13/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 0.0055 - mean_absolute_error: 0.0533 - val_loss: 0.0049 - val_mean_absolute_error: 0.0504\n",
+      "Epoch 14/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 0.0048 - mean_absolute_error: 0.0490 - val_loss: 0.0036 - val_mean_absolute_error: 0.0422\n",
+      "Epoch 15/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 0.0038 - mean_absolute_error: 0.0439 - val_loss: 0.0041 - val_mean_absolute_error: 0.0471\n",
+      "Epoch 16/100\n",
+      "18750/18750 [==============================] - 7s 358us/step - loss: 0.0030 - mean_absolute_error: 0.0394 - val_loss: 0.0019 - val_mean_absolute_error: 0.0304\n",
+      "Epoch 17/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 0.0023 - mean_absolute_error: 0.0344 - val_loss: 0.0025 - val_mean_absolute_error: 0.0373\n",
+      "Epoch 18/100\n",
+      "18750/18750 [==============================] - 7s 358us/step - loss: 0.0020 - mean_absolute_error: 0.0319 - val_loss: 0.0015 - val_mean_absolute_error: 0.0274\n",
+      "Epoch 19/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 0.0018 - mean_absolute_error: 0.0298 - val_loss: 0.0015 - val_mean_absolute_error: 0.0287\n",
+      "Epoch 20/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 0.0016 - mean_absolute_error: 0.0284 - val_loss: 0.0013 - val_mean_absolute_error: 0.0254\n",
+      "Epoch 21/100\n",
+      "18750/18750 [==============================] - 7s 358us/step - loss: 0.0015 - mean_absolute_error: 0.0272 - val_loss: 0.0013 - val_mean_absolute_error: 0.0251\n",
+      "Epoch 22/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 0.0014 - mean_absolute_error: 0.0260 - val_loss: 0.0014 - val_mean_absolute_error: 0.0258\n",
+      "Epoch 23/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 0.0013 - mean_absolute_error: 0.0253 - val_loss: 0.0012 - val_mean_absolute_error: 0.0251\n",
+      "Epoch 24/100\n",
+      "18750/18750 [==============================] - 7s 358us/step - loss: 0.0012 - mean_absolute_error: 0.0244 - val_loss: 0.0012 - val_mean_absolute_error: 0.0234\n",
+      "Epoch 25/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 0.0011 - mean_absolute_error: 0.0238 - val_loss: 8.2946e-04 - val_mean_absolute_error: 0.0206\n",
+      "Epoch 26/100\n",
+      "18750/18750 [==============================] - 7s 358us/step - loss: 0.0010 - mean_absolute_error: 0.0228 - val_loss: 0.0015 - val_mean_absolute_error: 0.0282\n",
+      "Epoch 27/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 9.8726e-04 - mean_absolute_error: 0.0226 - val_loss: 7.8138e-04 - val_mean_absolute_error: 0.0197\n",
+      "Epoch 28/100\n",
+      "18750/18750 [==============================] - 7s 358us/step - loss: 9.6040e-04 - mean_absolute_error: 0.0223 - val_loss: 0.0011 - val_mean_absolute_error: 0.0235\n",
+      "Epoch 29/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 9.1230e-04 - mean_absolute_error: 0.0218 - val_loss: 9.0021e-04 - val_mean_absolute_error: 0.0223\n",
+      "Epoch 30/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 8.6924e-04 - mean_absolute_error: 0.0213 - val_loss: 8.4191e-04 - val_mean_absolute_error: 0.0212\n",
+      "Epoch 31/100\n",
+      "18750/18750 [==============================] - 7s 365us/step - loss: 8.4018e-04 - mean_absolute_error: 0.0209 - val_loss: 0.0019 - val_mean_absolute_error: 0.0321\n",
+      "Epoch 32/100\n",
+      "18750/18750 [==============================] - 7s 357us/step - loss: 8.4033e-04 - mean_absolute_error: 0.0209 - val_loss: 9.7924e-04 - val_mean_absolute_error: 0.0232\n",
