PaperPlots_Gaussian.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import matplotlib as mpl\n",
    "\n",
    "from matplotlib import gridspec\n",
    "from scipy.stats import norm\n",
    "import energyflow as ef #needed for the Geant4 datasets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "import tensorflow as tf\n",
    "from keras.layers import Dense, Input\n",
    "from keras.models import Model\n",
    "import tensorflow.keras.backend as K\n",
    "\n",
    "import os\n",
    "# os.environ[\"CUDA_DEVICE_ORDER\"]=\"PCI_BUS_ID\"\n",
    "# os.environ['CUDA_VISIBLE_DEVICES']=\"3\"\n",
    "# gpus = tf.config.experimental.list_physical_devices('GPU')\n",
    "# tf.config.experimental.set_virtual_device_configuration(gpus[0], [tf.config.experimental.VirtualDeviceConfiguration(memory_limit=24220)]) #in MB\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Config"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "N = 20000\n",
    "sigma = 1\n",
    "mu = 0\n",
    "epsilon = 2\n",
    "\n",
    "\n",
    "# colors\n",
    "n_lines = 5\n",
    "c = np.arange(-3, n_lines + 1)\n",
    "\n",
    "norm = mpl.colors.Normalize(vmin=c.min(), vmax=c.max())\n",
    "cmap = mpl.cm.ScalarMappable(norm=norm, cmap=\"Blues\")\n",
    "cmap.set_array([])\n",
    "\n",
    "\n",
    "\n",
    "MSE_color = cmap.to_rgba(3)\n",
    "MSE_darkcolor = cmap.to_rgba(4)\n",
    "MLC_color = \"red\"\n",
    "MLC_darkcolor = \"darkred\"\n",
    "data_color = \"black\"\n",
    "capsize = 2"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "\n",
    "gauss_truth_prior = np.random.normal(mu,sigma,N)\n",
    "gauss_reco_prior = np.random.normal(gauss_truth_prior,epsilon,N)\n",
    "mycalibration = np.polyfit(gauss_reco_prior, gauss_truth_prior, 1)\n",
    "mypoly = np.poly1d(mycalibration)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig = plt.figure(figsize=(6, 6))\n",
    "gs = gridspec.GridSpec(1, 1) \n",
    "plt.subplots_adjust(hspace = .001)\n",
    "\n",
    "ax0 = plt.subplot(gs[0])\n",
    "ax0.yaxis.set_ticks_position('both')\n",
    "ax0.xaxis.set_ticks_position('both')\n",
    "ax0.tick_params(direction=\"in\",which=\"both\")\n",
    "plt.xticks(fontsize=20)\n",
    "plt.yticks(fontsize=20)\n",
    "ax0.minorticks_on()\n",
    "\n",
    "# plt.scatter(gauss_reco_prior,gauss_truth_prior,label=\"Data\",alpha=0.05, color = data_color)\n",
    "plt.hist2d(gauss_reco_prior,gauss_truth_prior,label=\"Data\", density = True, bins = (50, 50), range = [[-5, 5], [-5, 5]], cmap = \"binary\")\n",
    "plt.plot(np.linspace(-5,5,20),mypoly(np.linspace(-5,5,20)),label=\"MSE Fit\",ls=\"--\",color=\"blue\",lw=4)\n",
    "\n",
    "plt.xlim([-5,5])\n",
    "plt.ylim([-5,5])\n",
    "plt.xlabel(\"Measured $X_D$\",fontsize=20)\n",
    "plt.ylabel(\"True $Z_T$\",fontsize=20)\n",
    "leg = plt.legend(frameon=False,fontsize=15,ncol=1, loc = \"upper left\")\n",
    "for lh in leg.legendHandles: \n",
    "    lh.set_alpha(1)\n",
    "\n",
    "_=plt.title(r\"Simulation-Based Gaussian Example\",loc=\"right\",fontsize=18, fontstyle=\"italic\")\n",
    "_=plt.text(-0.5,4,r\"$(\\mu,\\sigma,\\epsilon)=(0,1,2)$\",fontsize=20)\n",
    "plt.savefig(\"Gaussian_prior.pdf\",bbox_inches='tight')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# GA Fit"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Start of epoch 0\n",
      "WARNING:tensorflow:Layer g_ifn_8 is casting an input tensor from dtype float64 to the layer's dtype of float32, which is new behavior in TensorFlow 2.  The layer has dtype float32 because it's dtype defaults to floatx.\n",
      "\n",
      "If you intended to run this layer in float32, you can safely ignore this warning. If in doubt, this warning is likely only an issue if you are porting a TensorFlow 1.X model to TensorFlow 2.\n",
      "\n",
      "To change all layers to have dtype float64 by default, call `tf.keras.backend.set_floatx('float64')`. To change just this layer, pass dtype='float64' to the layer constructor. If you are the author of this layer, you can disable autocasting by passing autocast=False to the base Layer constructor.\n",
      "\n",
      "WARNING:tensorflow:Layer dnn_33 is casting an input tensor from dtype float64 to the layer's dtype of float32, which is new behavior in TensorFlow 2.  The layer has dtype float32 because it's dtype defaults to floatx.\n",
      "\n",
      "If you intended to run this layer in float32, you can safely ignore this warning. If in doubt, this warning is likely only an issue if you are porting a TensorFlow 1.X model to TensorFlow 2.\n",
      "\n",
      "To change all layers to have dtype float64 by default, call `tf.keras.backend.set_floatx('float64')`. To change just this layer, pass dtype='float64' to the layer constructor. If you are the author of this layer, you can disable autocasting by passing autocast=False to the base Layer constructor.\n",
      "\n",
      "WARNING:tensorflow:Layer dnn_32 is casting an input tensor from dtype float64 to the layer's dtype of float32, which is new behavior in TensorFlow 2.  The layer has dtype float32 because it's dtype defaults to floatx.\n",
      "\n",
      "If you intended to run this layer in float32, you can safely ignore this warning. If in doubt, this warning is likely only an issue if you are porting a TensorFlow 1.X model to TensorFlow 2.\n",
      "\n",
      "To change all layers to have dtype float64 by default, call `tf.keras.backend.set_floatx('float64')`. To change just this layer, pass dtype='float64' to the layer constructor. If you are the author of this layer, you can disable autocasting by passing autocast=False to the base Layer constructor.\n",
      "\n",
      "WARNING:tensorflow:Layer dnn_34 is casting an input tensor from dtype float64 to the layer's dtype of float32, which is new behavior in TensorFlow 2.  The layer has dtype float32 because it's dtype defaults to floatx.\n",
      "\n",
      "If you intended to run this layer in float32, you can safely ignore this warning. If in doubt, this warning is likely only an issue if you are porting a TensorFlow 1.X model to TensorFlow 2.\n",
      "\n",
      "To change all layers to have dtype float64 by default, call `tf.keras.backend.set_floatx('float64')`. To change just this layer, pass dtype='float64' to the layer constructor. If you are the author of this layer, you can disable autocasting by passing autocast=False to the base Layer constructor.\n",
      "\n",
      "Pre-Train Epoch 0, Batch 0: Loss = 4380.181, a = 22.655 b = 45.272, c = -0.624, Joint = -29.796, Marginal = -32.639, MI = -23.027\n",
      "Pre-Train Epoch 0, Batch 1: Loss = 46859.465, a = 24.731 b = 52.861, c = -0.623, Joint = -57.065, Marginal = -67.283, MI = -50.406\n",
      "Pre-Train Epoch 0, Batch 2: Loss = 1434.648, a = 20.472 b = 50.440, c = -0.622, Joint = -24.476, Marginal = -27.721, MI = -18.502\n",
      "Pre-Train Epoch 0, Batch 3: Loss = 2480.652, a = 16.280 b = 42.560, c = -0.621, Joint = -24.384, Marginal = -30.000, MI = -18.044\n",
      "Pre-Train Epoch 0, Batch 4: Loss = 1162.633, a = 14.183 b = 33.633, c = -0.620, Joint = -19.515, Marginal = -25.247, MI = -13.098\n",
      "Pre-Train Epoch 0, Batch 5: Loss = 2145.364, a = 15.213 b = 40.505, c = -0.619, Joint = -26.915, Marginal = -34.049, MI = -21.395\n",
      "Pre-Train Epoch 0, Batch 6: Loss = 1803.715, a = 15.979 b = 43.171, c = -3.358, Joint = -23.331, Marginal = -27.827, MI = -17.748\n",
      "Pre-Train Epoch 0, Batch 7: Loss = 863.577, a = 13.178 b = 49.026, c = -0.618, Joint = -20.383, Marginal = -23.947, MI = -14.521\n",
      "Pre-Train Epoch 0, Batch 8: Loss = 1084.933, a = 13.297 b = 48.334, c = -0.617, Joint = -21.901, Marginal = -25.680, MI = -16.420\n",
      "Pre-Train Epoch 0, Batch 9: Loss = 14552.093, a = 15.259 b = 43.376, c = -0.616, Joint = -49.255, Marginal = -57.794, MI = -43.712\n",
      "Pre-Train Epoch 0, Batch 10: Loss = 2664.960, a = 13.214 b = 54.855, c = -8.813, Joint = -27.700, Marginal = -40.123, MI = -21.982\n",
      "Pre-Train Epoch 0, Batch 11: Loss = 884.787, a = 10.414 b = 41.497, c = -0.615, Joint = -20.408, Marginal = -26.335, MI = -14.138\n",
      "Pre-Train Epoch 0, Batch 12: Loss = 5290.576, a = 10.658 b = 32.666, c = -0.614, Joint = -29.787, Marginal = -32.474, MI = -24.355\n",
      "Pre-Train Epoch 0, Batch 13: Loss = 1702.642, a = 11.155 b = 39.734, c = -0.613, Joint = -24.016, Marginal = -26.437, MI = -18.611\n",
      "Pre-Train Epoch 0, Batch 14: Loss = 15948.728, a = 8.536 b = 34.322, c = -3.843, Joint = -30.956, Marginal = -35.197, MI = -25.352\n",
      "Pre-Train Epoch 0, Batch 15: Loss = 5624.464, a = 9.039 b = 42.806, c = -0.612, Joint = -31.134, Marginal = -31.440, MI = -25.246\n",
      "Pre-Train Epoch 0, Batch 16: Loss = 15351.374, a = 8.719 b = 49.369, c = -0.611, Joint = -40.201, Marginal = -35.246, MI = -35.011\n",
      "Pre-Train Epoch 0, Batch 17: Loss = 14517.901, a = 7.492 b = 48.134, c = -1.736, Joint = -35.585, Marginal = -43.899, MI = -30.171\n",
      "Pre-Train Epoch 0, Batch 18: Loss = 2806.689, a = 7.124 b = 35.277, c = -0.609, Joint = -26.260, Marginal = -37.499, MI = -20.035\n",
      "Pre-Train Epoch 0, Batch 19: Loss = 2086.120, a = 7.784 b = 38.403, c = -2.850, Joint = -22.463, Marginal = -33.927, MI = -16.857\n",
      "Pre-Train Epoch 0, Batch 20: Loss = 486.268, a = 5.315 b = 35.433, c = -0.607, Joint = -15.191, Marginal = -20.426, MI = -9.992\n",
      "Pre-Train Epoch 0, Batch 21: Loss = 3952.928, a = 6.740 b = 41.822, c = -0.607, Joint = -31.342, Marginal = -46.938, MI = -26.327\n",
      "Pre-Train Epoch 0, Batch 22: Loss = 870.661, a = 5.828 b = 33.771, c = -7.956, Joint = -18.719, Marginal = -22.727, MI = -13.767\n",
      "Pre-Train Epoch 0, Batch 23: Loss = 3776.035, a = 5.986 b = 39.828, c = -0.605, Joint = -28.743, Marginal = -32.962, MI = -23.992\n",
      "Pre-Train Epoch 0, Batch 24: Loss = 647.710, a = 5.108 b = 33.916, c = -0.604, Joint = -17.982, Marginal = -27.375, MI = -12.745\n",
      "Pre-Train Epoch 0, Batch 25: Loss = 12954.416, a = 5.253 b = 46.337, c = -0.603, Joint = -36.423, Marginal = -37.216, MI = -31.807\n",
      "Pre-Train Epoch 0, Batch 26: Loss = 2552.240, a = 5.041 b = 53.975, c = -0.603, Joint = -29.284, Marginal = -29.164, MI = -24.496\n",
      "Pre-Train Epoch 0, Batch 27: Loss = 2809.745, a = 4.186 b = 51.980, c = -0.602, Joint = -28.962, Marginal = -42.586, MI = -24.010\n",
      "Pre-Train Epoch 0, Batch 28: Loss = 3968.867, a = 4.463 b = 33.774, c = -2.754, Joint = -26.597, Marginal = -38.942, MI = -21.570\n",
      "Pre-Train Epoch 0, Batch 29: Loss = 1051.454, a = 3.579 b = 46.540, c = -0.600, Joint = -19.567, Marginal = -19.264, MI = -15.015\n",
      "Pre-Train Epoch 0, Batch 30: Loss = 4058.831, a = 3.263 b = 54.658, c = -0.599, Joint = -31.830, Marginal = -35.220, MI = -26.612\n",
      "Pre-Train Epoch 0, Batch 31: Loss = 8634.295, a = 3.249 b = 50.081, c = -13.786, Joint = -38.131, Marginal = -43.510, MI = -32.875\n",
      "Pre-Train Epoch 0, Batch 32: Loss = 1602.031, a = 2.773 b = 42.024, c = -0.598, Joint = -19.431, Marginal = -33.242, MI = -14.800\n",
      "Pre-Train Epoch 0, Batch 33: Loss = 15226.530, a = 2.934 b = 38.094, c = -0.597, Joint = -34.975, Marginal = -49.066, MI = -29.894\n",
      "Pre-Train Epoch 0, Batch 34: Loss = 635.293, a = 2.942 b = 44.238, c = -7.655, Joint = -16.988, Marginal = -23.061, MI = -12.722\n",
      "Pre-Train Epoch 0, Batch 35: Loss = 1669.119, a = 2.627 b = 47.440, c = -0.595, Joint = -24.180, Marginal = -26.985, MI = -19.799\n",
      "Pre-Train Epoch 0, Batch 36: Loss = 726.702, a = 2.383 b = 45.842, c = -4.698, Joint = -18.155, Marginal = -20.937, MI = -13.020\n",
      "Pre-Train Epoch 0, Batch 37: Loss = 2284.847, a = 2.354 b = 51.859, c = -0.593, Joint = -26.480, Marginal = -40.421, MI = -22.153\n",
      "Pre-Train Epoch 0, Batch 38: Loss = 470.322, a = 2.248 b = 37.957, c = -1.102, Joint = -14.636, Marginal = -19.867, MI = -10.440\n",
      "Pre-Train Epoch 0, Batch 39: Loss = 993.189, a = 2.115 b = 49.775, c = -1.297, Joint = -20.456, Marginal = -24.263, MI = -16.084\n",
      "Pre-Train Epoch 0, Batch 40: Loss = 145989.500, a = 2.797 b = 38.847, c = -0.591, Joint = -78.766, Marginal = -135.434, MI = -74.405\n",
      "Pre-Train Epoch 0, Batch 41: Loss = 1801.969, a = 2.550 b = 47.652, c = -0.590, Joint = -23.910, Marginal = -40.793, MI = -19.816\n",
      "Pre-Train Epoch 0, Batch 42: Loss = 902.881, a = 2.009 b = 45.557, c = -1.156, Joint = -18.869, Marginal = -25.165, MI = -14.578\n",
      "Pre-Train Epoch 0, Batch 43: Loss = 3006.188, a = 2.580 b = 50.001, c = -8.966, Joint = -27.386, Marginal = -31.508, MI = -22.590\n",
      "Pre-Train Epoch 0, Batch 44: Loss = 3619.139, a = 2.729 b = 44.374, c = -0.587, Joint = -26.763, Marginal = -44.643, MI = -22.708\n",
      "Pre-Train Epoch 0, Batch 45: Loss = 1356.649, a = 2.335 b = 43.063, c = -3.845, Joint = -20.707, Marginal = -24.079, MI = -16.345\n",
      "Pre-Train Epoch 0, Batch 46: Loss = 1504.158, a = 2.409 b = 48.493, c = -0.586, Joint = -23.921, Marginal = -34.202, MI = -19.896\n",
      "Pre-Train Epoch 0, Batch 47: Loss = 33832.508, a = 2.253 b = 37.417, c = -0.585, Joint = -42.031, Marginal = -54.412, MI = -36.599\n",
      "Pre-Train Epoch 0, Batch 48: Loss = 5925.305, a = 2.609 b = 44.695, c = -0.584, Joint = -36.195, Marginal = -55.140, MI = -31.926\n",
      "Pre-Train Epoch 0, Batch 49: Loss = 843.324, a = 1.872 b = 41.408, c = -0.583, Joint = -18.123, Marginal = -22.005, MI = -13.273\n",
      "Pre-Train Epoch 0, Batch 50: Loss = 3325.798, a = 2.828 b = 35.042, c = -0.582, Joint = -24.362, Marginal = -37.783, MI = -20.555\n",
      "Pre-Train Epoch 0, Batch 51: Loss = 3276.026, a = 2.890 b = 43.400, c = -2.494, Joint = -27.721, Marginal = -38.907, MI = -23.227\n",
      "Pre-Train Epoch 0, Batch 52: Loss = 654.878, a = 2.691 b = 36.177, c = -1.500, Joint = -15.481, Marginal = -21.490, MI = -11.823\n",
      "Pre-Train Epoch 0, Batch 53: Loss = 1754.366, a = 2.245 b = 29.631, c = -4.519, Joint = -20.647, Marginal = -25.169, MI = -17.054\n",
      "Pre-Train Epoch 0, Batch 54: Loss = 1558.694, a = 3.391 b = 47.726, c = -0.579, Joint = -20.071, Marginal = -25.782, MI = -15.831\n",
      "Pre-Train Epoch 0, Batch 55: Loss = 90569.266, a = 3.255 b = 46.123, c = -0.579, Joint = -58.480, Marginal = -61.581, MI = -54.327\n",
      "Pre-Train Epoch 0, Batch 56: Loss = 2821.430, a = 3.019 b = 43.493, c = -3.986, Joint = -23.747, Marginal = -25.474, MI = -19.931\n",
      "Pre-Train Epoch 0, Batch 57: Loss = 681.005, a = 2.779 b = 40.683, c = -8.429, Joint = -16.120, Marginal = -20.269, MI = -12.081\n",
      "Pre-Train Epoch 0, Batch 58: Loss = 1977.192, a = 4.156 b = 48.071, c = -5.777, Joint = -21.601, Marginal = -22.463, MI = -17.476\n",
      "Pre-Train Epoch 0, Batch 59: Loss = 657.900, a = 3.457 b = 37.386, c = -1.746, Joint = -14.298, Marginal = -17.941, MI = -10.998\n",
      "Pre-Train Epoch 0, Batch 60: Loss = 41592.082, a = 2.635 b = 41.208, c = -0.575, Joint = -46.123, Marginal = -68.930, MI = -42.330\n",
      "Pre-Train Epoch 0, Batch 61: Loss = 1263.967, a = 3.737 b = 41.041, c = -0.574, Joint = -20.051, Marginal = -27.855, MI = -16.600\n",
      "Pre-Train Epoch 0, Batch 62: Loss = 8414.641, a = 3.455 b = 38.883, c = -7.240, Joint = -32.985, Marginal = -38.036, MI = -29.450\n",
      "Pre-Train Epoch 0, Batch 63: Loss = 4044.647, a = 3.116 b = 36.249, c = -0.572, Joint = -24.292, Marginal = -42.644, MI = -20.870\n",
      "Pre-Train Epoch 0, Batch 64: Loss = 880.099, a = 4.696 b = 47.992, c = -0.571, Joint = -17.113, Marginal = -19.202, MI = -13.883\n",
      "Pre-Train Epoch 0, Batch 65: Loss = 18863.955, a = 2.706 b = 36.354, c = -2.329, Joint = -44.702, Marginal = -60.858, MI = -41.031\n",
      "Pre-Train Epoch 0, Batch 66: Loss = 973.167, a = 5.665 b = 51.013, c = -0.570, Joint = -18.924, Marginal = -25.994, MI = -15.434\n",
      "Pre-Train Epoch 0, Batch 67: Loss = 1882.163, a = 9.081 b = 58.504, c = -0.569, Joint = -22.920, Marginal = -40.456, MI = -19.655\n",
      "Pre-Train Epoch 0, Batch 68: Loss = 2909.573, a = 5.361 b = 38.037, c = -0.568, Joint = -21.546, Marginal = -34.395, MI = -17.811\n",
      "Pre-Train Epoch 0, Batch 69: Loss = 3799.015, a = 4.151 b = 39.133, c = -0.568, Joint = -25.188, Marginal = -41.637, MI = -21.434\n",
      "Pre-Train Epoch 0, Batch 70: Loss = 4518.474, a = 6.517 b = 49.789, c = -3.879, Joint = -23.838, Marginal = -26.962, MI = -20.242\n",
      "Pre-Train Epoch 0, Batch 71: Loss = 17775.400, a = 8.684 b = 60.323, c = -0.566, Joint = -39.834, Marginal = -48.591, MI = -36.328\n",
      "Pre-Train Epoch 0, Batch 72: Loss = 7378.311, a = 3.662 b = 35.683, c = -1.032, Joint = -33.517, Marginal = -51.873, MI = -30.271\n",
      "Pre-Train Epoch 0, Batch 73: Loss = 5430.236, a = 5.632 b = 32.005, c = -0.565, Joint = -23.481, Marginal = -34.826, MI = -20.194\n",
      "Pre-Train Epoch 0, Batch 74: Loss = 1241.649, a = 6.397 b = 41.402, c = -1.371, Joint = -20.737, Marginal = -37.393, MI = -17.508\n",
      "Pre-Train Epoch 0, Batch 75: Loss = 1427.955, a = 4.474 b = 34.723, c = -0.564, Joint = -18.812, Marginal = -24.722, MI = -15.701\n",
      "Pre-Train Epoch 0, Batch 76: Loss = 3274.747, a = 5.229 b = 41.095, c = -0.563, Joint = -21.699, Marginal = -31.961, MI = -18.571\n",
      "Pre-Train Epoch 0, Batch 77: Loss = 11923.917, a = 5.733 b = 40.941, c = -0.562, Joint = -34.425, Marginal = -40.596, MI = -30.724\n",
      "Pre-Train Epoch 0, Batch 78: Loss = 1313.838, a = 8.791 b = 49.749, c = -0.562, Joint = -17.950, Marginal = -20.919, MI = -14.585\n",
      "Pre-Train Epoch 0, Batch 79: Loss = 4446.805, a = 7.622 b = 39.946, c = -0.561, Joint = -23.350, Marginal = -33.779, MI = -20.852\n",
      "Pre-Train Epoch 0, Batch 80: Loss = 3041.917, a = 9.446 b = 47.352, c = -0.560, Joint = -21.890, Marginal = -27.369, MI = -18.530\n",
      "Pre-Train Epoch 0, Batch 81: Loss = 2469.723, a = 6.632 b = 41.176, c = -7.773, Joint = -21.407, Marginal = -29.890, MI = -18.361\n",
      "Pre-Train Epoch 0, Batch 82: Loss = 9592.649, a = 10.658 b = 45.584, c = -7.128, Joint = -27.593, Marginal = -51.362, MI = -24.884\n",
      "Pre-Train Epoch 0, Batch 83: Loss = 1064.364, a = 8.426 b = 39.408, c = -0.559, Joint = -16.235, Marginal = -33.435, MI = -13.277\n",
      "Pre-Train Epoch 0, Batch 84: Loss = 1955.860, a = 7.624 b = 37.521, c = -0.558, Joint = -18.985, Marginal = -31.437, MI = -16.040\n",
      "Pre-Train Epoch 0, Batch 85: Loss = 14605.358, a = 7.387 b = 40.953, c = -0.557, Joint = -34.111, Marginal = -44.205, MI = -31.353\n",
      "Pre-Train Epoch 0, Batch 86: Loss = 786.089, a = 8.546 b = 39.873, c = -0.557, Joint = -16.527, Marginal = -22.076, MI = -13.476\n",
      "Pre-Train Epoch 0, Batch 87: Loss = 3957.076, a = 6.640 b = 34.399, c = -0.556, Joint = -18.080, Marginal = -25.034, MI = -15.868\n",
      "Pre-Train Epoch 0, Batch 88: Loss = 456.220, a = 10.332 b = 43.086, c = -0.556, Joint = -12.812, Marginal = -14.190, MI = -10.423\n",
      "Pre-Train Epoch 0, Batch 89: Loss = 435.695, a = 11.825 b = 41.695, c = -2.862, Joint = -13.227, Marginal = -18.450, MI = -10.370\n",
      "Pre-Train Epoch 0, Batch 90: Loss = 2867.936, a = 11.471 b = 40.593, c = -0.554, Joint = -22.996, Marginal = -35.752, MI = -20.544\n",
      "Pre-Train Epoch 0, Batch 91: Loss = 569.962, a = 12.360 b = 41.502, c = -0.554, Joint = -13.062, Marginal = -16.745, MI = -10.219\n",
      "Pre-Train Epoch 0, Batch 92: Loss = 3621.874, a = 5.865 b = 33.080, c = -0.553, Joint = -27.793, Marginal = -49.100, MI = -25.194\n",
      "Pre-Train Epoch 0, Batch 93: Loss = 8985.057, a = 17.064 b = 54.674, c = -0.553, Joint = -34.552, Marginal = -39.126, MI = -31.678\n",
      "Pre-Train Epoch 0, Batch 94: Loss = 417.524, a = 10.436 b = 36.420, c = -0.552, Joint = -12.702, Marginal = -20.646, MI = -10.176\n",
      "Pre-Train Epoch 0, Batch 95: Loss = 6147.381, a = 16.788 b = 49.631, c = -0.552, Joint = -29.017, Marginal = -35.928, MI = -27.073\n",
      "Pre-Train Epoch 0, Batch 96: Loss = 1935.898, a = 14.617 b = 48.663, c = -0.551, Joint = -19.197, Marginal = -20.455, MI = -16.549\n",
      "Pre-Train Epoch 0, Batch 97: Loss = 585.592, a = 13.148 b = 42.940, c = -0.551, Joint = -15.108, Marginal = -15.194, MI = -12.940\n",
      "Pre-Train Epoch 0, Batch 98: Loss = 3535.571, a = 19.897 b = 46.497, c = -9.083, Joint = -24.436, Marginal = -42.533, MI = -22.485\n",
      "Pre-Train Epoch 0, Batch 99: Loss = 3091.487, a = 20.584 b = 46.640, c = -1.417, Joint = -21.837, Marginal = -25.279, MI = -20.048\n",
      "Pre-Train Epoch 0, Batch 100: Loss = 324.741, a = 15.481 b = 41.797, c = -0.549, Joint = -10.875, Marginal = -14.081, MI = -8.929\n",
      "Pre-Train Epoch 0, Batch 101: Loss = 8636.089, a = 18.292 b = 46.049, c = -7.703, Joint = -37.658, Marginal = -69.280, MI = -35.538\n",
      "Pre-Train Epoch 0, Batch 102: Loss = 2914.500, a = 19.425 b = 41.424, c = -0.548, Joint = -20.316, Marginal = -24.808, MI = -18.422\n",
      "Pre-Train Epoch 0, Batch 103: Loss = 719.461, a = 17.574 b = 42.016, c = -0.547, Joint = -13.842, Marginal = -19.682, MI = -11.156\n",
      "Pre-Train Epoch 0, Batch 104: Loss = 5310.905, a = 13.919 b = 40.379, c = -0.547, Joint = -26.433, Marginal = -32.855, MI = -24.402\n",
      "Pre-Train Epoch 0, Batch 105: Loss = 3515.372, a = 13.479 b = 37.898, c = -0.546, Joint = -23.351, Marginal = -24.208, MI = -20.814\n",
      "Pre-Train Epoch 0, Batch 106: Loss = 5278.566, a = 19.396 b = 44.494, c = -0.546, Joint = -28.327, Marginal = -60.417, MI = -26.352\n",
      "Pre-Train Epoch 0, Batch 107: Loss = 885.080, a = 20.504 b = 41.424, c = -4.556, Joint = -14.023, Marginal = -20.027, MI = -12.260\n",
      "Pre-Train Epoch 0, Batch 108: Loss = 6031.356, a = 22.962 b = 48.823, c = -0.545, Joint = -20.647, Marginal = -22.100, MI = -17.471\n",
      "Pre-Train Epoch 0, Batch 109: Loss = 16924.945, a = 12.269 b = 36.481, c = -2.948, Joint = -28.334, Marginal = -37.422, MI = -25.936\n",
      "Pre-Train Epoch 0, Batch 110: Loss = 13354.787, a = 20.029 b = 43.306, c = -1.179, Joint = -28.139, Marginal = -40.330, MI = -26.223\n",
      "Pre-Train Epoch 0, Batch 111: Loss = 25902.475, a = 21.816 b = 50.255, c = -0.543, Joint = -46.961, Marginal = -45.937, MI = -43.870\n",
      "Pre-Train Epoch 0, Batch 112: Loss = 2403.606, a = 18.881 b = 36.377, c = -1.228, Joint = -17.600, Marginal = -25.189, MI = -15.863\n",
      "Pre-Train Epoch 0, Batch 113: Loss = 2211.315, a = 30.673 b = 49.406, c = -0.541, Joint = -22.015, Marginal = -33.229, MI = -19.779\n",
      "Pre-Train Epoch 0, Batch 114: Loss = 933.727, a = 18.805 b = 39.568, c = -0.541, Joint = -13.375, Marginal = -13.819, MI = -12.038\n",
      "Pre-Train Epoch 0, Batch 115: Loss = 11770.699, a = 19.150 b = 38.190, c = -0.540, Joint = -26.280, Marginal = -38.897, MI = -24.594\n",
      "Pre-Train Epoch 0, Batch 116: Loss = 421.485, a = 21.981 b = 45.594, c = -0.539, Joint = -12.435, Marginal = -12.196, MI = -10.008\n",
      "Pre-Train Epoch 0, Batch 117: Loss = 2740.467, a = 28.847 b = 42.695, c = -0.539, Joint = -20.880, Marginal = -19.114, MI = -18.789\n",
      "Pre-Train Epoch 0, Batch 118: Loss = 1000.329, a = 33.034 b = 49.541, c = -0.538, Joint = -14.358, Marginal = -17.106, MI = -13.077\n",
      "Pre-Train Epoch 0, Batch 119: Loss = 1148.494, a = 26.127 b = 40.092, c = -0.538, Joint = -17.172, Marginal = -25.063, MI = -16.388\n",
      "Pre-Train Epoch 0, Batch 120: Loss = 293.067, a = 14.377 b = 33.554, c = -8.827, Joint = -10.260, Marginal = -11.961, MI = -8.432\n",
      "Pre-Train Epoch 0, Batch 121: Loss = 299.011, a = 20.988 b = 33.844, c = -0.802, Joint = -9.407, Marginal = -9.864, MI = -8.742\n",
      "Pre-Train Epoch 0, Batch 122: Loss = 690.946, a = 22.906 b = 39.057, c = -7.704, Joint = -14.700, Marginal = -16.705, MI = -13.410\n",
      "Pre-Train Epoch 0, Batch 123: Loss = 467.348, a = 39.667 b = 42.455, c = -2.398, Joint = -11.352, Marginal = -19.203, MI = -10.474\n",
      "Pre-Train Epoch 0, Batch 124: Loss = 338.762, a = 21.154 b = 32.954, c = -0.535, Joint = -10.360, Marginal = -17.713, MI = -9.721\n",
      "Pre-Train Epoch 0, Batch 125: Loss = 8206.272, a = 23.560 b = 34.127, c = -0.755, Joint = -27.263, Marginal = -45.095, MI = -26.800\n",
      "Pre-Train Epoch 0, Batch 126: Loss = 264.208, a = 28.071 b = 39.720, c = -0.534, Joint = -9.292, Marginal = -12.148, MI = -8.801\n",
      "Pre-Train Epoch 0, Batch 127: Loss = 732.191, a = 41.754 b = 49.374, c = -0.534, Joint = -13.702, Marginal = -20.972, MI = -12.906\n",
      "Pre-Train Epoch 0, Batch 128: Loss = 2063.682, a = 23.704 b = 32.218, c = -1.690, Joint = -13.339, Marginal = -16.677, MI = -12.475\n",
      "Pre-Train Epoch 0, Batch 129: Loss = 1002.009, a = 33.500 b = 40.653, c = -0.533, Joint = -15.055, Marginal = -20.486, MI = -14.522\n",
      "Pre-Train Epoch 0, Batch 130: Loss = 496.569, a = 38.128 b = 47.492, c = -0.532, Joint = -12.633, Marginal = -14.588, MI = -11.444\n",
      "Pre-Train Epoch 0, Batch 131: Loss = 874.056, a = 30.155 b = 39.556, c = -4.617, Joint = -11.482, Marginal = -13.617, MI = -11.382\n",
      "Pre-Train Epoch 0, Batch 132: Loss = 2535.067, a = 25.773 b = 37.593, c = -0.531, Joint = -13.720, Marginal = -14.784, MI = -13.095\n",
      "Pre-Train Epoch 0, Batch 133: Loss = 15917.601, a = 40.153 b = 45.925, c = -0.531, Joint = -30.826, Marginal = -49.227, MI = -30.131\n",
      "Pre-Train Epoch 0, Batch 134: Loss = 2111.405, a = 18.639 b = 28.849, c = -0.535, Joint = -14.127, Marginal = -25.156, MI = -14.127\n",
      "Pre-Train Epoch 0, Batch 135: Loss = 308.176, a = 31.647 b = 36.841, c = -0.530, Joint = -9.339, Marginal = -16.304, MI = -8.434\n",
      "Pre-Train Epoch 0, Batch 136: Loss = 699.926, a = 63.631 b = 54.958, c = -0.530, Joint = -12.758, Marginal = -17.002, MI = -12.406\n",
      "Pre-Train Epoch 0, Batch 137: Loss = 450.069, a = 38.910 b = 37.254, c = -0.529, Joint = -9.957, Marginal = -17.677, MI = -9.816\n",
      "Pre-Train Epoch 0, Batch 138: Loss = 1657.748, a = 32.938 b = 37.212, c = -0.529, Joint = -13.571, Marginal = -15.009, MI = -13.172\n",
      "Pre-Train Epoch 0, Batch 139: Loss = 1223.786, a = 55.635 b = 51.844, c = -3.435, Joint = -15.870, Marginal = -20.848, MI = -15.038\n",
      "Pre-Train Epoch 0, Batch 140: Loss = 22562.125, a = 45.520 b = 45.591, c = -4.670, Joint = -30.938, Marginal = -35.294, MI = -31.200\n",
      "Pre-Train Epoch 0, Batch 141: Loss = 788.965, a = 62.299 b = 44.574, c = -0.527, Joint = -13.000, Marginal = -33.041, MI = -12.056\n",
      "Pre-Train Epoch 0, Batch 142: Loss = 1931.840, a = 38.937 b = 36.137, c = -0.527, Joint = -13.949, Marginal = -26.300, MI = -13.867\n",
      "Pre-Train Epoch 0, Batch 143: Loss = 7655.973, a = 67.074 b = 49.308, c = -0.526, Joint = -20.314, Marginal = -26.932, MI = -20.514\n",
      "Pre-Train Epoch 0, Batch 144: Loss = 1201.539, a = 54.232 b = 46.907, c = -0.526, Joint = -13.136, Marginal = -17.655, MI = -13.839\n",
      "Pre-Train Epoch 0, Batch 145: Loss = 1115.390, a = 50.076 b = 39.351, c = -0.525, Joint = -13.915, Marginal = -21.828, MI = -13.571\n",
      "Pre-Train Epoch 0, Batch 146: Loss = 452.855, a = 56.377 b = 42.224, c = -0.525, Joint = -10.327, Marginal = -15.075, MI = -10.629\n",
      "Pre-Train Epoch 0, Batch 147: Loss = 1342.154, a = 91.624 b = 50.636, c = -0.524, Joint = -14.026, Marginal = -28.010, MI = -14.941\n",
      "Pre-Train Epoch 0, Batch 148: Loss = 1286.879, a = 56.803 b = 39.557, c = -0.524, Joint = -13.717, Marginal = -16.221, MI = -14.725\n",
      "Pre-Train Epoch 0, Batch 149: Loss = 264.429, a = 24.257 b = 27.115, c = -5.761, Joint = -7.076, Marginal = -11.576, MI = -6.868\n",
      "Pre-Train Epoch 0, Batch 150: Loss = 1096.649, a = 58.574 b = 41.667, c = -6.992, Joint = -12.015, Marginal = -16.999, MI = -12.502\n",
      "\n",
      "Start of epoch 1\n",
      "Pre-Train Epoch 1, Batch 0: Loss = 1835.928, a = 66.961 b = 42.044, c = -0.522, Joint = -13.198, Marginal = -12.011, MI = -14.515\n",
      "Pre-Train Epoch 1, Batch 1: Loss = 23530.305, a = 80.668 b = 48.620, c = -0.522, Joint = -32.020, Marginal = -50.578, MI = -33.117\n",
      "Pre-Train Epoch 1, Batch 2: Loss = 443.546, a = 74.502 b = 47.171, c = -0.521, Joint = -8.763, Marginal = -10.961, MI = -9.644\n",
      "Pre-Train Epoch 1, Batch 3: Loss = 978.657, a = 56.425 b = 40.219, c = -0.521, Joint = -10.160, Marginal = -15.068, MI = -10.586\n",
      "Pre-Train Epoch 1, Batch 4: Loss = 388.143, a = 40.429 b = 32.052, c = -0.520, Joint = -6.988, Marginal = -9.034, MI = -9.134\n",
      "Pre-Train Epoch 1, Batch 5: Loss = 702.640, a = 54.321 b = 38.121, c = -0.520, Joint = -11.485, Marginal = -15.603, MI = -11.310\n",
      "Pre-Train Epoch 1, Batch 6: Loss = 607.354, a = 70.705 b = 40.462, c = -2.675, Joint = -8.374, Marginal = -13.863, MI = -10.310\n",
      "Pre-Train Epoch 1, Batch 7: Loss = 248.503, a = 69.511 b = 46.389, c = -0.519, Joint = -7.147, Marginal = -9.504, MI = -8.483\n",
      "Pre-Train Epoch 1, Batch 8: Loss = 358.945, a = 75.489 b = 45.494, c = -0.518, Joint = -7.463, Marginal = -11.160, MI = -8.621\n",
      "Pre-Train Epoch 1, Batch 9: Loss = 6414.479, a = 73.014 b = 40.004, c = -0.518, Joint = -26.583, Marginal = -44.758, MI = -27.437\n",
      "Pre-Train Epoch 1, Batch 10: Loss = 940.564, a = 100.395 b = 51.123, c = -7.938, Joint = -10.638, Marginal = -17.943, MI = -11.272\n",
      "Pre-Train Epoch 1, Batch 11: Loss = 257.690, a = 62.583 b = 39.293, c = -0.517, Joint = -6.921, Marginal = -13.895, MI = -8.414\n",
      "Pre-Train Epoch 1, Batch 12: Loss = 2199.070, a = 46.143 b = 30.884, c = -0.517, Joint = -14.314, Marginal = -19.928, MI = -16.347\n",
      "Pre-Train Epoch 1, Batch 13: Loss = 584.491, a = 77.695 b = 37.090, c = -0.516, Joint = -8.771, Marginal = -9.888, MI = -10.314\n",
      "Pre-Train Epoch 1, Batch 14: Loss = 8029.577, a = 48.237 b = 32.669, c = -3.410, Joint = -15.960, Marginal = -18.996, MI = -17.230\n",
      "Pre-Train Epoch 1, Batch 15: Loss = 2699.038, a = 64.923 b = 40.403, c = -0.515, Joint = -15.034, Marginal = -11.978, MI = -17.309\n",
      "Pre-Train Epoch 1, Batch 16: Loss = 8105.228, a = 85.952 b = 46.324, c = -0.515, Joint = -21.381, Marginal = -22.865, MI = -22.422\n",
      "Pre-Train Epoch 1, Batch 17: Loss = 7060.229, a = 77.147 b = 45.608, c = -1.537, Joint = -18.579, Marginal = -24.368, MI = -20.934\n",
      "Pre-Train Epoch 1, Batch 18: Loss = 1049.729, a = 52.549 b = 33.512, c = -0.514, Joint = -11.919, Marginal = -18.942, MI = -12.478\n",
      "Pre-Train Epoch 1, Batch 19: Loss = 794.979, a = 90.668 b = 35.923, c = -2.512, Joint = -8.037, Marginal = -16.640, MI = -10.834\n",
      "Pre-Train Epoch 1, Batch 20: Loss = 155.599, a = 55.581 b = 34.015, c = -0.513, Joint = -4.711, Marginal = -8.484, MI = -5.546\n",
      "Pre-Train Epoch 1, Batch 21: Loss = 1551.114, a = 85.539 b = 39.102, c = -0.513, Joint = -14.430, Marginal = -29.356, MI = -14.796\n",
      "Pre-Train Epoch 1, Batch 22: Loss = 263.360, a = 62.496 b = 32.169, c = -7.241, Joint = -6.356, Marginal = -14.336, MI = -8.938\n",
      "Pre-Train Epoch 1, Batch 23: Loss = 1633.587, a = 78.283 b = 37.565, c = -0.512, Joint = -12.877, Marginal = -19.026, MI = -15.107\n",
      "Pre-Train Epoch 1, Batch 24: Loss = 197.697, a = 59.968 b = 32.383, c = -0.512, Joint = -5.552, Marginal = -12.853, MI = -8.206\n",
      "Pre-Train Epoch 1, Batch 25: Loss = 6855.320, a = 99.732 b = 43.388, c = -0.511, Joint = -18.456, Marginal = -19.923, MI = -21.591\n",
      "Pre-Train Epoch 1, Batch 26: Loss = 957.285, a = 133.312 b = 50.267, c = -0.511, Joint = -12.006, Marginal = -15.590, MI = -14.400\n",
      "Pre-Train Epoch 1, Batch 27: Loss = 1082.555, a = 115.356 b = 48.829, c = -0.511, Joint = -12.270, Marginal = -16.369, MI = -15.394\n",
      "Pre-Train Epoch 1, Batch 28: Loss = 1812.409, a = 65.920 b = 31.894, c = -2.666, Joint = -11.786, Marginal = -15.165, MI = -16.224\n",
      "Pre-Train Epoch 1, Batch 29: Loss = 387.928, a = 103.862 b = 44.098, c = -0.510, Joint = -6.792, Marginal = -8.134, MI = -9.512\n",
      "Pre-Train Epoch 1, Batch 30: Loss = 1776.825, a = 114.521 b = 51.588, c = -0.510, Joint = -14.835, Marginal = -16.564, MI = -17.318\n",
      "Pre-Train Epoch 1, Batch 31: Loss = 4044.767, a = 113.907 b = 47.060, c = -12.802, Joint = -20.269, Marginal = -35.329, MI = -22.967\n",
      "Pre-Train Epoch 1, Batch 32: Loss = 630.771, a = 85.401 b = 40.018, c = -0.509, Joint = -6.866, Marginal = -11.356, MI = -9.417\n",
      "Pre-Train Epoch 1, Batch 33: Loss = 7977.952, a = 73.443 b = 36.144, c = -0.508, Joint = -18.855, Marginal = -31.510, MI = -21.743\n",
      "Pre-Train Epoch 1, Batch 34: Loss = 249.395, a = 113.219 b = 41.735, c = -7.053, Joint = -3.568, Marginal = -5.828, MI = -7.761\n",
      "Pre-Train Epoch 1, Batch 35: Loss = 634.697, a = 109.101 b = 44.826, c = -0.508, Joint = -9.451, Marginal = -11.971, MI = -12.690\n",
      "Pre-Train Epoch 1, Batch 36: Loss = 272.007, a = 93.915 b = 43.761, c = -4.244, Joint = -6.104, Marginal = -9.604, MI = -9.311\n",
      "Pre-Train Epoch 1, Batch 37: Loss = 899.109, a = 112.391 b = 49.131, c = -0.507, Joint = -11.499, Marginal = -15.477, MI = -16.344\n",
      "Pre-Train Epoch 1, Batch 38: Loss = 189.058, a = 72.188 b = 36.678, c = -0.797, Joint = -4.338, Marginal = -5.924, MI = -8.164\n",
      "Pre-Train Epoch 1, Batch 39: Loss = 391.051, a = 116.314 b = 47.273, c = -0.978, Joint = -6.585, Marginal = -9.201, MI = -10.269\n",
      "Pre-Train Epoch 1, Batch 40: Loss = 85839.734, a = 111.925 b = 36.100, c = -0.506, Joint = -52.443, Marginal = -109.834, MI = -56.670\n",
      "Pre-Train Epoch 1, Batch 41: Loss = 638.511, a = 131.388 b = 44.904, c = -0.505, Joint = -8.775, Marginal = -24.639, MI = -12.925\n",
      "Pre-Train Epoch 1, Batch 42: Loss = 337.082, a = 99.088 b = 43.406, c = -0.935, Joint = -6.328, Marginal = -10.255, MI = -9.443\n",
      "Pre-Train Epoch 1, Batch 43: Loss = 1221.837, a = 124.367 b = 47.189, c = -8.272, Joint = -12.324, Marginal = -22.136, MI = -16.781\n",
      "Pre-Train Epoch 1, Batch 44: Loss = 1522.517, a = 119.324 b = 41.842, c = -0.503, Joint = -11.586, Marginal = -26.155, MI = -16.866\n",
      "Pre-Train Epoch 1, Batch 45: Loss = 580.865, a = 102.173 b = 40.934, c = -3.518, Joint = -7.545, Marginal = -9.631, MI = -11.989\n",
      "Pre-Train Epoch 1, Batch 46: Loss = 559.601, a = 122.445 b = 45.878, c = -0.502, Joint = -9.064, Marginal = -17.555, MI = -15.500\n",
      "Pre-Train Epoch 1, Batch 47: Loss = 19818.959, a = 93.492 b = 35.430, c = -0.501, Joint = -24.594, Marginal = -35.722, MI = -29.835\n",
      "Pre-Train Epoch 1, Batch 48: Loss = 2545.094, a = 132.009 b = 41.831, c = -0.501, Joint = -17.298, Marginal = -30.793, MI = -24.166\n",
      "Pre-Train Epoch 1, Batch 49: Loss = 367.080, a = 93.316 b = 39.606, c = -0.500, Joint = -5.604, Marginal = -13.041, MI = -12.849\n",
      "Pre-Train Epoch 1, Batch 50: Loss = 1452.714, a = 102.756 b = 33.260, c = -0.500, Joint = -10.766, Marginal = -24.484, MI = -15.062\n",
      "Pre-Train Epoch 1, Batch 51: Loss = 1367.020, a = 128.096 b = 40.862, c = -2.275, Joint = -11.746, Marginal = -16.946, MI = -17.830\n",
      "Pre-Train Epoch 1, Batch 52: Loss = 269.587, a = 94.068 b = 34.545, c = -1.265, Joint = -4.060, Marginal = -7.093, MI = -9.526\n",
      "Pre-Train Epoch 1, Batch 53: Loss = 732.494, a = 91.933 b = 28.117, c = -4.085, Joint = -7.123, Marginal = -9.192, MI = -13.807\n",
      "Pre-Train Epoch 1, Batch 54: Loss = 681.144, a = 117.548 b = 45.545, c = -0.498, Joint = -7.556, Marginal = -10.402, MI = -11.743\n",
      "Pre-Train Epoch 1, Batch 55: Loss = 52948.449, a = 132.655 b = 43.383, c = -0.497, Joint = -36.735, Marginal = -64.689, MI = -42.563\n",
      "Pre-Train Epoch 1, Batch 56: Loss = 1321.537, a = 111.379 b = 41.386, c = -3.618, Joint = -9.757, Marginal = -12.362, MI = -15.889\n",
      "Pre-Train Epoch 1, Batch 57: Loss = 292.604, a = 102.749 b = 38.933, c = -7.753, Joint = -4.112, Marginal = -8.556, MI = -10.726\n",