+      "Epoch 33/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 8.0410e-04 - mean_absolute_error: 0.0205 - val_loss: 8.3061e-04 - val_mean_absolute_error: 0.0202\n",
+      "Epoch 34/100\n",
+      "18750/18750 [==============================] - 7s 358us/step - loss: 7.9325e-04 - mean_absolute_error: 0.0203 - val_loss: 7.1261e-04 - val_mean_absolute_error: 0.0200\n",
+      "Epoch 35/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 7.8602e-04 - mean_absolute_error: 0.0202 - val_loss: 5.9531e-04 - val_mean_absolute_error: 0.0173\n",
+      "Epoch 36/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 7.4934e-04 - mean_absolute_error: 0.0198 - val_loss: 5.9050e-04 - val_mean_absolute_error: 0.0178\n",
+      "Epoch 37/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 7.2060e-04 - mean_absolute_error: 0.0194 - val_loss: 5.5108e-04 - val_mean_absolute_error: 0.0169\n",
+      "Epoch 38/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 7.0998e-04 - mean_absolute_error: 0.0192 - val_loss: 5.6978e-04 - val_mean_absolute_error: 0.0176\n",
+      "Epoch 39/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 7.0327e-04 - mean_absolute_error: 0.0192 - val_loss: 6.3382e-04 - val_mean_absolute_error: 0.0191\n",
+      "Epoch 40/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 6.8717e-04 - mean_absolute_error: 0.0189 - val_loss: 7.6121e-04 - val_mean_absolute_error: 0.0206\n",
+      "Epoch 41/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 6.5767e-04 - mean_absolute_error: 0.0185 - val_loss: 9.7121e-04 - val_mean_absolute_error: 0.0239\n",
+      "Epoch 42/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 6.5005e-04 - mean_absolute_error: 0.0184 - val_loss: 6.4900e-04 - val_mean_absolute_error: 0.0183\n",
+      "Epoch 43/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 6.3856e-04 - mean_absolute_error: 0.0183 - val_loss: 6.3351e-04 - val_mean_absolute_error: 0.0182\n",
+      "Epoch 44/100\n",
+      "18750/18750 [==============================] - 7s 358us/step - loss: 6.1954e-04 - mean_absolute_error: 0.0180 - val_loss: 4.9793e-04 - val_mean_absolute_error: 0.0160\n",
+      "Epoch 45/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 6.1985e-04 - mean_absolute_error: 0.0180 - val_loss: 4.4391e-04 - val_mean_absolute_error: 0.0152\n",
+      "Epoch 46/100\n",
+      "18750/18750 [==============================] - 7s 358us/step - loss: 5.9981e-04 - mean_absolute_error: 0.0178 - val_loss: 4.7827e-04 - val_mean_absolute_error: 0.0157\n",
+      "Epoch 47/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 5.9940e-04 - mean_absolute_error: 0.0178 - val_loss: 6.7507e-04 - val_mean_absolute_error: 0.0188\n",
+      "Epoch 48/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 5.9299e-04 - mean_absolute_error: 0.0176 - val_loss: 5.9828e-04 - val_mean_absolute_error: 0.0181\n",
+      "Epoch 49/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 5.8850e-04 - mean_absolute_error: 0.0176 - val_loss: 4.2146e-04 - val_mean_absolute_error: 0.0148\n",
+      "Epoch 50/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 5.7291e-04 - mean_absolute_error: 0.0173 - val_loss: 5.5165e-04 - val_mean_absolute_error: 0.0170\n",
+      "Epoch 51/100\n",
+      "18750/18750 [==============================] - 7s 365us/step - loss: 5.7243e-04 - mean_absolute_error: 0.0173 - val_loss: 5.4992e-04 - val_mean_absolute_error: 0.0168\n",
+      "Epoch 52/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 5.8064e-04 - mean_absolute_error: 0.0174 - val_loss: 7.6522e-04 - val_mean_absolute_error: 0.0192\n",