      "Pre-Train Epoch 1, Batch 58: Loss = 912.385, a = 133.746 b = 45.505, c = -5.397, Joint = -7.641, Marginal = -19.924, MI = -13.609\n",
      "Pre-Train Epoch 1, Batch 59: Loss = 326.601, a = 104.258 b = 35.755, c = -1.427, Joint = -3.035, Marginal = -5.624, MI = -10.071\n",
      "Pre-Train Epoch 1, Batch 60: Loss = 23181.518, a = 108.501 b = 39.151, c = -0.494, Joint = -27.459, Marginal = -41.172, MI = -34.892\n",
      "Pre-Train Epoch 1, Batch 61: Loss = 526.750, a = 123.783 b = 38.850, c = -0.494, Joint = -6.327, Marginal = -10.099, MI = -13.804\n",
      "Pre-Train Epoch 1, Batch 62: Loss = 4165.830, a = 123.638 b = 36.814, c = -6.691, Joint = -16.491, Marginal = -20.649, MI = -21.945\n",
      "Pre-Train Epoch 1, Batch 63: Loss = 1792.512, a = 108.175 b = 34.452, c = -0.493, Joint = -10.173, Marginal = -26.277, MI = -17.014\n",
      "Pre-Train Epoch 1, Batch 64: Loss = 390.373, a = 121.752 b = 45.735, c = -0.492, Joint = -4.921, Marginal = -8.521, MI = -12.350\n",
      "Pre-Train Epoch 1, Batch 65: Loss = 10090.764, a = 131.043 b = 34.032, c = -2.055, Joint = -24.877, Marginal = -45.472, MI = -32.725\n",
      "Pre-Train Epoch 1, Batch 66: Loss = 420.268, a = 129.717 b = 48.596, c = -0.491, Joint = -6.206, Marginal = -10.510, MI = -16.864\n",
      "Pre-Train Epoch 1, Batch 67: Loss = 765.552, a = 182.102 b = 54.931, c = -0.491, Joint = -8.056, Marginal = -23.558, MI = -14.093\n",
      "Pre-Train Epoch 1, Batch 68: Loss = 1262.058, a = 142.964 b = 35.869, c = -0.490, Joint = -6.887, Marginal = -27.584, MI = -16.918\n",
      "Pre-Train Epoch 1, Batch 69: Loss = 1724.490, a = 115.222 b = 37.232, c = -0.490, Joint = -11.049, Marginal = -14.348, MI = -20.480\n",
      "Pre-Train Epoch 1, Batch 70: Loss = 2376.188, a = 129.750 b = 47.414, c = -3.576, Joint = -10.449, Marginal = -11.938, MI = -20.646\n",
      "Pre-Train Epoch 1, Batch 71: Loss = 9306.073, a = 177.477 b = 56.761, c = -0.489, Joint = -20.975, Marginal = -50.210, MI = -27.418\n",
      "Pre-Train Epoch 1, Batch 72: Loss = 3426.431, a = 140.427 b = 33.607, c = -0.636, Joint = -15.873, Marginal = -35.390, MI = -26.276\n",
      "Pre-Train Epoch 1, Batch 73: Loss = 2544.269, a = 125.987 b = 30.320, c = -0.488, Joint = -9.373, Marginal = -19.582, MI = -16.460\n",
      "Pre-Train Epoch 1, Batch 74: Loss = 461.821, a = 131.342 b = 39.296, c = -1.191, Joint = -7.141, Marginal = -14.921, MI = -13.311\n",
      "Pre-Train Epoch 1, Batch 75: Loss = 639.233, a = 122.163 b = 33.132, c = -0.487, Joint = -5.551, Marginal = -10.713, MI = -13.139\n",
      "Pre-Train Epoch 1, Batch 76: Loss = 1620.810, a = 113.068 b = 39.322, c = -0.486, Joint = -8.996, Marginal = -8.745, MI = -17.135\n",
      "Pre-Train Epoch 1, Batch 77: Loss = 5968.797, a = 126.422 b = 38.891, c = -0.486, Joint = -18.326, Marginal = -33.978, MI = -22.957\n",
      "Pre-Train Epoch 1, Batch 78: Loss = 608.952, a = 110.969 b = 47.631, c = -0.486, Joint = -7.145, Marginal = -9.283, MI = -13.835\n",
      "Pre-Train Epoch 1, Batch 79: Loss = 2186.907, a = 131.229 b = 37.950, c = -0.485, Joint = -9.535, Marginal = -16.982, MI = -19.219\n",
      "Pre-Train Epoch 1, Batch 80: Loss = 1561.147, a = 130.506 b = 45.097, c = -0.485, Joint = -9.193, Marginal = -12.556, MI = -18.078\n",
      "Pre-Train Epoch 1, Batch 81: Loss = 1054.696, a = 118.210 b = 39.359, c = -7.196, Joint = -8.649, Marginal = -24.465, MI = -19.200\n",
      "Pre-Train Epoch 1, Batch 82: Loss = 4809.814, a = 162.726 b = 43.119, c = -6.517, Joint = -12.125, Marginal = -31.970, MI = -20.938\n",
      "Pre-Train Epoch 1, Batch 83: Loss = 445.141, a = 116.573 b = 37.698, c = -0.484, Joint = -5.044, Marginal = -16.494, MI = -13.142\n",
      "Pre-Train Epoch 1, Batch 84: Loss = 902.607, a = 129.589 b = 35.793, c = -0.483, Joint = -6.373, Marginal = -25.936, MI = -13.620\n",
      "Pre-Train Epoch 1, Batch 85: Loss = 7582.986, a = 133.142 b = 39.072, c = -0.483, Joint = -18.263, Marginal = -30.839, MI = -27.450\n",
      "Pre-Train Epoch 1, Batch 86: Loss = 345.374, a = 116.082 b = 38.115, c = -0.482, Joint = -5.239, Marginal = -6.505, MI = -13.415\n",
      "Pre-Train Epoch 1, Batch 87: Loss = 1977.715, a = 115.851 b = 33.264, c = -0.482, Joint = -6.315, Marginal = -15.959, MI = -16.547\n",
      "Pre-Train Epoch 1, Batch 88: Loss = 257.442, a = 108.504 b = 41.405, c = -0.482, Joint = -2.861, Marginal = -3.722, MI = -12.412\n",
      "Pre-Train Epoch 1, Batch 89: Loss = 264.339, a = 130.154 b = 39.852, c = -2.525, Joint = -2.354, Marginal = -11.476, MI = -11.484\n",
      "Pre-Train Epoch 1, Batch 90: Loss = 1160.913, a = 160.895 b = 38.495, c = -0.481, Joint = -8.608, Marginal = -28.763, MI = -18.432\n",
      "Pre-Train Epoch 1, Batch 91: Loss = 319.290, a = 115.489 b = 39.761, c = -0.481, Joint = -3.137, Marginal = -11.418, MI = -12.484\n",
      "Pre-Train Epoch 1, Batch 92: Loss = 1417.264, a = 163.859 b = 31.401, c = -0.480, Joint = -11.499, Marginal = -28.165, MI = -22.602\n",
      "Pre-Train Epoch 1, Batch 93: Loss = 4509.916, a = 174.001 b = 51.719, c = -0.480, Joint = -18.182, Marginal = -17.913, MI = -28.188\n",
      "Pre-Train Epoch 1, Batch 94: Loss = 239.949, a = 123.065 b = 34.915, c = -0.480, Joint = -2.277, Marginal = -11.840, MI = -15.394\n",
      "Pre-Train Epoch 1, Batch 95: Loss = 3082.098, a = 171.837 b = 46.920, c = -0.479, Joint = -13.614, Marginal = -20.876, MI = -23.836\n",
      "Pre-Train Epoch 1, Batch 96: Loss = 1001.455, a = 132.846 b = 46.538, c = -0.479, Joint = -7.446, Marginal = -9.634, MI = -15.864\n",
      "Pre-Train Epoch 1, Batch 97: Loss = 297.968, a = 125.833 b = 41.133, c = -0.479, Joint = -4.256, Marginal = -7.329, MI = -16.265\n",
      "Pre-Train Epoch 1, Batch 98: Loss = 1595.748, a = 187.583 b = 43.807, c = -8.445, Joint = -9.478, Marginal = -15.973, MI = -18.101\n",
      "Pre-Train Epoch 1, Batch 99: Loss = 1542.910, a = 180.232 b = 43.998, c = -1.126, Joint = -7.657, Marginal = -22.905, MI = -17.274\n",
      "Pre-Train Epoch 1, Batch 100: Loss = 259.198, a = 120.898 b = 40.161, c = -0.478, Joint = -1.361, Marginal = -4.810, MI = -12.898\n",
      "Pre-Train Epoch 1, Batch 101: Loss = 4214.486, a = 223.147 b = 43.088, c = -7.088, Joint = -18.464, Marginal = -45.939, MI = -30.303\n",
      "Pre-Train Epoch 1, Batch 102: Loss = 1560.681, a = 160.642 b = 39.222, c = -0.477, Joint = -7.553, Marginal = -11.775, MI = -17.358\n",
      "Pre-Train Epoch 1, Batch 103: Loss = 357.175, a = 127.403 b = 40.154, c = -0.476, Joint = -3.678, Marginal = -12.995, MI = -12.265\n",
      "Pre-Train Epoch 1, Batch 104: Loss = 2612.076, a = 150.903 b = 38.411, c = -0.476, Joint = -12.536, Marginal = -18.031, MI = -22.559\n",
      "Pre-Train Epoch 1, Batch 105: Loss = 1701.546, a = 138.256 b = 36.145, c = -0.476, Joint = -10.416, Marginal = -17.875, MI = -19.770\n",
      "Pre-Train Epoch 1, Batch 106: Loss = 2428.175, a = 181.743 b = 42.180, c = -0.475, Joint = -13.209, Marginal = -25.482, MI = -25.105\n",
      "Pre-Train Epoch 1, Batch 107: Loss = 475.292, a = 143.342 b = 39.501, c = -4.245, Joint = -3.249, Marginal = -19.929, MI = -10.953\n",
      "Pre-Train Epoch 1, Batch 108: Loss = 3305.321, a = 117.574 b = 46.787, c = -0.475, Joint = -10.630, Marginal = -16.969, MI = -20.569\n",
      "Pre-Train Epoch 1, Batch 109: Loss = 9304.672, a = 118.128 b = 35.112, c = -2.663, Joint = -15.866, Marginal = -32.125, MI = -26.094\n",
      "Pre-Train Epoch 1, Batch 110: Loss = 7324.898, a = 168.883 b = 41.280, c = -0.915, Joint = -14.134, Marginal = -30.700, MI = -26.045\n",
      "Pre-Train Epoch 1, Batch 111: Loss = 14436.397, a = 192.920 b = 47.637, c = -0.474, Joint = -28.611, Marginal = -41.500, MI = -40.874\n",
      "Pre-Train Epoch 1, Batch 112: Loss = 1160.125, a = 145.271 b = 34.781, c = -1.029, Joint = -5.937, Marginal = -13.012, MI = -17.523\n",
      "Pre-Train Epoch 1, Batch 113: Loss = 1012.726, a = 186.733 b = 46.885, c = -0.473, Joint = -8.946, Marginal = -18.476, MI = -22.385\n",
      "Pre-Train Epoch 1, Batch 114: Loss = 519.096, a = 112.909 b = 38.189, c = -0.472, Joint = -4.237, Marginal = -5.756, MI = -12.202\n",
      "Pre-Train Epoch 1, Batch 115: Loss = 6249.164, a = 165.957 b = 36.496, c = -0.472, Joint = -12.516, Marginal = -21.510, MI = -27.128\n",
      "Pre-Train Epoch 1, Batch 116: Loss = 276.457, a = 121.705 b = 44.019, c = -0.471, Joint = -3.752, Marginal = -5.211, MI = -14.900\n",
      "Pre-Train Epoch 1, Batch 117: Loss = 1459.636, a = 166.910 b = 40.558, c = -0.471, Joint = -8.840, Marginal = -10.512, MI = -19.087\n",
      "Pre-Train Epoch 1, Batch 118: Loss = 530.774, a = 148.648 b = 47.380, c = -0.470, Joint = -4.594, Marginal = -6.762, MI = -17.863\n",
      "Pre-Train Epoch 1, Batch 119: Loss = 495.239, a = 156.519 b = 38.222, c = -0.470, Joint = -5.989, Marginal = -12.931, MI = -18.121\n",
      "Pre-Train Epoch 1, Batch 120: Loss = 206.874, a = 100.495 b = 32.655, c = -8.320, Joint = -2.358, Marginal = -6.386, MI = -13.172\n",
      "Pre-Train Epoch 1, Batch 121: Loss = 240.869, a = 121.030 b = 32.611, c = -0.630, Joint = -0.699, Marginal = -1.823, MI = -12.434\n",
      "Pre-Train Epoch 1, Batch 122: Loss = 367.246, a = 163.223 b = 37.358, c = -7.138, Joint = -3.639, Marginal = -5.966, MI = -16.815\n",
      "Pre-Train Epoch 1, Batch 123: Loss = 404.276, a = 197.646 b = 40.167, c = -2.203, Joint = -0.255, Marginal = -5.272, MI = -13.961\n",
      "Pre-Train Epoch 1, Batch 124: Loss = 296.071, a = 155.680 b = 31.704, c = -0.468, Joint = -0.009, Marginal = -2.999, MI = -14.362\n",
      "Pre-Train Epoch 1, Batch 125: Loss = 4128.152, a = 180.472 b = 32.525, c = -0.757, Joint = -13.560, Marginal = -24.492, MI = -28.084\n",
      "Pre-Train Epoch 1, Batch 126: Loss = 256.692, a = 126.708 b = 38.313, c = -0.468, Joint = -1.095, Marginal = -4.935, MI = -15.097\n",
      "Pre-Train Epoch 1, Batch 127: Loss = 407.662, a = 154.852 b = 47.169, c = -0.467, Joint = -4.492, Marginal = -12.444, MI = -19.003\n",
      "Pre-Train Epoch 1, Batch 128: Loss = 1136.728, a = 129.692 b = 31.066, c = -1.567, Joint = -4.168, Marginal = -3.257, MI = -15.585\n",
      "Pre-Train Epoch 1, Batch 129: Loss = 476.213, a = 161.255 b = 38.926, c = -0.467, Joint = -4.797, Marginal = -8.354, MI = -16.915\n",
      "Pre-Train Epoch 1, Batch 130: Loss = 341.355, a = 145.699 b = 45.572, c = -0.466, Joint = -3.643, Marginal = -7.089, MI = -19.279\n",
      "Pre-Train Epoch 1, Batch 131: Loss = 535.646, a = 125.127 b = 38.222, c = -4.352, Joint = -3.003, Marginal = -2.910, MI = -18.614\n",
      "Pre-Train Epoch 1, Batch 132: Loss = 1400.617, a = 126.626 b = 36.413, c = -0.466, Joint = -4.854, Marginal = -12.928, MI = -17.175\n",
      "Pre-Train Epoch 1, Batch 133: Loss = 8788.123, a = 179.338 b = 43.788, c = -0.465, Joint = -17.799, Marginal = -22.734, MI = -30.914\n",
      "Pre-Train Epoch 1, Batch 134: Loss = 1125.080, a = 157.221 b = 27.864, c = -0.465, Joint = -3.551, Marginal = -13.020, MI = -18.337\n",
      "Pre-Train Epoch 1, Batch 135: Loss = 272.283, a = 139.757 b = 35.481, c = -0.465, Joint = -1.013, Marginal = -4.005, MI = -15.122\n",
      "Pre-Train Epoch 1, Batch 136: Loss = 448.474, a = 159.690 b = 52.451, c = -0.464, Joint = -4.410, Marginal = -5.970, MI = -20.128\n",
      "Pre-Train Epoch 1, Batch 137: Loss = 295.141, a = 137.409 b = 35.821, c = -0.464, Joint = -1.926, Marginal = -6.158, MI = -14.011\n",
      "Pre-Train Epoch 1, Batch 138: Loss = 884.395, a = 123.531 b = 35.958, c = -0.464, Joint = -5.150, Marginal = -3.869, MI = -18.841\n",
      "Pre-Train Epoch 1, Batch 139: Loss = 628.647, a = 160.075 b = 49.511, c = -3.161, Joint = -6.760, Marginal = -8.259, MI = -18.300\n",
      "Pre-Train Epoch 1, Batch 140: Loss = 13080.311, a = 154.899 b = 43.670, c = -4.295, Joint = -19.335, Marginal = -24.571, MI = -32.328\n",
      "Pre-Train Epoch 1, Batch 141: Loss = 490.660, a = 205.646 b = 42.170, c = -0.463, Joint = -2.649, Marginal = -7.949, MI = -16.487\n",
      "Pre-Train Epoch 1, Batch 142: Loss = 1012.563, a = 173.276 b = 34.641, c = -0.462, Joint = -3.959, Marginal = -14.474, MI = -19.244\n",
      "Pre-Train Epoch 1, Batch 143: Loss = 4503.590, a = 196.916 b = 46.820, c = -0.462, Joint = -9.279, Marginal = -12.320, MI = -22.607\n",
      "Pre-Train Epoch 1, Batch 144: Loss = 698.762, a = 149.578 b = 45.059, c = -0.462, Joint = -4.986, Marginal = -7.319, MI = -16.284\n",
      "Pre-Train Epoch 1, Batch 145: Loss = 550.709, a = 151.014 b = 37.625, c = -0.461, Joint = -5.192, Marginal = -15.630, MI = -16.307\n",
      "Pre-Train Epoch 1, Batch 146: Loss = 317.101, a = 152.485 b = 40.490, c = -0.461, Joint = -2.456, Marginal = -3.899, MI = -19.008\n",
      "Pre-Train Epoch 1, Batch 147: Loss = 750.887, a = 219.104 b = 47.884, c = -0.460, Joint = -4.087, Marginal = -13.999, MI = -16.250\n",
      "Pre-Train Epoch 1, Batch 148: Loss = 695.343, a = 162.092 b = 37.800, c = -0.460, Joint = -4.811, Marginal = -9.945, MI = -18.406\n",
      "Pre-Train Epoch 1, Batch 149: Loss = 218.511, a = 99.521 b = 26.575, c = -5.413, Joint = -0.709, Marginal = -2.471, MI = -13.860\n",
      "Pre-Train Epoch 1, Batch 150: Loss = 587.839, a = 157.690 b = 39.990, c = -6.665, Joint = -3.781, Marginal = -12.430, MI = -16.596\n",
      "\n",
      "Start of epoch 2\n",
      "Pre-Train Epoch 2, Batch 0: Loss = 1066.972, a = 155.606 b = 40.168, c = -0.459, Joint = -5.165, Marginal = -4.905, MI = -18.976\n",
      "Pre-Train Epoch 2, Batch 1: Loss = 13703.116, a = 208.316 b = 46.089, c = -0.459, Joint = -19.435, Marginal = -23.859, MI = -32.289\n",
      "Pre-Train Epoch 2, Batch 2: Loss = 373.508, a = 150.819 b = 45.210, c = -0.458, Joint = -2.068, Marginal = -5.909, MI = -17.637\n",
      "Pre-Train Epoch 2, Batch 3: Loss = 585.039, a = 123.455 b = 38.828, c = -0.458, Joint = -3.816, Marginal = -5.192, MI = -19.220\n",
      "Pre-Train Epoch 2, Batch 4: Loss = 300.023, a = 120.586 b = 31.150, c = -0.458, Joint = -0.779, Marginal = -3.188, MI = -13.242\n",
      "Pre-Train Epoch 2, Batch 5: Loss = 384.093, a = 145.322 b = 36.702, c = -0.457, Joint = -3.914, Marginal = -6.152, MI = -17.852\n",
      "Pre-Train Epoch 2, Batch 6: Loss = 445.056, a = 159.899 b = 38.841, c = -2.225, Joint = -1.171, Marginal = -2.427, MI = -14.239\n",
      "Pre-Train Epoch 2, Batch 7: Loss = 274.585, a = 117.020 b = 44.770, c = -0.456, Joint = -2.145, Marginal = -3.500, MI = -13.966\n",
      "Pre-Train Epoch 2, Batch 8: Loss = 322.272, a = 128.603 b = 43.743, c = -0.456, Joint = -2.121, Marginal = -6.490, MI = -16.935\n",
      "Pre-Train Epoch 2, Batch 9: Loss = 3232.429, a = 224.881 b = 37.968, c = -0.456, Joint = -14.445, Marginal = -33.609, MI = -26.421\n",
      "Pre-Train Epoch 2, Batch 10: Loss = 551.945, a = 161.693 b = 48.789, c = -7.352, Joint = -4.378, Marginal = -8.922, MI = -18.263\n",
      "Pre-Train Epoch 2, Batch 11: Loss = 259.559, a = 128.639 b = 37.948, c = -0.455, Joint = -1.524, Marginal = -15.142, MI = -15.579\n",
      "Pre-Train Epoch 2, Batch 12: Loss = 1125.348, a = 156.451 b = 29.843, c = -0.455, Joint = -5.781, Marginal = -11.281, MI = -17.706\n",
      "Pre-Train Epoch 2, Batch 13: Loss = 423.685, a = 168.556 b = 35.479, c = -0.454, Joint = -1.241, Marginal = -2.724, MI = -20.146\n",
      "Pre-Train Epoch 2, Batch 14: Loss = 4610.435, a = 120.343 b = 31.672, c = -3.122, Joint = -8.870, Marginal = -14.660, MI = -23.351\n",
      "Pre-Train Epoch 2, Batch 15: Loss = 1583.254, a = 144.785 b = 38.913, c = -0.454, Joint = -7.442, Marginal = -3.431, MI = -22.422\n",
      "Pre-Train Epoch 2, Batch 16: Loss = 4899.662, a = 158.507 b = 44.396, c = -0.453, Joint = -13.158, Marginal = -16.986, MI = -25.474\n",
      "Pre-Train Epoch 2, Batch 17: Loss = 3921.972, a = 138.882 b = 44.007, c = -1.407, Joint = -11.415, Marginal = -23.426, MI = -22.324\n",
      "Pre-Train Epoch 2, Batch 18: Loss = 536.803, a = 139.308 b = 32.424, c = -0.453, Joint = -5.060, Marginal = -11.636, MI = -19.097\n",
      "Pre-Train Epoch 2, Batch 19: Loss = 534.198, a = 173.463 b = 34.374, c = -2.284, Joint = -1.130, Marginal = -8.793, MI = -16.458\n",
      "Pre-Train Epoch 2, Batch 20: Loss = 205.105, a = 98.446 b = 33.122, c = -0.452, Joint = -0.780, Marginal = -5.934, MI = -14.239\n",
      "Pre-Train Epoch 2, Batch 21: Loss = 808.511, a = 182.784 b = 37.358, c = -0.452, Joint = -6.529, Marginal = -23.956, MI = -18.360\n",
      "Pre-Train Epoch 2, Batch 22: Loss = 246.470, a = 133.819 b = 31.165, c = -6.743, Joint = -0.746, Marginal = -5.233, MI = -13.003\n",
      "Pre-Train Epoch 2, Batch 23: Loss = 907.351, a = 159.963 b = 36.109, c = -0.451, Joint = -5.534, Marginal = -7.833, MI = -18.861\n",
      "Pre-Train Epoch 2, Batch 24: Loss = 226.578, a = 129.940 b = 31.408, c = -0.451, Joint = -0.290, Marginal = -4.300, MI = -13.462\n",
      "Pre-Train Epoch 2, Batch 25: Loss = 4160.413, a = 177.820 b = 41.439, c = -0.450, Joint = -10.370, Marginal = -9.549, MI = -23.784\n",
      "Pre-Train Epoch 2, Batch 26: Loss = 630.807, a = 195.355 b = 47.784, c = -0.450, Joint = -4.861, Marginal = -9.469, MI = -19.952\n",
      "Pre-Train Epoch 2, Batch 27: Loss = 597.088, a = 168.834 b = 46.698, c = -0.450, Joint = -6.083, Marginal = -15.542, MI = -18.483\n",
      "Pre-Train Epoch 2, Batch 28: Loss = 1058.201, a = 161.394 b = 30.666, c = -2.611, Joint = -4.649, Marginal = -10.621, MI = -20.840\n",
      "Pre-Train Epoch 2, Batch 29: Loss = 318.336, a = 126.407 b = 42.429, c = -0.449, Joint = -2.562, Marginal = -3.375, MI = -14.489\n",
      "Pre-Train Epoch 2, Batch 30: Loss = 968.283, a = 154.789 b = 49.462, c = -0.449, Joint = -8.865, Marginal = -12.594, MI = -22.118\n",
      "Pre-Train Epoch 2, Batch 31: Loss = 2144.269, a = 168.199 b = 44.974, c = -12.085, Joint = -12.815, Marginal = -21.108, MI = -26.678\n",
      "Pre-Train Epoch 2, Batch 32: Loss = 394.142, a = 118.314 b = 38.623, c = -0.448, Joint = -2.829, Marginal = -4.665, MI = -12.046\n",
      "Pre-Train Epoch 2, Batch 33: Loss = 4606.636, a = 140.492 b = 34.798, c = -0.448, Joint = -11.851, Marginal = -23.884, MI = -23.496\n",
      "Pre-Train Epoch 2, Batch 34: Loss = 323.718, a = 156.941 b = 39.991, c = -6.613, Joint = 0.979, Marginal = -4.088, MI = -12.122\n",
      "Pre-Train Epoch 2, Batch 35: Loss = 408.083, a = 151.567 b = 42.972, c = -0.447, Joint = -4.215, Marginal = -7.213, MI = -19.760\n",
      "Pre-Train Epoch 2, Batch 36: Loss = 269.619, a = 118.569 b = 42.267, c = -3.910, Joint = -2.438, Marginal = -3.428, MI = -15.545\n",
      "Pre-Train Epoch 2, Batch 37: Loss = 525.139, a = 150.476 b = 47.159, c = -0.446, Joint = -6.246, Marginal = -13.046, MI = -17.101\n",
      "Pre-Train Epoch 2, Batch 38: Loss = 203.748, a = 94.180 b = 35.755, c = -0.575, Joint = -1.189, Marginal = -2.143, MI = -16.407\n",
      "Pre-Train Epoch 2, Batch 39: Loss = 313.511, a = 138.481 b = 45.438, c = -0.745, Joint = -2.637, Marginal = -7.593, MI = -15.907\n",
      "Pre-Train Epoch 2, Batch 40: Loss = 54479.617, a = 230.159 b = 34.148, c = -0.445, Joint = -38.168, Marginal = -117.134, MI = -50.217\n",
      "Pre-Train Epoch 2, Batch 41: Loss = 417.190, a = 187.342 b = 42.921, c = -0.445, Joint = -3.194, Marginal = -13.116, MI = -15.829\n",
      "Pre-Train Epoch 2, Batch 42: Loss = 274.521, a = 124.461 b = 41.824, c = -0.776, Joint = -2.672, Marginal = -7.636, MI = -13.766\n",
      "Pre-Train Epoch 2, Batch 43: Loss = 689.146, a = 168.535 b = 45.146, c = -7.751, Joint = -6.630, Marginal = -15.337, MI = -19.149\n",
      "Pre-Train Epoch 2, Batch 44: Loss = 844.859, a = 181.182 b = 40.027, c = -0.443, Joint = -5.521, Marginal = -14.968, MI = -19.273\n",
      "Pre-Train Epoch 2, Batch 45: Loss = 428.822, a = 147.479 b = 39.399, c = -3.278, Joint = -2.805, Marginal = -3.081, MI = -18.485\n",
      "Pre-Train Epoch 2, Batch 46: Loss = 368.523, a = 164.629 b = 43.982, c = -0.442, Joint = -4.049, Marginal = -9.262, MI = -14.709\n",
      "Pre-Train Epoch 2, Batch 47: Loss = 12677.900, a = 173.354 b = 34.022, c = -0.442, Joint = -16.474, Marginal = -19.318, MI = -30.412\n",
      "Pre-Train Epoch 2, Batch 48: Loss = 1309.752, a = 232.807 b = 39.802, c = -0.441, Joint = -8.912, Marginal = -17.969, MI = -23.390\n",
      "Pre-Train Epoch 2, Batch 49: Loss = 327.861, a = 139.433 b = 38.305, c = -0.441, Joint = -1.777, Marginal = -8.344, MI = -15.291\n",
      "Pre-Train Epoch 2, Batch 50: Loss = 783.497, a = 156.547 b = 32.004, c = -0.440, Joint = -5.337, Marginal = -11.652, MI = -14.847\n",
      "Pre-Train Epoch 2, Batch 51: Loss = 773.312, a = 203.918 b = 39.062, c = -2.117, Joint = -5.227, Marginal = -20.611, MI = -19.717\n",
      "Pre-Train Epoch 2, Batch 52: Loss = 260.735, a = 129.199 b = 33.382, c = -1.094, Joint = -0.296, Marginal = -2.356, MI = -12.775\n",
      "Pre-Train Epoch 2, Batch 53: Loss = 515.219, a = 187.347 b = 27.089, c = -3.764, Joint = -0.896, Marginal = -7.781, MI = -16.174\n",
      "Pre-Train Epoch 2, Batch 54: Loss = 450.357, a = 135.919 b = 43.964, c = -0.438, Joint = -3.834, Marginal = -4.740, MI = -17.410\n",
      "Pre-Train Epoch 2, Batch 55: Loss = 33893.352, a = 218.367 b = 41.436, c = -0.438, Joint = -26.415, Marginal = -39.862, MI = -42.731\n",
      "Pre-Train Epoch 2, Batch 56: Loss = 817.343, a = 163.235 b = 39.884, c = -3.347, Joint = -4.682, Marginal = -4.339, MI = -19.326\n",
      "Pre-Train Epoch 2, Batch 57: Loss = 284.626, a = 140.652 b = 37.682, c = -7.252, Joint = -0.633, Marginal = -2.446, MI = -15.633\n",
      "Pre-Train Epoch 2, Batch 58: Loss = 586.791, a = 155.505 b = 43.653, c = -5.117, Joint = -3.708, Marginal = -5.067, MI = -17.257\n",
      "Pre-Train Epoch 2, Batch 59: Loss = 331.374, a = 142.153 b = 34.608, c = -1.194, Joint = 0.479, Marginal = -1.804, MI = -15.153\n",
      "Pre-Train Epoch 2, Batch 60: Loss = 14292.210, a = 193.671 b = 37.705, c = -0.435, Joint = -19.033, Marginal = -25.480, MI = -36.117\n",
      "Pre-Train Epoch 2, Batch 61: Loss = 417.260, a = 183.296 b = 37.310, c = -0.435, Joint = -1.528, Marginal = -7.168, MI = -17.435\n",
      "Pre-Train Epoch 2, Batch 62: Loss = 2405.731, a = 207.751 b = 35.380, c = -6.286, Joint = -9.505, Marginal = -15.718, MI = -25.816\n",
      "Pre-Train Epoch 2, Batch 63: Loss = 1017.607, a = 183.982 b = 33.207, c = -0.434, Joint = -4.550, Marginal = -11.234, MI = -18.972\n",
      "Pre-Train Epoch 2, Batch 64: Loss = 333.491, a = 144.677 b = 44.118, c = -0.433, Joint = -1.803, Marginal = -6.639, MI = -16.614\n",
      "Pre-Train Epoch 2, Batch 65: Loss = 6174.562, a = 260.788 b = 32.440, c = -1.856, Joint = -15.493, Marginal = -28.545, MI = -34.218\n",
      "Pre-Train Epoch 2, Batch 66: Loss = 330.647, a = 145.987 b = 46.862, c = -0.432, Joint = -3.009, Marginal = -6.953, MI = -18.797\n",
      "Pre-Train Epoch 2, Batch 67: Loss = 499.093, a = 195.965 b = 52.379, c = -0.432, Joint = -4.267, Marginal = -17.775, MI = -17.825\n",
      "Pre-Train Epoch 2, Batch 68: Loss = 786.734, a = 220.025 b = 34.377, c = -0.432, Joint = -1.509, Marginal = -17.294, MI = -21.210\n",
      "Pre-Train Epoch 2, Batch 69: Loss = 1008.726, a = 181.581 b = 35.907, c = -0.431, Joint = -5.767, Marginal = -10.485, MI = -24.467\n",
      "Pre-Train Epoch 2, Batch 70: Loss = 1513.520, a = 155.777 b = 45.720, c = -3.355, Joint = -6.597, Marginal = -8.939, MI = -26.395\n",
      "Pre-Train Epoch 2, Batch 71: Loss = 5596.084, a = 233.556 b = 54.232, c = -0.430, Joint = -14.144, Marginal = -36.257, MI = -32.023\n",
      "Pre-Train Epoch 2, Batch 72: Loss = 1974.160, a = 260.201 b = 32.199, c = -0.430, Joint = -8.188, Marginal = -10.359, MI = -25.941\n",
      "Pre-Train Epoch 2, Batch 73: Loss = 1474.506, a = 198.612 b = 29.180, c = -0.429, Joint = -3.807, Marginal = -14.211, MI = -21.277\n",
      "Pre-Train Epoch 2, Batch 74: Loss = 357.317, a = 192.437 b = 37.830, c = -1.060, Joint = -2.489, Marginal = -8.928, MI = -17.216\n",
      "Pre-Train Epoch 2, Batch 75: Loss = 511.990, a = 203.439 b = 32.055, c = -0.429, Joint = -0.632, Marginal = -0.578, MI = -19.245\n",
      "Pre-Train Epoch 2, Batch 76: Loss = 1044.363, a = 167.168 b = 38.082, c = -0.428, Joint = -4.772, Marginal = -2.624, MI = -20.115\n",
      "Pre-Train Epoch 2, Batch 77: Loss = 3484.156, a = 194.846 b = 37.459, c = -0.428, Joint = -12.035, Marginal = -19.719, MI = -29.801\n",
      "Pre-Train Epoch 2, Batch 78: Loss = 395.092, a = 112.010 b = 46.101, c = -0.427, Joint = -4.805, Marginal = -6.010, MI = -16.668\n",
      "Pre-Train Epoch 2, Batch 79: Loss = 1347.267, a = 183.611 b = 36.563, c = -0.427, Joint = -4.824, Marginal = -11.746, MI = -18.891\n",
      "Pre-Train Epoch 2, Batch 80: Loss = 1033.244, a = 158.579 b = 43.498, c = -0.427, Joint = -5.492, Marginal = -7.341, MI = -17.380\n",
      "Pre-Train Epoch 2, Batch 81: Loss = 635.148, a = 166.916 b = 38.083, c = -6.774, Joint = -4.600, Marginal = -9.398, MI = -20.259\n",
      "Pre-Train Epoch 2, Batch 82: Loss = 2884.991, a = 224.637 b = 41.402, c = -6.072, Joint = -6.791, Marginal = -27.744, MI = -28.207\n",
      "Pre-Train Epoch 2, Batch 83: Loss = 324.945, a = 148.715 b = 36.493, c = -0.426, Joint = -2.047, Marginal = -3.982, MI = -20.449\n",
      "Pre-Train Epoch 2, Batch 84: Loss = 637.575, a = 187.251 b = 34.597, c = -0.425, Joint = -2.342, Marginal = -5.349, MI = -16.276\n",
      "Pre-Train Epoch 2, Batch 85: Loss = 4584.611, a = 195.555 b = 37.762, c = -0.425, Joint = -12.370, Marginal = -33.425, MI = -29.867\n",
      "Pre-Train Epoch 2, Batch 86: Loss = 301.672, a = 150.759 b = 36.876, c = -0.425, Joint = -2.128, Marginal = -4.791, MI = -15.792\n",
      "Pre-Train Epoch 2, Batch 87: Loss = 1254.321, a = 168.557 b = 32.495, c = -0.424, Joint = -2.497, Marginal = -6.917, MI = -21.337\n",
      "Pre-Train Epoch 2, Batch 88: Loss = 260.410, a = 121.507 b = 40.206, c = -0.424, Joint = -0.735, Marginal = -1.841, MI = -16.943\n",
      "Pre-Train Epoch 2, Batch 89: Loss = 302.702, a = 157.731 b = 38.556, c = -2.279, Joint = 0.242, Marginal = -4.200, MI = -13.836\n",
      "Pre-Train Epoch 2, Batch 90: Loss = 704.169, a = 224.068 b = 37.046, c = -0.423, Joint = -3.776, Marginal = -15.198, MI = -19.811\n",
      "Pre-Train Epoch 2, Batch 91: Loss = 295.274, a = 127.154 b = 38.524, c = -0.423, Joint = -0.998, Marginal = -4.746, MI = -13.599\n",
      "Pre-Train Epoch 2, Batch 92: Loss = 833.622, a = 263.685 b = 30.277, c = -0.423, Joint = -5.244, Marginal = -20.775, MI = -26.039\n",
      "Pre-Train Epoch 2, Batch 93: Loss = 2739.328, a = 212.268 b = 49.625, c = -0.422, Joint = -12.657, Marginal = -24.796, MI = -26.974\n",
      "Pre-Train Epoch 2, Batch 94: Loss = 274.407, a = 156.328 b = 33.863, c = -0.422, Joint = 0.206, Marginal = -10.359, MI = -14.135\n",
      "Pre-Train Epoch 2, Batch 95: Loss = 1933.979, a = 212.050 b = 45.012, c = -0.422, Joint = -8.555, Marginal = -17.965, MI = -23.669\n",
      "Pre-Train Epoch 2, Batch 96: Loss = 707.058, a = 151.346 b = 45.021, c = -0.421, Joint = -4.477, Marginal = -9.055, MI = -16.336\n",
      "Pre-Train Epoch 2, Batch 97: Loss = 283.302, a = 147.424 b = 39.851, c = -0.421, Joint = -1.670, Marginal = -4.751, MI = -18.124\n",
      "Pre-Train Epoch 2, Batch 98: Loss = 996.597, a = 228.337 b = 41.928, c = -7.978, Joint = -4.891, Marginal = -13.642, MI = -20.323\n",
      "Pre-Train Epoch 2, Batch 99: Loss = 1037.248, a = 211.277 b = 42.146, c = -0.913, Joint = -3.619, Marginal = -10.222, MI = -20.559\n",
      "Pre-Train Epoch 2, Batch 100: Loss = 282.353, a = 130.902 b = 38.998, c = -0.420, Joint = 0.440, Marginal = -2.354, MI = -16.500\n",
      "Pre-Train Epoch 2, Batch 101: Loss = 2640.583, a = 297.996 b = 41.036, c = -6.638, Joint = -11.391, Marginal = -26.175, MI = -27.347\n",
      "Pre-Train Epoch 2, Batch 102: Loss = 1096.172, a = 185.797 b = 37.677, c = -0.420, Joint = -4.020, Marginal = -7.873, MI = -20.605\n",
      "Pre-Train Epoch 2, Batch 103: Loss = 289.957, a = 136.901 b = 38.825, c = -0.419, Joint = -1.591, Marginal = -3.752, MI = -16.721\n",
      "Pre-Train Epoch 2, Batch 104: Loss = 1627.012, a = 189.284 b = 37.028, c = -0.419, Joint = -8.087, Marginal = -15.157, MI = -23.471\n",
      "Pre-Train Epoch 2, Batch 105: Loss = 1087.530, a = 170.189 b = 34.914, c = -0.419, Joint = -6.398, Marginal = -6.211, MI = -23.975\n",
      "Pre-Train Epoch 2, Batch 106: Loss = 1478.120, a = 223.044 b = 40.560, c = -0.418, Joint = -8.287, Marginal = -16.439, MI = -27.036\n",
      "Pre-Train Epoch 2, Batch 107: Loss = 386.731, a = 155.979 b = 38.139, c = -4.015, Joint = -0.951, Marginal = -5.637, MI = -15.201\n",
      "Pre-Train Epoch 2, Batch 108: Loss = 2148.531, a = 111.595 b = 45.306, c = -0.418, Joint = -8.194, Marginal = -8.630, MI = -23.690\n",
      "Pre-Train Epoch 2, Batch 109: Loss = 5962.623, a = 141.311 b = 34.141, c = -2.452, Joint = -11.773, Marginal = -20.508, MI = -30.251\n",
      "Pre-Train Epoch 2, Batch 110: Loss = 4747.434, a = 198.893 b = 39.855, c = -0.719, Joint = -9.710, Marginal = -18.653, MI = -26.083\n",
      "Pre-Train Epoch 2, Batch 111: Loss = 9419.309, a = 230.447 b = 45.783, c = -0.417, Joint = -21.630, Marginal = -27.533, MI = -37.970\n",
      "Pre-Train Epoch 2, Batch 112: Loss = 754.672, a = 162.444 b = 33.658, c = -0.879, Joint = -3.040, Marginal = -9.758, MI = -17.851\n",
      "Pre-Train Epoch 2, Batch 113: Loss = 672.136, a = 198.324 b = 45.090, c = -0.416, Joint = -5.650, Marginal = -10.616, MI = -19.971\n",
      "Pre-Train Epoch 2, Batch 114: Loss = 389.985, a = 114.049 b = 37.195, c = -0.415, Joint = -2.542, Marginal = -2.301, MI = -23.078\n",
      "Pre-Train Epoch 2, Batch 115: Loss = 3964.445, a = 190.465 b = 35.306, c = -0.415, Joint = -8.468, Marginal = -28.986, MI = -26.480\n",
      "Pre-Train Epoch 2, Batch 116: Loss = 260.480, a = 122.138 b = 42.871, c = -0.414, Joint = -2.201, Marginal = -6.252, MI = -17.675\n",
      "Pre-Train Epoch 2, Batch 117: Loss = 1013.146, a = 172.388 b = 39.033, c = -0.414, Joint = -5.831, Marginal = -6.396, MI = -19.993\n",
      "Pre-Train Epoch 2, Batch 118: Loss = 392.781, a = 139.703 b = 45.809, c = -0.414, Joint = -2.837, Marginal = -6.828, MI = -19.003\n",
      "Pre-Train Epoch 2, Batch 119: Loss = 348.386, a = 162.784 b = 36.882, c = -0.413, Joint = -3.393, Marginal = -10.426, MI = -17.812\n",
      "Pre-Train Epoch 2, Batch 120: Loss = 204.709, a = 105.232 b = 32.004, c = -7.942, Joint = -1.090, Marginal = -1.246, MI = -18.402\n",
      "Pre-Train Epoch 2, Batch 121: Loss = 233.312, a = 120.760 b = 31.727, c = -0.499, Joint = 0.539, Marginal = -1.355, MI = -14.050\n",
      "Pre-Train Epoch 2, Batch 122: Loss = 323.824, a = 175.455 b = 36.149, c = -6.717, Joint = -1.268, Marginal = -4.005, MI = -18.565\n",
      "Pre-Train Epoch 2, Batch 123: Loss = 398.136, a = 196.718 b = 38.542, c = -2.054, Joint = 1.583, Marginal = -8.229, MI = -14.584\n",
      "Pre-Train Epoch 2, Batch 124: Loss = 306.253, a = 165.551 b = 30.830, c = -0.412, Joint = 1.723, Marginal = 0.003, MI = -13.517\n",
      "Pre-Train Epoch 2, Batch 125: Loss = 2590.056, a = 197.864 b = 31.403, c = -0.753, Joint = -9.512, Marginal = -18.543, MI = -25.296\n",
      "Pre-Train Epoch 2, Batch 126: Loss = 250.915, a = 120.555 b = 37.293, c = -0.411, Joint = -0.052, Marginal = -3.250, MI = -18.976\n",
      "Pre-Train Epoch 2, Batch 127: Loss = 323.216, a = 142.645 b = 45.564, c = -0.411, Joint = -2.865, Marginal = -6.788, MI = -17.661\n",
      "Pre-Train Epoch 2, Batch 128: Loss = 784.552, a = 130.252 b = 30.246, c = -1.471, Joint = -2.439, Marginal = -3.216, MI = -13.627\n",
      "Pre-Train Epoch 2, Batch 129: Loss = 349.307, a = 160.372 b = 37.688, c = -0.410, Joint = -2.627, Marginal = -8.382, MI = -17.446\n",
      "Pre-Train Epoch 2, Batch 130: Loss = 300.782, a = 136.042 b = 44.171, c = -0.410, Joint = -2.065, Marginal = -4.448, MI = -16.414\n",
      "Pre-Train Epoch 2, Batch 131: Loss = 407.115, a = 117.592 b = 37.254, c = -4.150, Joint = -1.690, Marginal = -2.243, MI = -17.306\n",
      "Pre-Train Epoch 2, Batch 132: Loss = 950.159, a = 124.718 b = 35.560, c = -0.409, Joint = -3.139, Marginal = -5.980, MI = -17.566\n",
      "Pre-Train Epoch 2, Batch 133: Loss = 5744.416, a = 179.330 b = 42.244, c = -0.409, Joint = -13.712, Marginal = -20.817, MI = -26.880\n",
      "Pre-Train Epoch 2, Batch 134: Loss = 781.985, a = 168.309 b = 27.183, c = -0.409, Joint = -1.282, Marginal = -8.823, MI = -21.741\n",
      "Pre-Train Epoch 2, Batch 135: Loss = 263.826, a = 135.761 b = 34.502, c = -0.409, Joint = 0.238, Marginal = -5.024, MI = -18.961\n",
      "Pre-Train Epoch 2, Batch 136: Loss = 350.113, a = 133.909 b = 50.614, c = -0.408, Joint = -3.090, Marginal = -4.161, MI = -14.824\n",
      "Pre-Train Epoch 2, Batch 137: Loss = 250.418, a = 127.595 b = 34.786, c = -0.408, Joint = -0.719, Marginal = -5.994, MI = -18.131\n",
      "Pre-Train Epoch 2, Batch 138: Loss = 602.260, a = 115.508 b = 35.049, c = -0.408, Joint = -3.539, Marginal = -5.317, MI = -17.887\n",
      "Pre-Train Epoch 2, Batch 139: Loss = 439.462, a = 143.890 b = 47.799, c = -2.951, Joint = -4.722, Marginal = -7.646, MI = -17.972\n",
      "Pre-Train Epoch 2, Batch 140: Loss = 8783.185, a = 146.315 b = 42.264, c = -4.011, Joint = -15.528, Marginal = -25.659, MI = -27.674\n",
      "Pre-Train Epoch 2, Batch 141: Loss = 399.942, a = 192.364 b = 40.440, c = -0.407, Joint = -0.710, Marginal = -6.580, MI = -16.612\n",
      "Pre-Train Epoch 2, Batch 142: Loss = 691.376, a = 170.731 b = 33.574, c = -0.406, Joint = -1.843, Marginal = -10.419, MI = -16.152\n",
      "Pre-Train Epoch 2, Batch 143: Loss = 3097.783, a = 179.532 b = 45.012, c = -0.406, Joint = -6.655, Marginal = -13.970, MI = -19.805\n",
      "Pre-Train Epoch 2, Batch 144: Loss = 509.047, a = 133.888 b = 43.700, c = -0.406, Joint = -3.487, Marginal = -5.081, MI = -16.967\n",
      "Pre-Train Epoch 2, Batch 145: Loss = 378.544, a = 138.516 b = 36.368, c = -0.405, Joint = -3.344, Marginal = -8.684, MI = -20.501\n",
      "Pre-Train Epoch 2, Batch 146: Loss = 267.968, a = 136.522 b = 39.223, c = -0.405, Joint = -1.154, Marginal = -6.112, MI = -13.444\n",
      "Pre-Train Epoch 2, Batch 147: Loss = 536.086, a = 193.604 b = 45.887, c = -0.404, Joint = -2.111, Marginal = -9.364, MI = -20.483\n",
      "Pre-Train Epoch 2, Batch 148: Loss = 488.238, a = 147.377 b = 36.521, c = -0.404, Joint = -3.006, Marginal = -6.297, MI = -15.372\n",
      "Pre-Train Epoch 2, Batch 149: Loss = 192.537, a = 93.216 b = 26.190, c = -5.146, Joint = 0.044, Marginal = -0.824, MI = -16.869\n",
      "Pre-Train Epoch 2, Batch 150: Loss = 409.979, a = 142.612 b = 38.767, c = -6.412, Joint = -2.256, Marginal = -10.000, MI = -20.409\n",
      "\n",
      "Start of epoch 3\n",
      "Pre-Train Epoch 3, Batch 0: Loss = 746.751, a = 136.746 b = 38.794, c = -0.403, Joint = -3.596, Marginal = -5.887, MI = -15.690\n",
      "Pre-Train Epoch 3, Batch 1: Loss = 9208.276, a = 191.980 b = 44.241, c = -0.403, Joint = -15.123, Marginal = -20.089, MI = -31.261\n",
      "Pre-Train Epoch 3, Batch 2: Loss = 309.386, a = 128.550 b = 43.760, c = -0.402, Joint = -1.113, Marginal = -2.221, MI = -16.681\n",
      "Pre-Train Epoch 3, Batch 3: Loss = 420.067, a = 106.953 b = 37.798, c = -0.402, Joint = -2.740, Marginal = -3.753, MI = -16.406\n",
      "Pre-Train Epoch 3, Batch 4: Loss = 246.374, a = 109.766 b = 30.493, c = -0.402, Joint = 0.002, Marginal = -3.168, MI = -17.607\n",
      "Pre-Train Epoch 3, Batch 5: Loss = 285.119, a = 131.636 b = 35.663, c = -0.401, Joint = -2.406, Marginal = -6.406, MI = -11.727\n",
      "Pre-Train Epoch 3, Batch 6: Loss = 349.845, a = 140.390 b = 37.658, c = -1.882, Joint = -0.119, Marginal = -2.026, MI = -15.814\n",
      "Pre-Train Epoch 3, Batch 7: Loss = 238.676, a = 95.811 b = 43.556, c = -0.401, Joint = -1.533, Marginal = -1.774, MI = -17.097\n",
      "Pre-Train Epoch 3, Batch 8: Loss = 266.452, a = 106.320 b = 42.433, c = -0.400, Joint = -1.445, Marginal = -3.809, MI = -16.517\n",
      "Pre-Train Epoch 3, Batch 9: Loss = 2048.502, a = 213.190 b = 36.499, c = -0.400, Joint = -10.510, Marginal = -17.775, MI = -25.561\n",
      "Pre-Train Epoch 3, Batch 10: Loss = 391.072, a = 135.037 b = 47.050, c = -6.906, Joint = -3.129, Marginal = -11.441, MI = -18.987\n",