+      "Epoch 53/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 5.5837e-04 - mean_absolute_error: 0.0171 - val_loss: 3.7729e-04 - val_mean_absolute_error: 0.0139\n",
+      "Epoch 54/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 5.4775e-04 - mean_absolute_error: 0.0169 - val_loss: 7.2287e-04 - val_mean_absolute_error: 0.0193\n",
+      "Epoch 55/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 5.4507e-04 - mean_absolute_error: 0.0169 - val_loss: 5.2354e-04 - val_mean_absolute_error: 0.0169\n",
+      "Epoch 56/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 5.5791e-04 - mean_absolute_error: 0.0170 - val_loss: 3.6747e-04 - val_mean_absolute_error: 0.0136\n",
+      "Epoch 57/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 5.4389e-04 - mean_absolute_error: 0.0169 - val_loss: 5.2052e-04 - val_mean_absolute_error: 0.0164\n",
+      "Epoch 58/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 5.4896e-04 - mean_absolute_error: 0.0169 - val_loss: 7.9571e-04 - val_mean_absolute_error: 0.0216\n",
+      "Epoch 59/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 5.2328e-04 - mean_absolute_error: 0.0166 - val_loss: 4.5743e-04 - val_mean_absolute_error: 0.0151\n",
+      "Epoch 60/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 5.3059e-04 - mean_absolute_error: 0.0166 - val_loss: 5.1968e-04 - val_mean_absolute_error: 0.0161\n",
+      "Epoch 61/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 5.0781e-04 - mean_absolute_error: 0.0163 - val_loss: 4.2339e-04 - val_mean_absolute_error: 0.0153\n",
+      "Epoch 62/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 5.0901e-04 - mean_absolute_error: 0.0163 - val_loss: 5.4574e-04 - val_mean_absolute_error: 0.0169\n",
+      "Epoch 63/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 5.0503e-04 - mean_absolute_error: 0.0163 - val_loss: 4.7407e-04 - val_mean_absolute_error: 0.0151\n",
+      "Epoch 64/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 5.1626e-04 - mean_absolute_error: 0.0164 - val_loss: 5.8969e-04 - val_mean_absolute_error: 0.0180\n",
+      "Epoch 65/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 5.1441e-04 - mean_absolute_error: 0.0164 - val_loss: 3.5121e-04 - val_mean_absolute_error: 0.0134\n",
+      "Epoch 66/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 5.1447e-04 - mean_absolute_error: 0.0163 - val_loss: 4.4777e-04 - val_mean_absolute_error: 0.0151\n",
+      "Epoch 67/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 4.9721e-04 - mean_absolute_error: 0.0161 - val_loss: 4.6706e-04 - val_mean_absolute_error: 0.0160\n",
+      "Epoch 68/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 4.7959e-04 - mean_absolute_error: 0.0159 - val_loss: 4.7260e-04 - val_mean_absolute_error: 0.0160\n",
+      "Epoch 69/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 4.9529e-04 - mean_absolute_error: 0.0161 - val_loss: 5.1170e-04 - val_mean_absolute_error: 0.0174\n",
+      "Epoch 70/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 4.8029e-04 - mean_absolute_error: 0.0158 - val_loss: 5.5782e-04 - val_mean_absolute_error: 0.0159\n",
+      "Epoch 71/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 4.7628e-04 - mean_absolute_error: 0.0158 - val_loss: 4.7913e-04 - val_mean_absolute_error: 0.0160\n",
+      "Epoch 72/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 4.8591e-04 - mean_absolute_error: 0.0159 - val_loss: 3.7890e-04 - val_mean_absolute_error: 0.0138\n",
+      "Epoch 73/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 4.8060e-04 - mean_absolute_error: 0.0158 - val_loss: 4.1332e-04 - val_mean_absolute_error: 0.0145\n",
+      "Epoch 74/100\n",