      "Pre-Train Epoch 3, Batch 11: Loss = 230.901, a = 111.972 b = 36.949, c = -0.399, Joint = -0.732, Marginal = -11.671, MI = -18.079\n",
      "Pre-Train Epoch 3, Batch 12: Loss = 730.992, a = 146.204 b = 29.100, c = -0.399, Joint = -3.783, Marginal = -14.334, MI = -19.535\n",
      "Pre-Train Epoch 3, Batch 13: Loss = 332.529, a = 147.446 b = 34.309, c = -0.399, Joint = -0.073, Marginal = -2.829, MI = -14.770\n",
      "Pre-Train Epoch 3, Batch 14: Loss = 3080.443, a = 108.156 b = 30.938, c = -2.897, Joint = -6.969, Marginal = -16.767, MI = -18.900\n",
      "Pre-Train Epoch 3, Batch 15: Loss = 1101.965, a = 127.322 b = 37.814, c = -0.398, Joint = -5.618, Marginal = -4.224, MI = -20.680\n",
      "Pre-Train Epoch 3, Batch 16: Loss = 3432.068, a = 137.218 b = 42.963, c = -0.398, Joint = -10.569, Marginal = -7.171, MI = -24.446\n",
      "Pre-Train Epoch 3, Batch 17: Loss = 2571.402, a = 120.309 b = 42.810, c = -1.302, Joint = -9.133, Marginal = -31.268, MI = -22.221\n",
      "Pre-Train Epoch 3, Batch 18: Loss = 367.182, a = 126.813 b = 31.627, c = -0.397, Joint = -3.318, Marginal = -8.649, MI = -18.559\n",
      "Pre-Train Epoch 3, Batch 19: Loss = 393.848, a = 148.826 b = 33.250, c = -2.105, Joint = -0.161, Marginal = -6.796, MI = -13.111\n",
      "Pre-Train Epoch 3, Batch 20: Loss = 182.418, a = 82.699 b = 32.451, c = -0.397, Joint = -0.342, Marginal = -2.437, MI = -16.988\n",
      "Pre-Train Epoch 3, Batch 21: Loss = 544.485, a = 162.203 b = 36.080, c = -0.396, Joint = -4.457, Marginal = -10.675, MI = -19.060\n",
      "Pre-Train Epoch 3, Batch 22: Loss = 213.453, a = 116.165 b = 30.432, c = -6.358, Joint = 0.015, Marginal = -3.168, MI = -12.214\n",
      "Pre-Train Epoch 3, Batch 23: Loss = 621.225, a = 138.436 b = 35.039, c = -0.396, Joint = -3.919, Marginal = -8.650, MI = -20.767\n",
      "Pre-Train Epoch 3, Batch 24: Loss = 204.049, a = 112.689 b = 30.691, c = -0.395, Joint = 0.386, Marginal = -7.008, MI = -12.087\n",
      "Pre-Train Epoch 3, Batch 25: Loss = 2906.046, a = 152.532 b = 39.995, c = -0.395, Joint = -7.974, Marginal = -16.504, MI = -19.044\n",
      "Pre-Train Epoch 3, Batch 26: Loss = 463.533, a = 161.677 b = 45.940, c = -0.395, Joint = -3.213, Marginal = -5.454, MI = -21.081\n",
      "Pre-Train Epoch 3, Batch 27: Loss = 411.316, a = 140.720 b = 45.102, c = -0.395, Joint = -4.407, Marginal = -5.209, MI = -14.911\n",
      "Pre-Train Epoch 3, Batch 28: Loss = 733.989, a = 142.925 b = 29.771, c = -2.559, Joint = -2.943, Marginal = -9.874, MI = -15.243\n",
      "Pre-Train Epoch 3, Batch 29: Loss = 248.565, a = 99.433 b = 41.172, c = -0.394, Joint = -1.944, Marginal = -2.442, MI = -16.288\n",
      "Pre-Train Epoch 3, Batch 30: Loss = 650.968, a = 127.294 b = 47.855, c = -0.394, Joint = -6.910, Marginal = -13.140, MI = -20.330\n",
      "Pre-Train Epoch 3, Batch 31: Loss = 1386.768, a = 141.367 b = 43.411, c = -11.530, Joint = -9.973, Marginal = -15.970, MI = -21.743\n",
      "Pre-Train Epoch 3, Batch 32: Loss = 278.043, a = 96.174 b = 37.569, c = -0.393, Joint = -2.066, Marginal = -7.456, MI = -14.941\n",
      "Pre-Train Epoch 3, Batch 33: Loss = 3090.306, a = 121.184 b = 33.792, c = -0.393, Joint = -9.409, Marginal = -14.225, MI = -20.125\n",
      "Pre-Train Epoch 3, Batch 34: Loss = 271.567, a = 125.551 b = 38.691, c = -6.268, Joint = 1.216, Marginal = -4.503, MI = -11.171\n",
      "Pre-Train Epoch 3, Batch 35: Loss = 299.945, a = 123.182 b = 41.577, c = -0.392, Joint = -2.999, Marginal = -5.794, MI = -19.587\n",
      "Pre-Train Epoch 3, Batch 36: Loss = 221.046, a = 93.930 b = 41.133, c = -3.648, Joint = -1.825, Marginal = -2.460, MI = -16.956\n",
      "Pre-Train Epoch 3, Batch 37: Loss = 368.401, a = 122.123 b = 45.668, c = -0.391, Joint = -4.732, Marginal = -5.896, MI = -14.695\n",
      "Pre-Train Epoch 3, Batch 38: Loss = 168.104, a = 73.579 b = 35.055, c = -0.399, Joint = -0.882, Marginal = -3.436, MI = -14.840\n",
      "Pre-Train Epoch 3, Batch 39: Loss = 244.070, a = 108.703 b = 44.045, c = -0.560, Joint = -1.900, Marginal = -6.457, MI = -13.979\n",
      "Pre-Train Epoch 3, Batch 40: Loss = 38361.770, a = 202.460 b = 32.715, c = -0.391, Joint = -31.352, Marginal = -69.827, MI = -42.160\n",
      "Pre-Train Epoch 3, Batch 41: Loss = 307.110, a = 153.426 b = 41.441, c = -0.390, Joint = -1.968, Marginal = -19.679, MI = -16.148\n",
      "Pre-Train Epoch 3, Batch 42: Loss = 215.601, a = 98.429 b = 40.618, c = -0.646, Joint = -1.960, Marginal = -5.321, MI = -11.421\n",
      "Pre-Train Epoch 3, Batch 43: Loss = 466.896, a = 136.276 b = 43.605, c = -7.344, Joint = -4.977, Marginal = -9.575, MI = -16.350\n",
      "Pre-Train Epoch 3, Batch 44: Loss = 557.790, a = 148.456 b = 38.670, c = -0.388, Joint = -4.040, Marginal = -13.113, MI = -18.223\n",
      "Pre-Train Epoch 3, Batch 45: Loss = 323.709, a = 117.935 b = 38.242, c = -3.085, Joint = -1.918, Marginal = -1.827, MI = -14.786\n",
      "Pre-Train Epoch 3, Batch 46: Loss = 272.836, a = 131.116 b = 42.545, c = -0.388, Joint = -2.878, Marginal = -8.162, MI = -14.163\n",
      "Pre-Train Epoch 3, Batch 47: Loss = 9054.760, a = 144.904 b = 32.974, c = -0.387, Joint = -13.534, Marginal = -27.826, MI = -27.922\n",
      "Pre-Train Epoch 3, Batch 48: Loss = 837.753, a = 193.682 b = 38.292, c = -0.387, Joint = -6.409, Marginal = -26.794, MI = -18.578\n",
      "Pre-Train Epoch 3, Batch 49: Loss = 261.872, a = 111.633 b = 37.316, c = -0.386, Joint = -1.207, Marginal = -3.323, MI = -18.120\n",
      "Pre-Train Epoch 3, Batch 50: Loss = 513.475, a = 125.419 b = 31.067, c = -0.386, Joint = -4.161, Marginal = -8.762, MI = -15.826\n",
      "Pre-Train Epoch 3, Batch 51: Loss = 520.455, a = 165.568 b = 37.716, c = -1.986, Joint = -3.613, Marginal = -6.766, MI = -15.341\n",
      "Pre-Train Epoch 3, Batch 52: Loss = 204.229, a = 99.353 b = 32.504, c = -0.955, Joint = -0.091, Marginal = -2.270, MI = -11.092\n",
      "Pre-Train Epoch 3, Batch 53: Loss = 375.287, a = 153.262 b = 26.344, c = -3.509, Joint = 0.024, Marginal = -1.564, MI = -16.712\n",
      "Pre-Train Epoch 3, Batch 54: Loss = 319.471, a = 103.109 b = 42.755, c = -0.384, Joint = -3.146, Marginal = -3.735, MI = -15.576\n",
      "Pre-Train Epoch 3, Batch 55: Loss = 23965.223, a = 179.171 b = 39.972, c = -0.384, Joint = -21.924, Marginal = -35.827, MI = -34.030\n",
      "Pre-Train Epoch 3, Batch 56: Loss = 567.516, a = 127.942 b = 38.746, c = -3.130, Joint = -3.669, Marginal = -3.045, MI = -19.913\n",
      "Pre-Train Epoch 3, Batch 57: Loss = 224.362, a = 108.361 b = 36.730, c = -6.856, Joint = -0.412, Marginal = -3.946, MI = -12.874\n",
      "Pre-Train Epoch 3, Batch 58: Loss = 415.199, a = 117.830 b = 42.234, c = -4.892, Joint = -2.959, Marginal = -8.026, MI = -13.210\n",
      "Pre-Train Epoch 3, Batch 59: Loss = 255.841, a = 108.041 b = 33.743, c = -1.007, Joint = 0.437, Marginal = -0.797, MI = -15.298\n",
      "Pre-Train Epoch 3, Batch 60: Loss = 9883.742, a = 157.820 b = 36.616, c = -0.381, Joint = -15.727, Marginal = -26.024, MI = -29.632\n",
      "Pre-Train Epoch 3, Batch 61: Loss = 312.722, a = 143.113 b = 36.147, c = -0.381, Joint = -0.869, Marginal = -2.119, MI = -15.701\n",
      "Pre-Train Epoch 3, Batch 62: Loss = 1621.101, a = 166.217 b = 34.308, c = -5.964, Joint = -7.493, Marginal = -8.827, MI = -20.474\n",
      "Pre-Train Epoch 3, Batch 63: Loss = 668.884, a = 146.107 b = 32.275, c = -0.380, Joint = -3.401, Marginal = -9.506, MI = -17.177\n",
      "Pre-Train Epoch 3, Batch 64: Loss = 251.393, a = 108.479 b = 42.871, c = -0.380, Joint = -1.481, Marginal = -6.689, MI = -13.991\n",
      "Pre-Train Epoch 3, Batch 65: Loss = 4315.334, a = 214.018 b = 31.256, c = -1.694, Joint = -12.175, Marginal = -38.691, MI = -31.288\n",
      "Pre-Train Epoch 3, Batch 66: Loss = 247.660, a = 108.563 b = 45.521, c = -0.379, Joint = -2.492, Marginal = -5.173, MI = -17.258\n",
      "Pre-Train Epoch 3, Batch 67: Loss = 350.632, a = 147.012 b = 50.409, c = -0.378, Joint = -3.399, Marginal = -7.340, MI = -16.028\n",
      "Pre-Train Epoch 3, Batch 68: Loss = 531.974, a = 171.464 b = 33.261, c = -0.378, Joint = -0.818, Marginal = -10.661, MI = -18.680\n",
      "Pre-Train Epoch 3, Batch 69: Loss = 681.072, a = 142.323 b = 34.903, c = -0.378, Joint = -4.472, Marginal = -6.760, MI = -21.042\n",
      "Pre-Train Epoch 3, Batch 70: Loss = 1077.085, a = 117.332 b = 44.410, c = -3.175, Joint = -5.628, Marginal = -12.286, MI = -15.904\n",
      "Pre-Train Epoch 3, Batch 71: Loss = 3819.962, a = 181.731 b = 52.284, c = -0.377, Joint = -11.435, Marginal = -24.619, MI = -22.359\n",
      "Pre-Train Epoch 3, Batch 72: Loss = 1321.718, a = 208.297 b = 31.152, c = -0.376, Joint = -6.075, Marginal = -14.877, MI = -20.625\n",
      "Pre-Train Epoch 3, Batch 73: Loss = 970.190, a = 153.208 b = 28.332, c = -0.376, Joint = -2.917, Marginal = -11.085, MI = -17.174\n",
      "Pre-Train Epoch 3, Batch 74: Loss = 267.983, a = 148.006 b = 36.716, c = -0.950, Joint = -1.763, Marginal = -7.915, MI = -16.166\n",
      "Pre-Train Epoch 3, Batch 75: Loss = 375.870, a = 157.156 b = 31.253, c = -0.375, Joint = -0.173, Marginal = -3.841, MI = -14.428\n",
      "Pre-Train Epoch 3, Batch 76: Loss = 735.936, a = 128.150 b = 37.131, c = -0.375, Joint = -3.921, Marginal = -8.648, MI = -17.459\n",
      "Pre-Train Epoch 3, Batch 77: Loss = 2342.025, a = 152.270 b = 36.364, c = -0.375, Joint = -9.804, Marginal = -20.514, MI = -28.124\n",
      "Pre-Train Epoch 3, Batch 78: Loss = 278.420, a = 81.104 b = 44.899, c = -0.374, Joint = -4.180, Marginal = -4.638, MI = -16.373\n",
      "Pre-Train Epoch 3, Batch 79: Loss = 922.251, a = 138.564 b = 35.505, c = -0.374, Joint = -4.014, Marginal = -5.119, MI = -23.177\n",
      "Pre-Train Epoch 3, Batch 80: Loss = 745.093, a = 118.411 b = 42.259, c = -0.374, Joint = -4.616, Marginal = -6.445, MI = -16.665\n",
      "Pre-Train Epoch 3, Batch 81: Loss = 429.408, a = 127.471 b = 37.101, c = -6.436, Joint = -3.698, Marginal = -11.353, MI = -16.928\n",
      "Pre-Train Epoch 3, Batch 82: Loss = 1936.847, a = 172.139 b = 40.090, c = -5.716, Joint = -5.463, Marginal = -16.637, MI = -25.030\n",
      "Pre-Train Epoch 3, Batch 83: Loss = 234.802, a = 111.818 b = 35.563, c = -0.373, Joint = -1.611, Marginal = -6.324, MI = -13.329\n",
      "Pre-Train Epoch 3, Batch 84: Loss = 461.271, a = 143.405 b = 33.687, c = -0.372, Joint = -1.699, Marginal = -9.517, MI = -12.997\n",
      "Pre-Train Epoch 3, Batch 85: Loss = 3145.950, a = 150.783 b = 36.758, c = -0.372, Joint = -10.341, Marginal = -12.578, MI = -24.090\n",
      "Pre-Train Epoch 3, Batch 86: Loss = 230.453, a = 112.837 b = 35.917, c = -0.372, Joint = -1.714, Marginal = -4.637, MI = -15.086\n",
      "Pre-Train Epoch 3, Batch 87: Loss = 857.394, a = 127.163 b = 31.920, c = -0.371, Joint = -2.019, Marginal = -4.340, MI = -17.741\n",
      "Pre-Train Epoch 3, Batch 88: Loss = 199.331, a = 87.683 b = 39.269, c = -0.371, Joint = -0.725, Marginal = -2.247, MI = -15.922\n",
      "Pre-Train Epoch 3, Batch 89: Loss = 237.321, a = 115.983 b = 37.557, c = -2.079, Joint = 0.141, Marginal = -2.860, MI = -13.777\n",
      "Pre-Train Epoch 3, Batch 90: Loss = 470.307, a = 171.522 b = 35.943, c = -0.370, Joint = -2.724, Marginal = -11.224, MI = -12.552\n",
      "Pre-Train Epoch 3, Batch 91: Loss = 221.458, a = 91.786 b = 37.562, c = -0.370, Joint = -0.989, Marginal = -2.989, MI = -12.228\n",
      "Pre-Train Epoch 3, Batch 92: Loss = 554.679, a = 205.697 b = 29.443, c = -0.370, Joint = -3.688, Marginal = -14.334, MI = -21.327\n",
      "Pre-Train Epoch 3, Batch 93: Loss = 1885.290, a = 160.531 b = 47.996, c = -0.370, Joint = -10.516, Marginal = -9.607, MI = -27.434\n",
      "Pre-Train Epoch 3, Batch 94: Loss = 216.121, a = 116.812 b = 33.055, c = -0.369, Joint = 0.256, Marginal = -8.511, MI = -13.905\n",
      "Pre-Train Epoch 3, Batch 95: Loss = 1346.176, a = 159.157 b = 43.535, c = -0.369, Joint = -7.030, Marginal = -5.959, MI = -20.225\n",
      "Pre-Train Epoch 3, Batch 96: Loss = 517.736, a = 111.755 b = 43.834, c = -0.369, Joint = -3.780, Marginal = -4.789, MI = -20.338\n",
      "Pre-Train Epoch 3, Batch 97: Loss = 218.040, a = 108.587 b = 38.853, c = -0.368, Joint = -1.356, Marginal = -5.082, MI = -10.303\n",
      "Pre-Train Epoch 3, Batch 98: Loss = 676.430, a = 171.634 b = 40.480, c = -7.600, Joint = -3.819, Marginal = -10.675, MI = -14.375\n",
      "Pre-Train Epoch 3, Batch 99: Loss = 729.519, a = 156.564 b = 40.716, c = -0.738, Joint = -2.901, Marginal = -5.488, MI = -19.395\n",
      "Pre-Train Epoch 3, Batch 100: Loss = 216.382, a = 94.401 b = 38.094, c = -0.368, Joint = 0.299, Marginal = -3.790, MI = -12.879\n",
      "Pre-Train Epoch 3, Batch 101: Loss = 1856.394, a = 230.015 b = 39.464, c = -6.274, Joint = -8.868, Marginal = -28.712, MI = -23.749\n",
      "Pre-Train Epoch 3, Batch 102: Loss = 803.712, a = 138.146 b = 36.484, c = -0.367, Joint = -3.265, Marginal = -6.411, MI = -18.261\n",
      "Pre-Train Epoch 3, Batch 103: Loss = 212.957, a = 100.048 b = 37.786, c = -0.367, Joint = -1.369, Marginal = -4.985, MI = -13.768\n",
      "Pre-Train Epoch 3, Batch 104: Loss = 1124.440, a = 143.611 b = 35.958, c = -0.366, Joint = -6.538, Marginal = -13.108, MI = -19.565\n",
      "Pre-Train Epoch 3, Batch 105: Loss = 763.627, a = 127.860 b = 33.964, c = -0.366, Joint = -5.134, Marginal = -8.827, MI = -22.271\n",
      "Pre-Train Epoch 3, Batch 106: Loss = 1008.636, a = 170.103 b = 39.312, c = -0.366, Joint = -6.566, Marginal = -13.971, MI = -17.336\n",
      "Pre-Train Epoch 3, Batch 107: Loss = 284.170, a = 114.155 b = 37.082, c = -3.824, Joint = -0.810, Marginal = -4.068, MI = -12.929\n",
      "Pre-Train Epoch 3, Batch 108: Loss = 1531.217, a = 81.392 b = 44.133, c = -0.365, Joint = -7.071, Marginal = -7.444, MI = -20.509\n",
      "Pre-Train Epoch 3, Batch 109: Loss = 4185.688, a = 106.355 b = 33.390, c = -2.277, Joint = -10.041, Marginal = -16.076, MI = -20.919\n",
      "Pre-Train Epoch 3, Batch 110: Loss = 3348.199, a = 150.427 b = 38.754, c = -0.557, Joint = -8.011, Marginal = -12.246, MI = -26.799\n",
      "Pre-Train Epoch 3, Batch 111: Loss = 6735.398, a = 175.805 b = 44.341, c = -0.364, Joint = -18.166, Marginal = -27.154, MI = -31.711\n",
      "Pre-Train Epoch 3, Batch 112: Loss = 521.110, a = 119.444 b = 32.794, c = -0.753, Joint = -2.498, Marginal = -10.677, MI = -15.973\n",
      "Pre-Train Epoch 3, Batch 113: Loss = 474.309, a = 147.360 b = 43.689, c = -0.363, Joint = -4.663, Marginal = -12.921, MI = -21.653\n",
      "Pre-Train Epoch 3, Batch 114: Loss = 285.492, a = 81.393 b = 36.417, c = -0.363, Joint = -2.395, Marginal = -2.797, MI = -16.701\n",
      "Pre-Train Epoch 3, Batch 115: Loss = 2743.475, a = 142.142 b = 34.391, c = -0.363, Joint = -7.141, Marginal = -13.590, MI = -22.676\n",
      "Pre-Train Epoch 3, Batch 116: Loss = 202.414, a = 89.229 b = 41.966, c = -0.362, Joint = -1.838, Marginal = -6.810, MI = -17.236\n",
      "Pre-Train Epoch 3, Batch 117: Loss = 744.415, a = 126.404 b = 37.840, c = -0.362, Joint = -4.893, Marginal = -8.684, MI = -13.919\n",
      "Pre-Train Epoch 3, Batch 118: Loss = 283.559, a = 100.960 b = 44.568, c = -0.362, Joint = -2.394, Marginal = -7.256, MI = -13.229\n",
      "Pre-Train Epoch 3, Batch 119: Loss = 249.003, a = 119.827 b = 35.833, c = -0.361, Joint = -2.647, Marginal = -7.194, MI = -16.977\n",
      "Pre-Train Epoch 3, Batch 120: Loss = 159.046, a = 76.242 b = 31.497, c = -7.631, Joint = -1.014, Marginal = -2.053, MI = -12.040\n",
      "Pre-Train Epoch 3, Batch 121: Loss = 175.137, a = 85.840 b = 31.039, c = -0.388, Joint = 0.316, Marginal = -2.855, MI = -10.024\n",
      "Pre-Train Epoch 3, Batch 122: Loss = 242.955, a = 129.189 b = 35.209, c = -6.373, Joint = -0.867, Marginal = -6.820, MI = -17.991\n",
      "Pre-Train Epoch 3, Batch 123: Loss = 302.040, a = 142.836 b = 37.278, c = -1.928, Joint = 1.423, Marginal = -8.017, MI = -12.422\n",
      "Pre-Train Epoch 3, Batch 124: Loss = 233.442, a = 119.836 b = 30.165, c = -0.360, Joint = 1.463, Marginal = -2.133, MI = -14.147\n",
      "Pre-Train Epoch 3, Batch 125: Loss = 1786.327, a = 147.586 b = 30.542, c = -0.742, Joint = -8.013, Marginal = -17.036, MI = -23.980\n",
      "Pre-Train Epoch 3, Batch 126: Loss = 191.697, a = 85.989 b = 36.493, c = -0.359, Joint = -0.156, Marginal = -2.927, MI = -16.880\n",
      "Pre-Train Epoch 3, Batch 127: Loss = 238.826, a = 103.556 b = 44.292, c = -0.359, Joint = -2.400, Marginal = -7.037, MI = -13.793\n",
      "Pre-Train Epoch 3, Batch 128: Loss = 560.235, a = 94.246 b = 29.613, c = -1.387, Joint = -2.203, Marginal = -7.288, MI = -12.090\n",
      "Pre-Train Epoch 3, Batch 129: Loss = 249.153, a = 117.648 b = 36.722, c = -0.358, Joint = -2.086, Marginal = -4.599, MI = -11.699\n",
      "Pre-Train Epoch 3, Batch 130: Loss = 230.024, a = 98.303 b = 43.064, c = -0.358, Joint = -1.687, Marginal = -2.959, MI = -14.225\n",
      "Pre-Train Epoch 3, Batch 131: Loss = 297.608, a = 83.789 b = 36.497, c = -3.980, Joint = -1.621, Marginal = -1.895, MI = -14.849\n",
      "Pre-Train Epoch 3, Batch 132: Loss = 674.328, a = 90.351 b = 34.896, c = -0.358, Joint = -2.763, Marginal = -3.835, MI = -16.242\n",
      "Pre-Train Epoch 3, Batch 133: Loss = 4064.994, a = 134.178 b = 41.031, c = -0.357, Joint = -11.621, Marginal = -18.885, MI = -23.738\n",
      "Pre-Train Epoch 3, Batch 134: Loss = 553.425, a = 124.648 b = 26.673, c = -0.357, Joint = -0.931, Marginal = -6.670, MI = -18.214\n",
      "Pre-Train Epoch 3, Batch 135: Loss = 203.089, a = 98.793 b = 33.740, c = -0.357, Joint = 0.220, Marginal = -2.978, MI = -15.523\n",
      "Pre-Train Epoch 3, Batch 136: Loss = 255.481, a = 95.294 b = 49.155, c = -0.357, Joint = -2.768, Marginal = -4.050, MI = -13.674\n",
      "Pre-Train Epoch 3, Batch 137: Loss = 187.799, a = 92.649 b = 33.977, c = -0.356, Joint = -0.597, Marginal = -5.934, MI = -15.102\n",
      "Pre-Train Epoch 3, Batch 138: Loss = 428.107, a = 83.324 b = 34.339, c = -0.356, Joint = -3.145, Marginal = -3.481, MI = -15.083\n",
      "Pre-Train Epoch 3, Batch 139: Loss = 313.516, a = 105.268 b = 46.438, c = -2.776, Joint = -3.850, Marginal = -5.578, MI = -16.790\n",
      "Pre-Train Epoch 3, Batch 140: Loss = 6329.859, a = 108.695 b = 41.153, c = -3.776, Joint = -13.365, Marginal = -20.261, MI = -28.944\n",
      "Pre-Train Epoch 3, Batch 141: Loss = 294.505, a = 141.157 b = 39.084, c = -0.355, Joint = -0.388, Marginal = -3.791, MI = -17.564\n",
      "Pre-Train Epoch 3, Batch 142: Loss = 482.003, a = 126.483 b = 32.752, c = -0.355, Joint = -1.341, Marginal = -9.316, MI = -12.746\n",
      "Pre-Train Epoch 3, Batch 143: Loss = 2257.054, a = 130.957 b = 43.585, c = -0.354, Joint = -5.682, Marginal = -7.648, MI = -19.580\n",
      "Pre-Train Epoch 3, Batch 144: Loss = 373.354, a = 97.189 b = 42.626, c = -0.354, Joint = -3.047, Marginal = -6.862, MI = -12.606\n",
      "Pre-Train Epoch 3, Batch 145: Loss = 266.645, a = 101.216 b = 35.380, c = -0.354, Joint = -2.695, Marginal = -4.485, MI = -15.853\n",
      "Pre-Train Epoch 3, Batch 146: Loss = 202.535, a = 99.045 b = 38.226, c = -0.353, Joint = -0.916, Marginal = -5.704, MI = -13.210\n",
      "Pre-Train Epoch 3, Batch 147: Loss = 377.821, a = 141.493 b = 44.316, c = -0.353, Joint = -1.609, Marginal = -5.071, MI = -17.572\n",
      "Pre-Train Epoch 3, Batch 148: Loss = 349.435, a = 106.950 b = 35.518, c = -0.353, Joint = -2.508, Marginal = -2.851, MI = -16.480\n",
      "Pre-Train Epoch 3, Batch 149: Loss = 145.873, a = 66.837 b = 25.902, c = -4.924, Joint = -0.095, Marginal = -1.585, MI = -12.231\n",
      "Pre-Train Epoch 3, Batch 150: Loss = 287.440, a = 103.807 b = 37.807, c = -6.201, Joint = -1.897, Marginal = -7.936, MI = -15.058\n",
      "\n",
      "Start of epoch 4\n",
      "Pre-Train Epoch 4, Batch 0: Loss = 538.834, a = 98.915 b = 37.712, c = -0.352, Joint = -3.107, Marginal = -3.137, MI = -17.103\n",
      "Pre-Train Epoch 4, Batch 1: Loss = 6602.037, a = 143.058 b = 42.784, c = -0.352, Joint = -12.720, Marginal = -18.308, MI = -25.835\n",
      "Pre-Train Epoch 4, Batch 2: Loss = 231.886, a = 91.988 b = 42.609, c = -0.351, Joint = -1.009, Marginal = -6.054, MI = -11.394\n",
      "Pre-Train Epoch 4, Batch 3: Loss = 304.999, a = 77.020 b = 36.984, c = -0.351, Joint = -2.435, Marginal = -3.458, MI = -16.375\n",
      "Pre-Train Epoch 4, Batch 4: Loss = 182.104, a = 79.402 b = 29.987, c = -0.351, Joint = -0.034, Marginal = -3.700, MI = -13.648\n",
      "Pre-Train Epoch 4, Batch 5: Loss = 207.454, a = 96.117 b = 34.850, c = -0.350, Joint = -1.966, Marginal = -4.977, MI = -16.595\n",
      "Pre-Train Epoch 4, Batch 6: Loss = 255.238, a = 101.195 b = 36.732, c = -1.601, Joint = -0.087, Marginal = -4.456, MI = -13.343\n",
      "Pre-Train Epoch 4, Batch 7: Loss = 183.358, a = 67.851 b = 42.587, c = -0.350, Joint = -1.430, Marginal = -3.077, MI = -14.090\n",
      "Pre-Train Epoch 4, Batch 8: Loss = 201.642, a = 75.658 b = 41.389, c = -0.349, Joint = -1.364, Marginal = -3.378, MI = -13.863\n",
      "Pre-Train Epoch 4, Batch 9: Loss = 1403.659, a = 161.045 b = 35.357, c = -0.349, Joint = -8.559, Marginal = -16.553, MI = -21.820\n",
      "Pre-Train Epoch 4, Batch 10: Loss = 279.304, a = 97.918 b = 45.662, c = -6.538, Joint = -2.653, Marginal = -11.163, MI = -13.976\n",
      "Pre-Train Epoch 4, Batch 11: Loss = 179.845, a = 81.389 b = 36.160, c = -0.349, Joint = -0.618, Marginal = -4.277, MI = -15.784\n",
      "Pre-Train Epoch 4, Batch 12: Loss = 499.440, a = 108.312 b = 28.535, c = -0.348, Joint = -3.073, Marginal = -12.430, MI = -15.199\n",
      "Pre-Train Epoch 4, Batch 13: Loss = 243.082, a = 106.471 b = 33.398, c = -0.348, Joint = -0.025, Marginal = -3.785, MI = -10.861\n",
      "Pre-Train Epoch 4, Batch 14: Loss = 2191.303, a = 79.664 b = 30.367, c = -2.710, Joint = -6.032, Marginal = -10.507, MI = -17.062\n",
      "Pre-Train Epoch 4, Batch 15: Loss = 803.369, a = 92.465 b = 36.950, c = -0.348, Joint = -4.882, Marginal = -4.638, MI = -19.379\n",
      "Pre-Train Epoch 4, Batch 16: Loss = 2546.483, a = 100.727 b = 41.826, c = -0.347, Joint = -9.151, Marginal = -9.545, MI = -22.377\n",
      "Pre-Train Epoch 4, Batch 17: Loss = 1802.542, a = 89.102 b = 41.860, c = -1.210, Joint = -7.809, Marginal = -15.097, MI = -15.218\n",
      "Pre-Train Epoch 4, Batch 18: Loss = 259.817, a = 94.851 b = 31.009, c = -0.347, Joint = -2.550, Marginal = -4.804, MI = -18.656\n",
      "Pre-Train Epoch 4, Batch 19: Loss = 281.385, a = 107.934 b = 32.378, c = -1.953, Joint = -0.178, Marginal = -6.943, MI = -16.043\n",
      "Pre-Train Epoch 4, Batch 20: Loss = 142.076, a = 59.437 b = 31.925, c = -0.346, Joint = -0.355, Marginal = -2.822, MI = -14.747\n",
      "Pre-Train Epoch 4, Batch 21: Loss = 382.098, a = 120.987 b = 35.077, c = -0.346, Joint = -3.568, Marginal = -8.316, MI = -15.340\n",
      "Pre-Train Epoch 4, Batch 22: Loss = 165.081, a = 84.728 b = 29.867, c = -6.039, Joint = 0.014, Marginal = -3.069, MI = -11.453\n",
      "Pre-Train Epoch 4, Batch 23: Loss = 443.790, a = 101.142 b = 34.203, c = -0.345, Joint = -3.386, Marginal = -3.348, MI = -17.979\n",
      "Pre-Train Epoch 4, Batch 24: Loss = 160.226, a = 82.005 b = 30.136, c = -0.345, Joint = 0.324, Marginal = -4.360, MI = -13.477\n",
      "Pre-Train Epoch 4, Batch 25: Loss = 2138.799, a = 112.482 b = 38.854, c = -0.345, Joint = -6.768, Marginal = -9.361, MI = -19.520\n",
      "Pre-Train Epoch 4, Batch 26: Loss = 336.874, a = 117.602 b = 44.477, c = -0.344, Joint = -2.586, Marginal = -6.066, MI = -15.818\n",
      "Pre-Train Epoch 4, Batch 27: Loss = 293.433, a = 103.318 b = 43.831, c = -0.344, Joint = -3.593, Marginal = -9.292, MI = -14.057\n",
      "Pre-Train Epoch 4, Batch 28: Loss = 525.856, a = 106.201 b = 29.079, c = -2.511, Joint = -2.304, Marginal = -8.913, MI = -12.397\n",
      "Pre-Train Epoch 4, Batch 29: Loss = 186.513, a = 70.759 b = 40.171, c = -0.344, Joint = -1.773, Marginal = -3.699, MI = -11.463\n",
      "Pre-Train Epoch 4, Batch 30: Loss = 463.292, a = 93.511 b = 46.569, c = -0.343, Joint = -5.748, Marginal = -6.303, MI = -15.469\n",
      "Pre-Train Epoch 4, Batch 31: Loss = 960.062, a = 105.516 b = 42.167, c = -11.072, Joint = -8.167, Marginal = -16.182, MI = -17.317\n",
      "Pre-Train Epoch 4, Batch 32: Loss = 198.445, a = 69.783 b = 36.729, c = -0.343, Joint = -1.792, Marginal = -6.601, MI = -13.024\n",
      "Pre-Train Epoch 4, Batch 33: Loss = 2199.738, a = 90.586 b = 32.999, c = -0.343, Joint = -7.973, Marginal = -17.362, MI = -18.367\n",
      "Pre-Train Epoch 4, Batch 34: Loss = 205.482, a = 89.489 b = 37.664, c = -5.982, Joint = 0.873, Marginal = -3.726, MI = -12.478\n",
      "Pre-Train Epoch 4, Batch 35: Loss = 220.568, a = 89.774 b = 40.467, c = -0.342, Joint = -2.443, Marginal = -5.010, MI = -14.298\n",
      "Pre-Train Epoch 4, Batch 36: Loss = 170.020, a = 67.677 b = 40.228, c = -3.428, Joint = -1.561, Marginal = -3.112, MI = -14.004\n",
      "Pre-Train Epoch 4, Batch 37: Loss = 266.519, a = 89.710 b = 44.476, c = -0.341, Joint = -3.874, Marginal = -10.802, MI = -11.965\n",
      "Pre-Train Epoch 4, Batch 38: Loss = 129.450, a = 52.095 b = 34.502, c = -0.341, Joint = -0.871, Marginal = -2.180, MI = -12.131\n",
      "Pre-Train Epoch 4, Batch 39: Loss = 184.380, a = 78.210 b = 42.931, c = -0.405, Joint = -1.610, Marginal = -6.610, MI = -13.191\n",
      "Pre-Train Epoch 4, Batch 40: Loss = 28395.037, a = 155.140 b = 31.595, c = -0.341, Joint = -26.757, Marginal = -50.501, MI = -40.337\n",
      "Pre-Train Epoch 4, Batch 41: Loss = 225.114, a = 113.589 b = 40.270, c = -0.340, Joint = -1.470, Marginal = -7.658, MI = -10.488\n",
      "Pre-Train Epoch 4, Batch 42: Loss = 164.128, a = 71.645 b = 39.652, c = -0.535, Joint = -1.614, Marginal = -3.257, MI = -9.240\n",
      "Pre-Train Epoch 4, Batch 43: Loss = 329.181, a = 100.968 b = 42.375, c = -7.005, Joint = -4.019, Marginal = -5.777, MI = -13.078\n",
      "Pre-Train Epoch 4, Batch 44: Loss = 383.595, a = 110.416 b = 37.595, c = -0.339, Joint = -3.276, Marginal = -9.826, MI = -14.642\n",
      "Pre-Train Epoch 4, Batch 45: Loss = 241.807, a = 86.002 b = 37.322, c = -2.922, Joint = -1.545, Marginal = -3.726, MI = -11.786\n",
      "Pre-Train Epoch 4, Batch 46: Loss = 202.816, a = 96.312 b = 41.397, c = -0.338, Joint = -2.266, Marginal = -6.504, MI = -12.930\n",
      "Pre-Train Epoch 4, Batch 47: Loss = 6780.867, a = 108.832 b = 32.150, c = -0.337, Joint = -11.639, Marginal = -11.094, MI = -26.318\n",
      "Pre-Train Epoch 4, Batch 48: Loss = 568.475, a = 145.425 b = 37.100, c = -0.337, Joint = -5.016, Marginal = -23.662, MI = -16.734\n",
      "Pre-Train Epoch 4, Batch 49: Loss = 202.039, a = 81.911 b = 36.529, c = -0.336, Joint = -0.993, Marginal = -2.013, MI = -18.213\n",
      "Pre-Train Epoch 4, Batch 50: Loss = 355.471, a = 92.501 b = 30.333, c = -0.336, Joint = -3.552, Marginal = -13.623, MI = -15.087\n",
      "Pre-Train Epoch 4, Batch 51: Loss = 364.064, a = 123.126 b = 36.651, c = -1.874, Joint = -2.776, Marginal = -9.097, MI = -13.888\n",
      "Pre-Train Epoch 4, Batch 52: Loss = 153.603, a = 70.926 b = 31.811, c = -0.836, Joint = -0.181, Marginal = -1.335, MI = -12.576\n",
      "Pre-Train Epoch 4, Batch 53: Loss = 271.359, a = 112.884 b = 25.777, c = -3.296, Joint = 0.247, Marginal = -2.514, MI = -14.049\n",
      "Pre-Train Epoch 4, Batch 54: Loss = 231.723, a = 74.362 b = 41.785, c = -0.335, Joint = -2.748, Marginal = -6.884, MI = -12.516\n",
      "Pre-Train Epoch 4, Batch 55: Loss = 17743.971, a = 135.544 b = 38.809, c = -0.334, Joint = -18.821, Marginal = -40.396, MI = -28.207\n",
      "Pre-Train Epoch 4, Batch 56: Loss = 407.204, a = 93.139 b = 37.842, c = -2.947, Joint = -3.141, Marginal = -1.974, MI = -13.982\n",
      "Pre-Train Epoch 4, Batch 57: Loss = 170.987, a = 78.133 b = 35.972, c = -6.527, Joint = -0.448, Marginal = -5.025, MI = -10.829\n",
      "Pre-Train Epoch 4, Batch 58: Loss = 302.859, a = 85.131 b = 41.091, c = -4.701, Joint = -2.525, Marginal = -5.321, MI = -14.977\n",
      "Pre-Train Epoch 4, Batch 59: Loss = 190.645, a = 76.994 b = 33.061, c = -0.849, Joint = 0.198, Marginal = -1.802, MI = -10.984\n",
      "Pre-Train Epoch 4, Batch 60: Loss = 7176.386, a = 118.712 b = 35.753, c = -0.332, Joint = -13.490, Marginal = -17.311, MI = -25.411\n",
      "Pre-Train Epoch 4, Batch 61: Loss = 231.800, a = 104.327 b = 35.224, c = -0.332, Joint = -0.662, Marginal = -1.222, MI = -15.596\n",
      "Pre-Train Epoch 4, Batch 62: Loss = 1149.362, a = 123.377 b = 33.464, c = -5.695, Joint = -6.293, Marginal = -13.138, MI = -19.212\n",
      "Pre-Train Epoch 4, Batch 63: Loss = 459.586, a = 107.719 b = 31.542, c = -0.331, Joint = -2.829, Marginal = -6.628, MI = -12.100\n",
      "Pre-Train Epoch 4, Batch 64: Loss = 187.679, a = 77.696 b = 41.865, c = -0.331, Joint = -1.356, Marginal = -3.080, MI = -11.647\n",
      "Pre-Train Epoch 4, Batch 65: Loss = 3164.415, a = 161.812 b = 30.326, c = -1.556, Joint = -10.082, Marginal = -27.187, MI = -23.257\n",
      "Pre-Train Epoch 4, Batch 66: Loss = 186.287, a = 77.744 b = 44.436, c = -0.330, Joint = -2.197, Marginal = -4.387, MI = -15.218\n",
      "Pre-Train Epoch 4, Batch 67: Loss = 253.163, a = 107.187 b = 48.813, c = -0.330, Joint = -2.825, Marginal = -11.220, MI = -17.150\n",
      "Pre-Train Epoch 4, Batch 68: Loss = 370.647, a = 125.698 b = 32.381, c = -0.329, Joint = -0.637, Marginal = -15.385, MI = -15.086\n",
      "Pre-Train Epoch 4, Batch 69: Loss = 480.719, a = 104.873 b = 34.105, c = -0.329, Joint = -3.698, Marginal = -9.089, MI = -16.264\n",
      "Pre-Train Epoch 4, Batch 70: Loss = 799.196, a = 85.012 b = 43.351, c = -3.022, Joint = -4.981, Marginal = -6.986, MI = -13.081\n",
      "Pre-Train Epoch 4, Batch 71: Loss = 2737.897, a = 135.167 b = 50.708, c = -0.328, Joint = -9.554, Marginal = -20.646, MI = -15.242\n",
      "Pre-Train Epoch 4, Batch 72: Loss = 928.117, a = 155.559 b = 30.335, c = -0.328, Joint = -4.876, Marginal = -11.017, MI = -20.018\n",
      "Pre-Train Epoch 4, Batch 73: Loss = 667.548, a = 111.684 b = 27.673, c = -0.327, Joint = -2.545, Marginal = -8.636, MI = -16.194\n",
      "Pre-Train Epoch 4, Batch 74: Loss = 201.521, a = 108.198 b = 35.828, c = -0.856, Joint = -1.396, Marginal = -6.332, MI = -10.262\n",
      "Pre-Train Epoch 4, Batch 75: Loss = 275.302, a = 114.228 b = 30.629, c = -0.327, Joint = -0.129, Marginal = -2.055, MI = -14.011\n",
      "Pre-Train Epoch 4, Batch 76: Loss = 537.402, a = 93.219 b = 36.373, c = -0.326, Joint = -3.431, Marginal = -6.775, MI = -13.417\n",
      "Pre-Train Epoch 4, Batch 77: Loss = 1656.245, a = 113.077 b = 35.489, c = -0.326, Joint = -8.277, Marginal = -21.966, MI = -18.802\n",
      "Pre-Train Epoch 4, Batch 78: Loss = 203.856, a = 57.562 b = 43.917, c = -0.326, Joint = -3.684, Marginal = -3.110, MI = -12.731\n",
      "Pre-Train Epoch 4, Batch 79: Loss = 657.366, a = 99.964 b = 34.662, c = -0.326, Joint = -3.583, Marginal = -6.579, MI = -13.480\n",
      "Pre-Train Epoch 4, Batch 80: Loss = 556.697, a = 85.495 b = 41.257, c = -0.325, Joint = -4.028, Marginal = -3.091, MI = -19.566\n",
      "Pre-Train Epoch 4, Batch 81: Loss = 303.120, a = 93.168 b = 36.314, c = -6.152, Joint = -3.154, Marginal = -8.782, MI = -14.023\n",
      "Pre-Train Epoch 4, Batch 82: Loss = 1360.055, a = 126.674 b = 39.041, c = -5.418, Joint = -4.630, Marginal = -18.147, MI = -19.408\n",
      "Pre-Train Epoch 4, Batch 83: Loss = 173.330, a = 81.102 b = 34.816, c = -0.324, Joint = -1.376, Marginal = -9.651, MI = -9.711\n",
      "Pre-Train Epoch 4, Batch 84: Loss = 341.529, a = 105.208 b = 32.965, c = -0.324, Joint = -1.383, Marginal = -5.894, MI = -16.503\n",
      "Pre-Train Epoch 4, Batch 85: Loss = 2264.289, a = 111.252 b = 35.956, c = -0.324, Joint = -8.965, Marginal = -20.756, MI = -23.204\n",
      "Pre-Train Epoch 4, Batch 86: Loss = 176.185, a = 81.448 b = 35.147, c = -0.323, Joint = -1.496, Marginal = -3.524, MI = -13.021\n",
      "Pre-Train Epoch 4, Batch 87: Loss = 608.438, a = 91.936 b = 31.476, c = -0.323, Joint = -1.814, Marginal = -5.489, MI = -13.137\n",
      "Pre-Train Epoch 4, Batch 88: Loss = 151.418, a = 61.582 b = 38.513, c = -0.323, Joint = -0.779, Marginal = -1.227, MI = -11.561\n",
      "Pre-Train Epoch 4, Batch 89: Loss = 182.479, a = 82.838 b = 36.756, c = -1.909, Joint = -0.030, Marginal = -3.420, MI = -12.765\n",
      "Pre-Train Epoch 4, Batch 90: Loss = 327.447, a = 126.563 b = 35.067, c = -0.322, Joint = -2.124, Marginal = -7.881, MI = -13.684\n",
      "Pre-Train Epoch 4, Batch 91: Loss = 165.961, a = 64.697 b = 36.786, c = -0.322, Joint = -1.044, Marginal = -4.151, MI = -13.952\n",
      "Pre-Train Epoch 4, Batch 92: Loss = 385.710, a = 153.433 b = 28.797, c = -0.322, Joint = -2.817, Marginal = -12.017, MI = -19.918\n",
      "Pre-Train Epoch 4, Batch 93: Loss = 1357.787, a = 118.179 b = 46.674, c = -0.322, Joint = -8.974, Marginal = -7.662, MI = -19.846\n",
      "Pre-Train Epoch 4, Batch 94: Loss = 167.110, a = 84.689 b = 32.411, c = -0.321, Joint = 0.207, Marginal = -3.392, MI = -10.015\n",
      "Pre-Train Epoch 4, Batch 95: Loss = 976.459, a = 116.187 b = 42.343, c = -0.321, Joint = -6.016, Marginal = -10.800, MI = -15.674\n",
      "Pre-Train Epoch 4, Batch 96: Loss = 390.231, a = 80.700 b = 42.872, c = -0.321, Joint = -3.283, Marginal = -5.402, MI = -19.138\n",
      "Pre-Train Epoch 4, Batch 97: Loss = 167.637, a = 78.056 b = 38.048, c = -0.321, Joint = -1.179, Marginal = -2.519, MI = -13.379\n",
      "Pre-Train Epoch 4, Batch 98: Loss = 477.309, a = 125.797 b = 39.318, c = -7.283, Joint = -3.157, Marginal = -8.754, MI = -14.696\n",
      "Pre-Train Epoch 4, Batch 99: Loss = 529.586, a = 113.325 b = 39.566, c = -0.589, Joint = -2.502, Marginal = -3.416, MI = -15.249\n",
      "Pre-Train Epoch 4, Batch 100: Loss = 163.725, a = 66.664 b = 37.368, c = -0.320, Joint = 0.151, Marginal = -1.979, MI = -12.219\n",
      "Pre-Train Epoch 4, Batch 101: Loss = 1365.422, a = 172.302 b = 38.208, c = -5.969, Joint = -7.189, Marginal = -18.805, MI = -19.315\n",
      "Pre-Train Epoch 4, Batch 102: Loss = 607.322, a = 100.548 b = 35.527, c = -0.319, Joint = -2.792, Marginal = -8.087, MI = -13.955\n",
      "Pre-Train Epoch 4, Batch 103: Loss = 159.194, a = 71.967 b = 36.946, c = -0.319, Joint = -1.231, Marginal = -3.489, MI = -13.216\n",
      "Pre-Train Epoch 4, Batch 104: Loss = 810.327, a = 106.231 b = 35.100, c = -0.319, Joint = -5.460, Marginal = -12.653, MI = -16.948\n",
      "Pre-Train Epoch 4, Batch 105: Loss = 558.222, a = 93.751 b = 33.204, c = -0.318, Joint = -4.264, Marginal = -7.564, MI = -19.611\n",
      "Pre-Train Epoch 4, Batch 106: Loss = 720.648, a = 126.880 b = 38.313, c = -0.318, Joint = -5.368, Marginal = -18.616, MI = -15.373\n",
      "Pre-Train Epoch 4, Batch 107: Loss = 211.953, a = 82.203 b = 36.232, c = -3.661, Joint = -0.770, Marginal = -5.651, MI = -14.586\n",
      "Pre-Train Epoch 4, Batch 108: Loss = 1136.272, a = 59.206 b = 43.174, c = -0.318, Joint = -6.132, Marginal = -10.122, MI = -15.149\n",
      "Pre-Train Epoch 4, Batch 109: Loss = 3062.841, a = 78.658 b = 32.790, c = -2.128, Joint = -8.733, Marginal = -20.236, MI = -16.873\n",
      "Pre-Train Epoch 4, Batch 110: Loss = 2457.378, a = 111.545 b = 37.873, c = -0.420, Joint = -6.785, Marginal = -13.293, MI = -17.032\n",
      "Pre-Train Epoch 4, Batch 111: Loss = 5013.629, a = 131.589 b = 43.177, c = -0.317, Joint = -15.589, Marginal = -15.885, MI = -29.694\n",
      "Pre-Train Epoch 4, Batch 112: Loss = 373.777, a = 86.379 b = 32.105, c = -0.645, Joint = -2.165, Marginal = -5.638, MI = -15.530\n",
      "Pre-Train Epoch 4, Batch 113: Loss = 347.728, a = 108.247 b = 42.555, c = -0.316, Joint = -3.930, Marginal = -10.846, MI = -19.908\n",
      "Pre-Train Epoch 4, Batch 114: Loss = 214.879, a = 57.519 b = 35.790, c = -0.316, Joint = -2.265, Marginal = -1.185, MI = -16.403\n",
      "Pre-Train Epoch 4, Batch 115: Loss = 1978.138, a = 104.504 b = 33.662, c = -0.315, Joint = -6.197, Marginal = -15.700, MI = -19.931\n",
      "Pre-Train Epoch 4, Batch 116: Loss = 158.185, a = 64.594 b = 41.230, c = -0.315, Joint = -1.529, Marginal = -2.184, MI = -13.498\n",