+      "18750/18750 [==============================] - 7s 359us/step - loss: 4.7238e-04 - mean_absolute_error: 0.0157 - val_loss: 5.1453e-04 - val_mean_absolute_error: 0.0166\n",
+      "Epoch 75/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 4.7576e-04 - mean_absolute_error: 0.0157 - val_loss: 3.6188e-04 - val_mean_absolute_error: 0.0133\n",
+      "Epoch 76/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 4.7006e-04 - mean_absolute_error: 0.0156 - val_loss: 4.0508e-04 - val_mean_absolute_error: 0.0146\n",
+      "Epoch 77/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 4.6711e-04 - mean_absolute_error: 0.0156 - val_loss: 4.3654e-04 - val_mean_absolute_error: 0.0154\n",
+      "Epoch 78/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 4.7930e-04 - mean_absolute_error: 0.0156 - val_loss: 5.3041e-04 - val_mean_absolute_error: 0.0165\n",
+      "Epoch 79/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 4.6607e-04 - mean_absolute_error: 0.0156 - val_loss: 3.9712e-04 - val_mean_absolute_error: 0.0145\n",
+      "Epoch 80/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 4.5676e-04 - mean_absolute_error: 0.0154 - val_loss: 3.7035e-04 - val_mean_absolute_error: 0.0140\n",
+      "Epoch 81/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 4.5534e-04 - mean_absolute_error: 0.0154 - val_loss: 6.7935e-04 - val_mean_absolute_error: 0.0175\n",
+      "Epoch 82/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 4.4209e-04 - mean_absolute_error: 0.0152 - val_loss: 3.0094e-04 - val_mean_absolute_error: 0.0124\n",
+      "Epoch 83/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 4.5072e-04 - mean_absolute_error: 0.0153 - val_loss: 4.9114e-04 - val_mean_absolute_error: 0.0163\n",
+      "Epoch 84/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 4.4901e-04 - mean_absolute_error: 0.0153 - val_loss: 3.8599e-04 - val_mean_absolute_error: 0.0140\n",
+      "Epoch 85/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 4.5136e-04 - mean_absolute_error: 0.0152 - val_loss: 3.8554e-04 - val_mean_absolute_error: 0.0141\n",
+      "Epoch 86/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 4.3593e-04 - mean_absolute_error: 0.0151 - val_loss: 3.7972e-04 - val_mean_absolute_error: 0.0135\n",
+      "Epoch 87/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 4.4672e-04 - mean_absolute_error: 0.0151 - val_loss: 4.8179e-04 - val_mean_absolute_error: 0.0155\n",
+      "Epoch 88/100\n",
+      "18750/18750 [==============================] - 7s 364us/step - loss: 4.3549e-04 - mean_absolute_error: 0.0150 - val_loss: 4.6216e-04 - val_mean_absolute_error: 0.0158\n",
+      "Epoch 89/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 4.3476e-04 - mean_absolute_error: 0.0150 - val_loss: 3.6614e-04 - val_mean_absolute_error: 0.0136\n",
+      "Epoch 90/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 4.3906e-04 - mean_absolute_error: 0.0150 - val_loss: 4.9844e-04 - val_mean_absolute_error: 0.0164\n",
+      "Epoch 91/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 4.2275e-04 - mean_absolute_error: 0.0148 - val_loss: 4.0900e-04 - val_mean_absolute_error: 0.0140\n",
+      "Epoch 92/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 4.2736e-04 - mean_absolute_error: 0.0149 - val_loss: 4.3093e-04 - val_mean_absolute_error: 0.0158\n",
+      "Epoch 93/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 4.1034e-04 - mean_absolute_error: 0.0146 - val_loss: 3.6711e-04 - val_mean_absolute_error: 0.0134\n",
+      "Epoch 94/100\n",
+      "18750/18750 [==============================] - 7s 362us/step - loss: 4.1896e-04 - mean_absolute_error: 0.0148 - val_loss: 5.1838e-04 - val_mean_absolute_error: 0.0161\n",