      "Pre-Train Epoch 4, Batch 117: Loss = 565.639, a = 91.519 b = 36.876, c = -0.315, Joint = -4.229, Marginal = -8.968, MI = -15.472\n",
      "Pre-Train Epoch 4, Batch 118: Loss = 210.486, a = 72.464 b = 43.555, c = -0.314, Joint = -2.052, Marginal = -6.770, MI = -12.278\n",
      "Pre-Train Epoch 4, Batch 119: Loss = 184.289, a = 87.226 b = 34.986, c = -0.314, Joint = -2.135, Marginal = -10.402, MI = -10.752\n",
      "Pre-Train Epoch 4, Batch 120: Loss = 124.383, a = 54.634 b = 31.094, c = -7.368, Joint = -0.938, Marginal = -2.283, MI = -11.388\n",
      "Pre-Train Epoch 4, Batch 121: Loss = 132.449, a = 60.241 b = 30.489, c = -0.314, Joint = 0.089, Marginal = -2.183, MI = -10.680\n",
      "Pre-Train Epoch 4, Batch 122: Loss = 184.422, a = 93.788 b = 34.455, c = -6.083, Joint = -0.661, Marginal = -1.697, MI = -11.715\n",
      "Pre-Train Epoch 4, Batch 123: Loss = 228.684, a = 102.551 b = 36.260, c = -1.820, Joint = 1.163, Marginal = -2.469, MI = -10.183\n",
      "Pre-Train Epoch 4, Batch 124: Loss = 177.522, a = 85.507 b = 29.643, c = -0.313, Joint = 1.159, Marginal = -0.634, MI = -11.791\n",
      "Pre-Train Epoch 4, Batch 125: Loss = 1285.686, a = 108.895 b = 29.860, c = -0.729, Joint = -6.971, Marginal = -15.740, MI = -18.459\n",
      "Pre-Train Epoch 4, Batch 126: Loss = 147.112, a = 60.843 b = 35.847, c = -0.312, Joint = -0.254, Marginal = -2.573, MI = -10.441\n",
      "Pre-Train Epoch 4, Batch 127: Loss = 180.581, a = 74.873 b = 43.255, c = -0.312, Joint = -2.052, Marginal = -8.019, MI = -13.507\n",
      "Pre-Train Epoch 4, Batch 128: Loss = 413.587, a = 67.639 b = 29.111, c = -1.313, Joint = -2.064, Marginal = -2.615, MI = -14.518\n",
      "Pre-Train Epoch 4, Batch 129: Loss = 183.333, a = 85.616 b = 35.946, c = -0.312, Joint = -1.738, Marginal = -3.656, MI = -15.421\n",
      "Pre-Train Epoch 4, Batch 130: Loss = 177.964, a = 70.605 b = 42.163, c = -0.311, Joint = -1.421, Marginal = -2.049, MI = -13.152\n",
      "Pre-Train Epoch 4, Batch 131: Loss = 223.312, a = 59.271 b = 35.888, c = -3.835, Joint = -1.584, Marginal = -0.916, MI = -17.493\n",
      "Pre-Train Epoch 4, Batch 132: Loss = 495.566, a = 64.959 b = 34.366, c = -0.311, Joint = -2.512, Marginal = -4.690, MI = -11.692\n",
      "Pre-Train Epoch 4, Batch 133: Loss = 2990.897, a = 99.983 b = 40.048, c = -0.311, Joint = -10.034, Marginal = -14.864, MI = -19.596\n",
      "Pre-Train Epoch 4, Batch 134: Loss = 404.198, a = 91.173 b = 26.282, c = -0.310, Joint = -0.802, Marginal = -6.963, MI = -11.968\n",
      "Pre-Train Epoch 4, Batch 135: Loss = 157.158, a = 71.229 b = 33.131, c = -0.310, Joint = 0.120, Marginal = -1.421, MI = -11.535\n",
      "Pre-Train Epoch 4, Batch 136: Loss = 190.675, a = 67.799 b = 47.961, c = -0.310, Joint = -2.520, Marginal = -3.821, MI = -14.216\n",
      "Pre-Train Epoch 4, Batch 137: Loss = 143.682, a = 66.930 b = 33.327, c = -0.310, Joint = -0.527, Marginal = -2.642, MI = -12.165\n",
      "Pre-Train Epoch 4, Batch 138: Loss = 315.711, a = 59.838 b = 33.769, c = -0.309, Joint = -2.860, Marginal = -3.484, MI = -12.231\n",
      "Pre-Train Epoch 4, Batch 139: Loss = 231.548, a = 76.642 b = 45.325, c = -2.626, Joint = -3.214, Marginal = -6.248, MI = -12.793\n",
      "Pre-Train Epoch 4, Batch 140: Loss = 4733.498, a = 80.479 b = 40.248, c = -3.577, Joint = -11.714, Marginal = -10.956, MI = -21.421\n",
      "Pre-Train Epoch 4, Batch 141: Loss = 220.882, a = 102.908 b = 37.987, c = -0.309, Joint = -0.260, Marginal = -6.138, MI = -8.568\n",
      "Pre-Train Epoch 4, Batch 142: Loss = 347.698, a = 93.163 b = 32.098, c = -0.308, Joint = -1.035, Marginal = -3.746, MI = -13.840\n",
      "Pre-Train Epoch 4, Batch 143: Loss = 1701.701, a = 95.077 b = 42.425, c = -0.308, Joint = -5.011, Marginal = -6.700, MI = -14.787\n",
      "Pre-Train Epoch 4, Batch 144: Loss = 282.292, a = 70.454 b = 41.751, c = -0.308, Joint = -2.702, Marginal = -4.019, MI = -15.435\n",
      "Pre-Train Epoch 4, Batch 145: Loss = 195.198, a = 73.771 b = 34.579, c = -0.307, Joint = -2.222, Marginal = -6.880, MI = -12.134\n",
      "Pre-Train Epoch 4, Batch 146: Loss = 155.836, a = 71.587 b = 37.419, c = -0.307, Joint = -0.767, Marginal = -4.432, MI = -11.887\n",
      "Pre-Train Epoch 4, Batch 147: Loss = 274.259, a = 103.331 b = 43.041, c = -0.307, Joint = -1.304, Marginal = -6.831, MI = -12.585\n",
      "Pre-Train Epoch 4, Batch 148: Loss = 258.722, a = 77.364 b = 34.707, c = -0.306, Joint = -2.190, Marginal = -2.729, MI = -12.276\n",
      "Pre-Train Epoch 4, Batch 149: Loss = 112.543, a = 47.712 b = 25.683, c = -4.735, Joint = -0.218, Marginal = -1.095, MI = -11.059\n",
      "Pre-Train Epoch 4, Batch 150: Loss = 208.960, a = 75.625 b = 37.033, c = -6.021, Joint = -1.633, Marginal = -6.668, MI = -13.092\n",
      "\n",
      "Start of epoch 5\n",
      "Pre-Train Epoch 5, Batch 0: Loss = 401.143, a = 71.619 b = 36.833, c = -0.306, Joint = -2.717, Marginal = -2.322, MI = -11.628\n",
      "Pre-Train Epoch 5, Batch 1: Loss = 4908.162, a = 106.727 b = 41.598, c = -0.305, Joint = -10.897, Marginal = -18.627, MI = -21.038\n",
      "Pre-Train Epoch 5, Batch 2: Loss = 176.763, a = 65.910 b = 41.672, c = -0.305, Joint = -0.934, Marginal = -2.025, MI = -15.084\n",
      "Pre-Train Epoch 5, Batch 3: Loss = 228.711, a = 55.666 b = 36.324, c = -0.305, Joint = -2.170, Marginal = -2.577, MI = -12.106\n",
      "Pre-Train Epoch 5, Batch 4: Loss = 137.564, a = 57.393 b = 29.586, c = -0.304, Joint = -0.073, Marginal = -3.955, MI = -13.367\n",
      "Pre-Train Epoch 5, Batch 5: Loss = 156.375, a = 70.305 b = 34.195, c = -0.304, Joint = -1.647, Marginal = -5.933, MI = -11.197\n",
      "Pre-Train Epoch 5, Batch 6: Loss = 190.297, a = 73.052 b = 35.987, c = -1.363, Joint = -0.099, Marginal = -3.505, MI = -11.591\n",
      "Pre-Train Epoch 5, Batch 7: Loss = 142.851, a = 48.316 b = 41.794, c = -0.304, Joint = -1.301, Marginal = -2.325, MI = -9.871\n",
      "Pre-Train Epoch 5, Batch 8: Loss = 155.484, a = 54.114 b = 40.535, c = -0.303, Joint = -1.264, Marginal = -2.309, MI = -12.159\n",
      "Pre-Train Epoch 5, Batch 9: Loss = 1001.939, a = 121.883 b = 34.441, c = -0.303, Joint = -7.149, Marginal = -18.106, MI = -14.853\n",
      "Pre-Train Epoch 5, Batch 10: Loss = 206.733, a = 71.528 b = 44.524, c = -6.228, Joint = -2.269, Marginal = -7.218, MI = -10.058\n",
      "Pre-Train Epoch 5, Batch 11: Loss = 142.120, a = 59.365 b = 35.522, c = -0.303, Joint = -0.534, Marginal = -8.820, MI = -12.266\n",
      "Pre-Train Epoch 5, Batch 12: Loss = 355.016, a = 80.350 b = 28.094, c = -0.302, Joint = -2.570, Marginal = -6.929, MI = -13.946\n",
      "Pre-Train Epoch 5, Batch 13: Loss = 181.677, a = 77.058 b = 32.668, c = -0.302, Joint = -0.022, Marginal = -0.842, MI = -11.453\n",
      "Pre-Train Epoch 5, Batch 14: Loss = 1615.865, a = 59.113 b = 29.911, c = -2.549, Joint = -5.273, Marginal = -11.444, MI = -15.475\n",
      "Pre-Train Epoch 5, Batch 15: Loss = 605.093, a = 67.347 b = 36.253, c = -0.302, Joint = -4.312, Marginal = -4.459, MI = -14.474\n",
      "Pre-Train Epoch 5, Batch 16: Loss = 1953.339, a = 74.367 b = 40.899, c = -0.301, Joint = -8.029, Marginal = -8.664, MI = -15.470\n",
      "Pre-Train Epoch 5, Batch 17: Loss = 1312.786, a = 66.596 b = 41.087, c = -1.130, Joint = -6.736, Marginal = -9.862, MI = -14.890\n",
      "Pre-Train Epoch 5, Batch 18: Loss = 191.614, a = 71.127 b = 30.517, c = -0.301, Joint = -1.972, Marginal = -9.219, MI = -11.589\n",
      "Pre-Train Epoch 5, Batch 19: Loss = 206.473, a = 78.522 b = 31.682, c = -1.823, Joint = -0.235, Marginal = -5.152, MI = -9.840\n",
      "Pre-Train Epoch 5, Batch 20: Loss = 112.155, a = 42.843 b = 31.502, c = -0.300, Joint = -0.364, Marginal = -1.605, MI = -10.112\n",
      "Pre-Train Epoch 5, Batch 21: Loss = 278.849, a = 90.522 b = 34.269, c = -0.300, Joint = -2.929, Marginal = -13.346, MI = -13.712\n",
      "Pre-Train Epoch 5, Batch 22: Loss = 129.898, a = 61.871 b = 29.421, c = -5.769, Joint = -0.019, Marginal = -1.950, MI = -13.167\n",
      "Pre-Train Epoch 5, Batch 23: Loss = 329.090, a = 74.143 b = 33.532, c = -0.300, Joint = -2.999, Marginal = -7.758, MI = -15.441\n",
      "Pre-Train Epoch 5, Batch 24: Loss = 127.544, a = 59.713 b = 29.696, c = -0.299, Joint = 0.235, Marginal = -1.430, MI = -8.955\n",
      "Pre-Train Epoch 5, Batch 25: Loss = 1627.062, a = 83.109 b = 37.929, c = -0.299, Joint = -5.874, Marginal = -10.092, MI = -14.799\n",
      "Pre-Train Epoch 5, Batch 26: Loss = 251.039, a = 85.752 b = 43.283, c = -0.299, Joint = -2.149, Marginal = -3.208, MI = -13.002\n",
      "Pre-Train Epoch 5, Batch 27: Loss = 217.885, a = 76.262 b = 42.792, c = -0.299, Joint = -2.978, Marginal = -7.852, MI = -11.671\n",
      "Pre-Train Epoch 5, Batch 28: Loss = 389.281, a = 78.944 b = 28.529, c = -2.467, Joint = -1.879, Marginal = -5.132, MI = -14.205\n",
      "Pre-Train Epoch 5, Batch 29: Loss = 143.336, a = 50.626 b = 39.353, c = -0.298, Joint = -1.610, Marginal = -3.459, MI = -8.307\n",
      "Pre-Train Epoch 5, Batch 30: Loss = 342.628, a = 69.042 b = 45.514, c = -0.298, Joint = -4.829, Marginal = -7.596, MI = -12.076\n",
      "Pre-Train Epoch 5, Batch 31: Loss = 691.131, a = 79.180 b = 41.150, c = -10.684, Joint = -6.772, Marginal = -11.439, MI = -14.808\n",
      "Pre-Train Epoch 5, Batch 32: Loss = 147.238, a = 50.762 b = 36.045, c = -0.297, Joint = -1.599, Marginal = -5.411, MI = -10.797\n",
      "Pre-Train Epoch 5, Batch 33: Loss = 1623.051, a = 67.892 b = 32.358, c = -0.297, Joint = -6.905, Marginal = -12.083, MI = -15.390\n",
      "Pre-Train Epoch 5, Batch 34: Loss = 157.505, a = 63.817 b = 36.831, c = -5.738, Joint = 0.560, Marginal = -4.546, MI = -11.228\n",
      "Pre-Train Epoch 5, Batch 35: Loss = 167.582, a = 65.675 b = 39.560, c = -0.297, Joint = -2.031, Marginal = -3.233, MI = -13.560\n",
      "Pre-Train Epoch 5, Batch 36: Loss = 133.271, a = 48.865 b = 39.488, c = -3.241, Joint = -1.343, Marginal = -1.186, MI = -11.209\n",
      "Pre-Train Epoch 5, Batch 37: Loss = 200.359, a = 66.279 b = 43.499, c = -0.296, Joint = -3.205, Marginal = -9.127, MI = -13.034\n",
      "Pre-Train Epoch 5, Batch 38: Loss = 102.029, a = 36.958 b = 34.057, c = -0.296, Joint = -0.859, Marginal = -1.918, MI = -12.588\n",
      "Pre-Train Epoch 5, Batch 39: Loss = 143.149, a = 56.347 b = 42.018, c = -0.296, Joint = -1.411, Marginal = -4.230, MI = -8.408\n",
      "Pre-Train Epoch 5, Batch 40: Loss = 21706.248, a = 120.477 b = 30.695, c = -0.295, Joint = -23.258, Marginal = -52.895, MI = -35.313\n",
      "Pre-Train Epoch 5, Batch 41: Loss = 171.153, a = 84.206 b = 39.320, c = -0.295, Joint = -1.152, Marginal = -6.037, MI = -11.829\n",
      "Pre-Train Epoch 5, Batch 42: Loss = 128.262, a = 51.913 b = 38.859, c = -0.439, Joint = -1.394, Marginal = -3.759, MI = -10.513\n",
      "Pre-Train Epoch 5, Batch 43: Loss = 241.125, a = 74.635 b = 41.368, c = -6.718, Joint = -3.336, Marginal = -10.103, MI = -11.031\n",
      "Pre-Train Epoch 5, Batch 44: Loss = 274.009, a = 81.712 b = 36.723, c = -0.294, Joint = -2.760, Marginal = -8.456, MI = -15.024\n",
      "Pre-Train Epoch 5, Batch 45: Loss = 185.143, a = 62.147 b = 36.574, c = -2.782, Joint = -1.345, Marginal = -1.381, MI = -12.082\n",
      "Pre-Train Epoch 5, Batch 46: Loss = 155.528, a = 69.979 b = 40.457, c = -0.293, Joint = -1.887, Marginal = -7.037, MI = -13.240\n",
      "Pre-Train Epoch 5, Batch 47: Loss = 5240.917, a = 81.276 b = 31.485, c = -0.292, Joint = -10.209, Marginal = -9.110, MI = -22.126\n",
      "Pre-Train Epoch 5, Batch 48: Loss = 401.866, a = 108.406 b = 36.135, c = -0.292, Joint = -4.071, Marginal = -10.528, MI = -15.138\n",
      "Pre-Train Epoch 5, Batch 49: Loss = 158.657, a = 59.396 b = 35.889, c = -0.292, Joint = -0.891, Marginal = -3.559, MI = -11.543\n",
      "Pre-Train Epoch 5, Batch 50: Loss = 256.333, a = 67.700 b = 29.743, c = -0.291, Joint = -3.136, Marginal = -13.997, MI = -9.314\n",
      "Pre-Train Epoch 5, Batch 51: Loss = 264.034, a = 90.685 b = 35.787, c = -1.777, Joint = -2.258, Marginal = -5.772, MI = -13.875\n",
      "Pre-Train Epoch 5, Batch 52: Loss = 117.820, a = 50.034 b = 31.250, c = -0.734, Joint = -0.286, Marginal = -0.642, MI = -8.494\n",
      "Pre-Train Epoch 5, Batch 53: Loss = 200.782, a = 82.385 b = 25.333, c = -3.113, Joint = 0.283, Marginal = 0.078, MI = -12.006\n",
      "Pre-Train Epoch 5, Batch 54: Loss = 173.674, a = 52.860 b = 40.989, c = -0.290, Joint = -2.477, Marginal = -3.876, MI = -14.668\n",
      "Pre-Train Epoch 5, Batch 55: Loss = 13566.694, a = 101.863 b = 37.863, c = -0.290, Joint = -16.471, Marginal = -36.391, MI = -24.397\n",
      "Pre-Train Epoch 5, Batch 56: Loss = 301.477, a = 67.039 b = 37.105, c = -2.790, Joint = -2.796, Marginal = -4.190, MI = -12.778\n",
      "Pre-Train Epoch 5, Batch 57: Loss = 132.819, a = 55.703 b = 35.356, c = -6.247, Joint = -0.532, Marginal = -1.264, MI = -11.639\n",
      "Pre-Train Epoch 5, Batch 58: Loss = 227.781, a = 60.761 b = 40.148, c = -4.538, Joint = -2.225, Marginal = -4.588, MI = -11.546\n",
      "Pre-Train Epoch 5, Batch 59: Loss = 144.268, a = 54.166 b = 32.511, c = -0.714, Joint = -0.015, Marginal = -1.752, MI = -9.982\n",
      "Pre-Train Epoch 5, Batch 60: Loss = 5387.115, a = 89.116 b = 35.054, c = -0.288, Joint = -11.781, Marginal = -23.858, MI = -22.326\n",
      "Pre-Train Epoch 5, Batch 61: Loss = 175.577, a = 75.419 b = 34.473, c = -0.287, Joint = -0.563, Marginal = -1.986, MI = -11.840\n",
      "Pre-Train Epoch 5, Batch 62: Loss = 843.432, a = 91.094 b = 32.784, c = -5.465, Joint = -5.440, Marginal = -10.529, MI = -16.730\n",
      "Pre-Train Epoch 5, Batch 63: Loss = 327.454, a = 79.177 b = 30.952, c = -0.287, Joint = -2.434, Marginal = -10.221, MI = -11.547\n",
      "Pre-Train Epoch 5, Batch 64: Loss = 143.600, a = 55.095 b = 41.035, c = -0.286, Joint = -1.282, Marginal = -7.353, MI = -9.753\n",
      "Pre-Train Epoch 5, Batch 65: Loss = 2397.130, a = 122.231 b = 29.577, c = -1.438, Joint = -8.560, Marginal = -20.429, MI = -18.213\n",
      "Pre-Train Epoch 5, Batch 66: Loss = 144.084, a = 55.176 b = 43.538, c = -0.286, Joint = -1.974, Marginal = -6.154, MI = -11.430\n",
      "Pre-Train Epoch 5, Batch 67: Loss = 189.098, a = 77.363 b = 47.488, c = -0.285, Joint = -2.436, Marginal = -7.849, MI = -10.433\n",
      "Pre-Train Epoch 5, Batch 68: Loss = 267.094, a = 91.925 b = 31.672, c = -0.285, Joint = -0.583, Marginal = -9.378, MI = -12.094\n",
      "Pre-Train Epoch 5, Batch 69: Loss = 351.441, a = 77.212 b = 33.455, c = -0.285, Joint = -3.133, Marginal = -6.424, MI = -13.722\n",
      "Pre-Train Epoch 5, Batch 70: Loss = 611.720, a = 61.348 b = 42.476, c = -2.890, Joint = -4.488, Marginal = -8.394, MI = -15.322\n",
      "Pre-Train Epoch 5, Batch 71: Loss = 2029.216, a = 100.416 b = 49.401, c = -0.284, Joint = -8.134, Marginal = -13.880, MI = -17.978\n",
      "Pre-Train Epoch 5, Batch 72: Loss = 674.904, a = 116.403 b = 29.679, c = -0.284, Joint = -4.041, Marginal = -8.894, MI = -15.276\n",
      "Pre-Train Epoch 5, Batch 73: Loss = 475.252, a = 81.527 b = 27.146, c = -0.283, Joint = -2.310, Marginal = -7.163, MI = -14.663\n",
      "Pre-Train Epoch 5, Batch 74: Loss = 155.898, a = 79.146 b = 35.103, c = -0.773, Joint = -1.142, Marginal = -7.692, MI = -11.373\n",
      "Pre-Train Epoch 5, Batch 75: Loss = 206.448, a = 83.120 b = 30.131, c = -0.283, Joint = -0.151, Marginal = -2.491, MI = -13.491\n",
      "Pre-Train Epoch 5, Batch 76: Loss = 404.904, a = 67.947 b = 35.753, c = -0.283, Joint = -3.057, Marginal = -8.518, MI = -12.889\n",
      "Pre-Train Epoch 5, Batch 77: Loss = 1212.531, a = 84.422 b = 34.773, c = -0.282, Joint = -7.092, Marginal = -18.890, MI = -14.240\n",
      "Pre-Train Epoch 5, Batch 78: Loss = 154.805, a = 40.702 b = 43.098, c = -0.282, Joint = -3.265, Marginal = -6.372, MI = -10.971\n",
      "Pre-Train Epoch 5, Batch 79: Loss = 483.819, a = 72.245 b = 33.974, c = -0.282, Joint = -3.265, Marginal = -3.736, MI = -18.288\n",
      "Pre-Train Epoch 5, Batch 80: Loss = 428.480, a = 61.676 b = 40.429, c = -0.281, Joint = -3.569, Marginal = -2.813, MI = -16.450\n",
      "Pre-Train Epoch 5, Batch 81: Loss = 222.517, a = 68.542 b = 35.668, c = -5.909, Joint = -2.729, Marginal = -5.460, MI = -12.654\n",
      "Pre-Train Epoch 5, Batch 82: Loss = 987.005, a = 93.596 b = 38.182, c = -5.163, Joint = -4.008, Marginal = -21.775, MI = -16.782\n",
      "Pre-Train Epoch 5, Batch 83: Loss = 132.536, a = 59.050 b = 34.203, c = -0.281, Joint = -1.194, Marginal = -13.604, MI = -8.422\n",
      "Pre-Train Epoch 5, Batch 84: Loss = 260.257, a = 77.574 b = 32.379, c = -0.280, Joint = -1.165, Marginal = -5.218, MI = -11.637\n",
      "Pre-Train Epoch 5, Batch 85: Loss = 1684.552, a = 82.710 b = 35.302, c = -0.280, Joint = -7.911, Marginal = -14.654, MI = -12.976\n",
      "Pre-Train Epoch 5, Batch 86: Loss = 138.256, a = 59.021 b = 34.514, c = -0.280, Joint = -1.318, Marginal = -4.400, MI = -7.639\n",
      "Pre-Train Epoch 5, Batch 87: Loss = 445.792, a = 66.764 b = 31.126, c = -0.280, Joint = -1.679, Marginal = -2.783, MI = -14.983\n",
      "Pre-Train Epoch 5, Batch 88: Loss = 117.921, a = 43.237 b = 37.887, c = -0.279, Joint = -0.801, Marginal = -1.509, MI = -10.007\n",
      "Pre-Train Epoch 5, Batch 89: Loss = 143.064, a = 59.374 b = 36.099, c = -1.763, Joint = -0.156, Marginal = -3.298, MI = -10.047\n",
      "Pre-Train Epoch 5, Batch 90: Loss = 237.182, a = 93.859 b = 34.354, c = -0.279, Joint = -1.712, Marginal = -8.072, MI = -11.689\n",
      "Pre-Train Epoch 5, Batch 91: Loss = 127.654, a = 45.521 b = 36.146, c = -0.279, Joint = -1.084, Marginal = -2.200, MI = -12.718\n",
      "Pre-Train Epoch 5, Batch 92: Loss = 279.101, a = 115.267 b = 28.286, c = -0.278, Joint = -2.231, Marginal = -10.210, MI = -15.281\n",
      "Pre-Train Epoch 5, Batch 93: Loss = 1009.857, a = 87.333 b = 45.577, c = -0.278, Joint = -7.771, Marginal = -15.543, MI = -16.848\n",
      "Pre-Train Epoch 5, Batch 94: Loss = 131.658, a = 61.557 b = 31.888, c = -0.278, Joint = 0.144, Marginal = -2.551, MI = -12.079\n",