+      "Epoch 95/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 4.1208e-04 - mean_absolute_error: 0.0147 - val_loss: 3.3842e-04 - val_mean_absolute_error: 0.0137\n",
+      "Epoch 96/100\n",
+      "18750/18750 [==============================] - 7s 361us/step - loss: 4.0929e-04 - mean_absolute_error: 0.0146 - val_loss: 4.1079e-04 - val_mean_absolute_error: 0.0151\n",
+      "Epoch 97/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 4.0987e-04 - mean_absolute_error: 0.0146 - val_loss: 4.6050e-04 - val_mean_absolute_error: 0.0149\n",
+      "Epoch 98/100\n",
+      "18750/18750 [==============================] - 7s 363us/step - loss: 4.0293e-04 - mean_absolute_error: 0.0144 - val_loss: 3.4880e-04 - val_mean_absolute_error: 0.0137\n",
+      "Epoch 99/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 3.9736e-04 - mean_absolute_error: 0.0143 - val_loss: 4.7751e-04 - val_mean_absolute_error: 0.0152\n",
+      "Epoch 100/100\n",
+      "18750/18750 [==============================] - 7s 360us/step - loss: 4.0610e-04 - mean_absolute_error: 0.0145 - val_loss: 4.6150e-04 - val_mean_absolute_error: 0.0150\n"
+     ]
+    }
+   ],
+   "source": [
+    "hist_mixed=model_mixed.fit(X_train,y_train,\n",
+    "                           validation_data=(X_val,y_val),\n",
+    "                           batch_size=32,\n",
+    "                           epochs=100)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 37,
+   "id": "10235cc1-e181-4a89-a135-0785488a8684",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "image/png": "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",
+      "text/plain": [
+       "<Figure size 640x480 with 1 Axes>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    }
+   ],
+   "source": [
+    "colors=['blue','orange','green','olive','purple','gray','red']\n",
+    "legend=['RandomNormal','RandomUniform','GlorotNormal','GlorotUniform','HeNormal','HeUniform','Glorot+He']\n",
+    "for i in range(len(my_initializers)):\n",
+    "    plt.plot(histories[i].history['loss'],color=colors[i])\n",
+    "    plt.yscale('log')\n",
+    "plt.plot(hist_mixed.history['loss'],color=colors[-1])\n",
+    "plt.title('Model loss on the training set \\n for different weights initializers')\n",
+    "plt.xlabel('epoch')\n",
+    "plt.ylabel('mean squared error')\n",
+    "plt.legend(legend,loc='upper right')\n",
+    "plt.show()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 38,
+   "id": "74f43d53-00e5-4968-9198-3b2b316a9015",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "image/png": "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",
+      "text/plain": [
+       "<Figure size 640x480 with 1 Axes>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    }
+   ],
+   "source": [
+    "colors=['blue','orange','green','olive','purple','gray','red']\n",
+    "legend=['RandomNormal','RandomUniform','GlorotNormal','GlorotUniform','HeNormal','HeUniform','Glorot+He']\n",
+    "for i in range(len(my_initializers)):\n",
+    "    plt.plot(histories[i].history['val_loss'],color=colors[i])\n",
+    "plt.plot(hist_mixed.history['val_loss'], color=colors[-1])\n",
+    "plt.title('Model loss on the validation set \\n for different weights initializers')\n",
+    "plt.xlabel('epoch')\n",
+    "plt.ylabel('mean squared error')\n",
+    "plt.legend(legend,loc='upper right')\n",
+    "plt.show()"
+   ]
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3 (ipykernel)",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.9.19"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}