      "Pre-Train Epoch 5, Batch 95: Loss = 730.917, a = 85.022 b = 41.358, c = -0.278, Joint = -5.270, Marginal = -9.937, MI = -17.825\n",
      "Pre-Train Epoch 5, Batch 96: Loss = 302.486, a = 58.345 b = 42.073, c = -0.277, Joint = -2.885, Marginal = -2.785, MI = -11.538\n",
      "Pre-Train Epoch 5, Batch 97: Loss = 131.951, a = 56.204 b = 37.385, c = -0.277, Joint = -1.046, Marginal = -4.142, MI = -11.833\n",
      "Pre-Train Epoch 5, Batch 98: Loss = 348.403, a = 92.513 b = 38.362, c = -7.011, Joint = -2.682, Marginal = -9.355, MI = -10.825\n",
      "Pre-Train Epoch 5, Batch 99: Loss = 395.820, a = 82.037 b = 38.618, c = -0.461, Joint = -2.241, Marginal = -2.964, MI = -12.333\n",
      "Pre-Train Epoch 5, Batch 100: Loss = 126.296, a = 46.898 b = 36.770, c = -0.277, Joint = 0.017, Marginal = -1.572, MI = -10.878\n",
      "Pre-Train Epoch 5, Batch 101: Loss = 1038.544, a = 129.668 b = 37.179, c = -5.707, Joint = -5.969, Marginal = -19.106, MI = -16.015\n",
      "Pre-Train Epoch 5, Batch 102: Loss = 471.354, a = 73.211 b = 34.740, c = -0.276, Joint = -2.465, Marginal = -3.833, MI = -13.199\n",
      "Pre-Train Epoch 5, Batch 103: Loss = 122.709, a = 51.822 b = 36.251, c = -0.276, Joint = -1.120, Marginal = -4.190, MI = -9.199\n",
      "Pre-Train Epoch 5, Batch 104: Loss = 602.721, a = 78.799 b = 34.396, c = -0.276, Joint = -4.661, Marginal = -7.377, MI = -15.682\n",
      "Pre-Train Epoch 5, Batch 105: Loss = 421.299, a = 68.886 b = 32.581, c = -0.275, Joint = -3.617, Marginal = -5.846, MI = -11.267\n",
      "Pre-Train Epoch 5, Batch 106: Loss = 533.369, a = 94.984 b = 37.494, c = -0.275, Joint = -4.488, Marginal = -10.468, MI = -11.641\n",
      "Pre-Train Epoch 5, Batch 107: Loss = 162.200, a = 59.071 b = 35.533, c = -3.521, Joint = -0.788, Marginal = -4.448, MI = -9.378\n",
      "Pre-Train Epoch 5, Batch 108: Loss = 869.106, a = 42.979 b = 42.373, c = -0.275, Joint = -5.382, Marginal = -3.987, MI = -15.237\n",
      "Pre-Train Epoch 5, Batch 109: Loss = 2311.170, a = 58.448 b = 32.300, c = -1.999, Joint = -7.728, Marginal = -20.439, MI = -14.312\n",
      "Pre-Train Epoch 5, Batch 110: Loss = 1858.464, a = 82.833 b = 37.151, c = -0.301, Joint = -5.875, Marginal = -11.452, MI = -16.589\n",
      "Pre-Train Epoch 5, Batch 111: Loss = 3843.716, a = 98.909 b = 42.216, c = -0.274, Joint = -13.592, Marginal = -16.472, MI = -24.511\n",
      "Pre-Train Epoch 5, Batch 112: Loss = 277.059, a = 62.606 b = 31.544, c = -0.551, Joint = -1.951, Marginal = -9.909, MI = -11.453\n",
      "Pre-Train Epoch 5, Batch 113: Loss = 263.494, a = 79.761 b = 41.618, c = -0.273, Joint = -3.355, Marginal = -11.345, MI = -15.494\n",
      "Pre-Train Epoch 5, Batch 114: Loss = 166.729, a = 40.668 b = 35.274, c = -0.273, Joint = -2.144, Marginal = -1.283, MI = -12.678\n",
      "Pre-Train Epoch 5, Batch 115: Loss = 1470.972, a = 77.451 b = 33.068, c = -0.273, Joint = -5.456, Marginal = -14.092, MI = -16.399\n",
      "Pre-Train Epoch 5, Batch 116: Loss = 126.400, a = 46.856 b = 40.620, c = -0.272, Joint = -1.278, Marginal = -3.735, MI = -10.629\n",
      "Pre-Train Epoch 5, Batch 117: Loss = 441.991, a = 66.363 b = 36.078, c = -0.272, Joint = -3.727, Marginal = -4.556, MI = -15.756\n",
      "Pre-Train Epoch 5, Batch 118: Loss = 161.120, a = 52.019 b = 42.710, c = -0.272, Joint = -1.788, Marginal = -3.779, MI = -11.170\n",
      "Pre-Train Epoch 5, Batch 119: Loss = 141.272, a = 63.873 b = 34.286, c = -0.271, Joint = -1.744, Marginal = -9.183, MI = -10.912\n",
      "Pre-Train Epoch 5, Batch 120: Loss = 99.644, a = 39.436 b = 30.766, c = -7.142, Joint = -0.842, Marginal = -0.929, MI = -10.648\n",
      "Pre-Train Epoch 5, Batch 121: Loss = 102.731, a = 42.330 b = 30.039, c = -0.271, Joint = -0.097, Marginal = -2.093, MI = -8.798\n",
      "Pre-Train Epoch 5, Batch 122: Loss = 143.531, a = 68.443 b = 33.837, c = -5.834, Joint = -0.529, Marginal = -3.749, MI = -9.914\n",
      "Pre-Train Epoch 5, Batch 123: Loss = 176.183, a = 73.741 b = 35.421, c = -1.725, Joint = 0.904, Marginal = -2.710, MI = -9.689\n",
      "Pre-Train Epoch 5, Batch 124: Loss = 137.492, a = 61.339 b = 29.224, c = -0.270, Joint = 0.908, Marginal = -0.475, MI = -9.119\n",
      "Pre-Train Epoch 5, Batch 125: Loss = 955.706, a = 81.097 b = 29.306, c = -0.714, Joint = -6.178, Marginal = -6.561, MI = -13.682\n",
      "Pre-Train Epoch 5, Batch 126: Loss = 115.321, a = 43.061 b = 35.315, c = -0.270, Joint = -0.333, Marginal = -2.329, MI = -12.777\n",
      "Pre-Train Epoch 5, Batch 127: Loss = 140.703, a = 54.145 b = 42.389, c = -0.269, Joint = -1.795, Marginal = -5.714, MI = -10.502\n",
      "Pre-Train Epoch 5, Batch 128: Loss = 314.229, a = 48.774 b = 28.703, c = -1.247, Joint = -1.957, Marginal = -2.117, MI = -9.485\n",
      "Pre-Train Epoch 5, Batch 129: Loss = 139.594, a = 62.544 b = 35.307, c = -0.269, Joint = -1.488, Marginal = -2.634, MI = -12.792\n",
      "Pre-Train Epoch 5, Batch 130: Loss = 140.783, a = 50.644 b = 41.414, c = -0.269, Joint = -1.226, Marginal = -2.257, MI = -10.311\n",
      "Pre-Train Epoch 5, Batch 131: Loss = 172.318, a = 41.911 b = 35.388, c = -3.708, Joint = -1.549, Marginal = -2.550, MI = -13.823\n",
      "Pre-Train Epoch 5, Batch 132: Loss = 374.949, a = 46.845 b = 33.934, c = -0.268, Joint = -2.311, Marginal = -2.107, MI = -9.730\n",
      "Pre-Train Epoch 5, Batch 133: Loss = 2266.626, a = 74.900 b = 39.233, c = -0.268, Joint = -8.792, Marginal = -15.288, MI = -16.736\n",
      "Pre-Train Epoch 5, Batch 134: Loss = 303.519, a = 67.030 b = 25.974, c = -0.268, Joint = -0.751, Marginal = -5.903, MI = -10.822\n",
      "Pre-Train Epoch 5, Batch 135: Loss = 123.961, a = 51.435 b = 32.632, c = -0.268, Joint = 0.007, Marginal = -2.649, MI = -12.626\n",
      "Pre-Train Epoch 5, Batch 136: Loss = 146.456, a = 48.073 b = 46.963, c = -0.267, Joint = -2.309, Marginal = -2.264, MI = -11.390\n",
      "Pre-Train Epoch 5, Batch 137: Loss = 112.857, a = 48.376 b = 32.794, c = -0.267, Joint = -0.474, Marginal = -2.122, MI = -7.892\n",
      "Pre-Train Epoch 5, Batch 138: Loss = 240.361, a = 43.152 b = 33.302, c = -0.267, Joint = -2.611, Marginal = -4.065, MI = -10.697\n",
      "Pre-Train Epoch 5, Batch 139: Loss = 176.495, a = 55.780 b = 44.394, c = -2.496, Joint = -2.720, Marginal = -4.335, MI = -10.975\n",
      "Pre-Train Epoch 5, Batch 140: Loss = 3641.385, a = 59.961 b = 39.496, c = -3.405, Joint = -10.384, Marginal = -14.735, MI = -17.955\n",
      "Pre-Train Epoch 5, Batch 141: Loss = 169.572, a = 74.992 b = 37.080, c = -0.266, Joint = -0.219, Marginal = -5.089, MI = -11.934\n",
      "Pre-Train Epoch 5, Batch 142: Loss = 258.229, a = 68.832 b = 31.566, c = -0.266, Joint = -0.837, Marginal = -3.100, MI = -10.132\n",
      "Pre-Train Epoch 5, Batch 143: Loss = 1317.967, a = 69.044 b = 41.460, c = -0.266, Joint = -4.514, Marginal = -9.106, MI = -15.111\n",
      "Pre-Train Epoch 5, Batch 144: Loss = 219.421, a = 51.188 b = 41.025, c = -0.265, Joint = -2.415, Marginal = -4.580, MI = -8.775\n",
      "Pre-Train Epoch 5, Batch 145: Loss = 147.877, a = 53.836 b = 33.916, c = -0.265, Joint = -1.876, Marginal = -10.074, MI = -11.508\n",
      "Pre-Train Epoch 5, Batch 146: Loss = 122.653, a = 51.696 b = 36.751, c = -0.265, Joint = -0.673, Marginal = -3.135, MI = -10.673\n",
      "Pre-Train Epoch 5, Batch 147: Loss = 204.714, a = 75.320 b = 41.982, c = -0.265, Joint = -1.116, Marginal = -10.250, MI = -11.311\n",
      "Pre-Train Epoch 5, Batch 148: Loss = 197.332, a = 56.002 b = 34.038, c = -0.264, Joint = -1.971, Marginal = -7.395, MI = -11.513\n",
      "Pre-Train Epoch 5, Batch 149: Loss = 88.946, a = 34.112 b = 25.515, c = -4.571, Joint = -0.323, Marginal = -0.682, MI = -9.413\n",
      "Pre-Train Epoch 5, Batch 150: Loss = 156.860, a = 55.167 b = 36.394, c = -5.864, Joint = -1.427, Marginal = -7.421, MI = -10.580\n",
      "\n",
      "Start of epoch 6\n",
      "Pre-Train Epoch 6, Batch 0: Loss = 306.603, a = 51.619 b = 36.105, c = -0.264, Joint = -2.428, Marginal = -2.419, MI = -12.145\n",
      "Pre-Train Epoch 6, Batch 1: Loss = 3753.299, a = 79.651 b = 40.613, c = -0.263, Joint = -9.503, Marginal = -13.118, MI = -19.840\n",
      "Pre-Train Epoch 6, Batch 2: Loss = 137.590, a = 46.923 b = 40.891, c = -0.263, Joint = -0.898, Marginal = -2.221, MI = -12.678\n",
      "Pre-Train Epoch 6, Batch 3: Loss = 176.485, a = 40.129 b = 35.778, c = -0.263, Joint = -1.963, Marginal = -2.953, MI = -8.285\n",
      "Pre-Train Epoch 6, Batch 4: Loss = 106.390, a = 41.371 b = 29.264, c = -0.263, Joint = -0.137, Marginal = -1.195, MI = -9.195\n",
      "Pre-Train Epoch 6, Batch 5: Loss = 121.694, a = 51.416 b = 33.656, c = -0.262, Joint = -1.418, Marginal = -5.346, MI = -9.498\n",
      "Pre-Train Epoch 6, Batch 6: Loss = 145.137, a = 52.429 b = 35.374, c = -1.159, Joint = -0.156, Marginal = -2.468, MI = -9.626\n",
      "Pre-Train Epoch 6, Batch 7: Loss = 113.441, a = 34.059 b = 41.132, c = -0.262, Joint = -1.209, Marginal = -1.819, MI = -10.545\n",
      "Pre-Train Epoch 6, Batch 8: Loss = 122.449, a = 38.385 b = 39.821, c = -0.262, Joint = -1.195, Marginal = -2.612, MI = -12.110\n",
      "Pre-Train Epoch 6, Batch 9: Loss = 738.652, a = 92.378 b = 33.690, c = -0.261, Joint = -6.127, Marginal = -13.972, MI = -14.685\n",
      "Pre-Train Epoch 6, Batch 10: Loss = 157.876, a = 51.939 b = 43.570, c = -5.961, Joint = -2.000, Marginal = -4.885, MI = -8.679\n",
      "Pre-Train Epoch 6, Batch 11: Loss = 114.267, a = 43.108 b = 34.994, c = -0.261, Joint = -0.514, Marginal = -1.699, MI = -13.694\n",
      "Pre-Train Epoch 6, Batch 12: Loss = 260.662, a = 59.588 b = 27.742, c = -0.261, Joint = -2.230, Marginal = -7.579, MI = -12.134\n",
      "Pre-Train Epoch 6, Batch 13: Loss = 138.560, a = 55.381 b = 32.071, c = -0.260, Joint = -0.065, Marginal = -2.972, MI = -9.070\n",
      "Pre-Train Epoch 6, Batch 14: Loss = 1226.481, a = 43.934 b = 29.539, c = -2.410, Joint = -4.671, Marginal = -8.257, MI = -10.802\n",
      "Pre-Train Epoch 6, Batch 15: Loss = 468.026, a = 48.884 b = 35.677, c = -0.260, Joint = -3.873, Marginal = -3.741, MI = -14.044\n",
      "Pre-Train Epoch 6, Batch 16: Loss = 1538.424, a = 54.812 b = 40.127, c = -0.260, Joint = -7.149, Marginal = -11.932, MI = -13.014\n",
      "Pre-Train Epoch 6, Batch 17: Loss = 985.566, a = 49.893 b = 40.444, c = -1.059, Joint = -5.867, Marginal = -12.666, MI = -13.836\n",
      "Pre-Train Epoch 6, Batch 18: Loss = 146.037, a = 53.322 b = 30.117, c = -0.259, Joint = -1.546, Marginal = -4.047, MI = -11.114\n",
      "Pre-Train Epoch 6, Batch 19: Loss = 155.154, a = 56.900 b = 31.115, c = -1.710, Joint = -0.314, Marginal = -3.109, MI = -12.222\n",
      "Pre-Train Epoch 6, Batch 20: Loss = 90.174, a = 30.702 b = 31.156, c = -0.259, Joint = -0.402, Marginal = -3.815, MI = -8.502\n",
      "Pre-Train Epoch 6, Batch 21: Loss = 210.098, a = 67.703 b = 33.604, c = -0.259, Joint = -2.476, Marginal = -6.953, MI = -13.299\n",
      "Pre-Train Epoch 6, Batch 22: Loss = 104.104, a = 45.148 b = 29.060, c = -5.536, Joint = -0.070, Marginal = -2.744, MI = -7.361\n",
      "Pre-Train Epoch 6, Batch 23: Loss = 251.597, a = 54.447 b = 32.981, c = -0.258, Joint = -2.694, Marginal = -6.323, MI = -13.627\n",
      "Pre-Train Epoch 6, Batch 24: Loss = 103.110, a = 43.499 b = 29.340, c = -0.258, Joint = 0.146, Marginal = -1.392, MI = -11.589\n",
      "Pre-Train Epoch 6, Batch 25: Loss = 1270.725, a = 61.249 b = 37.161, c = -0.258, Joint = -5.201, Marginal = -5.817, MI = -14.070\n",
      "Pre-Train Epoch 6, Batch 26: Loss = 191.438, a = 62.141 b = 42.288, c = -0.258, Joint = -1.852, Marginal = -1.959, MI = -12.756\n",
      "Pre-Train Epoch 6, Batch 27: Loss = 167.200, a = 56.121 b = 41.924, c = -0.257, Joint = -2.532, Marginal = -9.194, MI = -11.452\n",
      "Pre-Train Epoch 6, Batch 28: Loss = 295.794, a = 58.558 b = 28.082, c = -2.425, Joint = -1.609, Marginal = -4.045, MI = -9.598\n",
      "Pre-Train Epoch 6, Batch 29: Loss = 112.836, a = 35.905 b = 38.672, c = -0.257, Joint = -1.481, Marginal = -4.183, MI = -7.308\n",
      "Pre-Train Epoch 6, Batch 30: Loss = 261.562, a = 50.835 b = 44.628, c = -0.257, Joint = -4.110, Marginal = -5.085, MI = -12.681\n",
      "Pre-Train Epoch 6, Batch 31: Loss = 513.529, a = 59.203 b = 40.301, c = -10.350, Joint = -5.703, Marginal = -6.944, MI = -13.207\n",
      "Pre-Train Epoch 6, Batch 32: Loss = 113.012, a = 36.733 b = 35.476, c = -0.256, Joint = -1.477, Marginal = -3.268, MI = -13.403\n",
      "Pre-Train Epoch 6, Batch 33: Loss = 1232.356, a = 50.984 b = 31.827, c = -0.256, Joint = -6.070, Marginal = -9.375, MI = -14.264\n",
      "Pre-Train Epoch 6, Batch 34: Loss = 122.508, a = 45.048 b = 36.140, c = -5.528, Joint = 0.261, Marginal = -2.729, MI = -8.607\n",
      "Pre-Train Epoch 6, Batch 35: Loss = 130.805, a = 47.817 b = 38.803, c = -0.255, Joint = -1.735, Marginal = -4.521, MI = -8.424\n",
      "Pre-Train Epoch 6, Batch 36: Loss = 106.496, a = 34.931 b = 38.870, c = -3.079, Joint = -1.193, Marginal = -2.001, MI = -9.630\n",
      "Pre-Train Epoch 6, Batch 37: Loss = 155.535, a = 48.673 b = 42.681, c = -0.255, Joint = -2.710, Marginal = -6.579, MI = -11.026\n",
      "Pre-Train Epoch 6, Batch 38: Loss = 82.319, a = 25.991 b = 33.691, c = -0.255, Joint = -0.858, Marginal = -1.636, MI = -10.887\n",
      "Pre-Train Epoch 6, Batch 39: Loss = 113.813, a = 40.242 b = 41.254, c = -0.254, Joint = -1.283, Marginal = -3.108, MI = -9.540\n",
      "Pre-Train Epoch 6, Batch 40: Loss = 17038.025, a = 93.446 b = 29.956, c = -0.254, Joint = -20.549, Marginal = -37.369, MI = -30.327\n",
      "Pre-Train Epoch 6, Batch 41: Loss = 134.049, a = 62.313 b = 38.531, c = -0.254, Joint = -0.938, Marginal = -7.688, MI = -10.858\n",
      "Pre-Train Epoch 6, Batch 42: Loss = 102.779, a = 37.544 b = 38.196, c = -0.355, Joint = -1.233, Marginal = -2.775, MI = -8.324\n",
      "Pre-Train Epoch 6, Batch 43: Loss = 182.456, a = 55.206 b = 40.526, c = -6.469, Joint = -2.832, Marginal = -7.067, MI = -8.921\n",
      "Pre-Train Epoch 6, Batch 44: Loss = 202.160, a = 60.782 b = 35.999, c = -0.253, Joint = -2.370, Marginal = -6.128, MI = -10.493\n",
      "Pre-Train Epoch 6, Batch 45: Loss = 145.316, a = 45.089 b = 35.951, c = -2.660, Joint = -1.190, Marginal = -0.842, MI = -8.982\n",
      "Pre-Train Epoch 6, Batch 46: Loss = 122.948, a = 51.163 b = 39.670, c = -0.252, Joint = -1.588, Marginal = -5.494, MI = -9.750\n",
      "Pre-Train Epoch 6, Batch 47: Loss = 4150.948, a = 61.311 b = 30.937, c = -0.252, Joint = -9.052, Marginal = -14.198, MI = -17.142\n",
      "Pre-Train Epoch 6, Batch 48: Loss = 293.774, a = 81.676 b = 35.335, c = -0.251, Joint = -3.345, Marginal = -20.674, MI = -12.292\n",
      "Pre-Train Epoch 6, Batch 49: Loss = 127.503, a = 43.434 b = 35.356, c = -0.251, Joint = -0.796, Marginal = -3.132, MI = -9.616\n",
      "Pre-Train Epoch 6, Batch 50: Loss = 191.513, a = 50.197 b = 29.258, c = -0.251, Joint = -2.801, Marginal = -7.359, MI = -10.793\n",
      "Pre-Train Epoch 6, Batch 51: Loss = 197.855, a = 67.520 b = 35.071, c = -1.691, Joint = -1.868, Marginal = -5.662, MI = -11.630\n",
      "Pre-Train Epoch 6, Batch 52: Loss = 92.847, a = 35.431 b = 30.785, c = -0.644, Joint = -0.346, Marginal = -0.822, MI = -6.052\n",
      "Pre-Train Epoch 6, Batch 53: Loss = 152.529, a = 60.541 b = 24.979, c = -2.954, Joint = 0.266, Marginal = -2.712, MI = -10.616\n",
      "Pre-Train Epoch 6, Batch 54: Loss = 134.462, a = 37.274 b = 40.322, c = -0.249, Joint = -2.275, Marginal = -5.484, MI = -8.882\n",
      "Pre-Train Epoch 6, Batch 55: Loss = 10632.143, a = 77.071 b = 37.075, c = -0.249, Joint = -14.639, Marginal = -27.291, MI = -21.484\n",
      "Pre-Train Epoch 6, Batch 56: Loss = 229.680, a = 48.251 b = 36.491, c = -2.653, Joint = -2.550, Marginal = -2.416, MI = -9.246\n",
      "Pre-Train Epoch 6, Batch 57: Loss = 106.197, a = 39.684 b = 34.844, c = -6.004, Joint = -0.617, Marginal = -4.736, MI = -10.430\n",
      "Pre-Train Epoch 6, Batch 58: Loss = 176.203, a = 43.100 b = 39.355, c = -4.395, Joint = -2.006, Marginal = -5.210, MI = -9.508\n",
      "Pre-Train Epoch 6, Batch 59: Loss = 111.878, a = 37.729 b = 32.057, c = -0.596, Joint = -0.206, Marginal = -0.656, MI = -8.096\n",
      "Pre-Train Epoch 6, Batch 60: Loss = 4147.515, a = 67.807 b = 34.474, c = -0.247, Joint = -10.420, Marginal = -17.368, MI = -16.462\n",
      "Pre-Train Epoch 6, Batch 61: Loss = 136.607, a = 54.433 b = 33.849, c = -0.247, Joint = -0.529, Marginal = -0.903, MI = -11.391\n",
      "Pre-Train Epoch 6, Batch 62: Loss = 635.669, a = 67.772 b = 32.224, c = -5.265, Joint = -4.805, Marginal = -8.413, MI = -14.938\n",
      "Pre-Train Epoch 6, Batch 63: Loss = 240.815, a = 58.707 b = 30.466, c = -0.246, Joint = -2.147, Marginal = -8.703, MI = -10.864\n",
      "Pre-Train Epoch 6, Batch 64: Loss = 113.333, a = 38.684 b = 40.336, c = -0.246, Joint = -1.252, Marginal = -5.714, MI = -9.478\n",
      "Pre-Train Epoch 6, Batch 65: Loss = 1861.308, a = 93.802 b = 28.960, c = -1.333, Joint = -7.387, Marginal = -14.606, MI = -15.230\n",
      "Pre-Train Epoch 6, Batch 66: Loss = 115.050, a = 38.809 b = 42.779, c = -0.246, Joint = -1.789, Marginal = -3.308, MI = -10.735\n",
      "Pre-Train Epoch 6, Batch 67: Loss = 146.345, a = 55.200 b = 46.365, c = -0.245, Joint = -2.178, Marginal = -4.654, MI = -12.329\n",
      "Pre-Train Epoch 6, Batch 68: Loss = 198.795, a = 67.832 b = 31.086, c = -0.245, Joint = -0.552, Marginal = -7.432, MI = -10.942\n",
      "Pre-Train Epoch 6, Batch 69: Loss = 264.741, a = 57.446 b = 32.915, c = -0.245, Joint = -2.691, Marginal = -5.494, MI = -12.313\n",
      "Pre-Train Epoch 6, Batch 70: Loss = 480.108, a = 44.217 b = 41.736, c = -2.774, Joint = -4.102, Marginal = -3.894, MI = -17.301\n",
      "Pre-Train Epoch 6, Batch 71: Loss = 1542.875, a = 74.735 b = 48.296, c = -0.244, Joint = -7.040, Marginal = -20.020, MI = -12.997\n",
      "Pre-Train Epoch 6, Batch 72: Loss = 504.857, a = 88.558 b = 29.142, c = -0.244, Joint = -3.391, Marginal = -10.844, MI = -12.621\n",
      "Pre-Train Epoch 6, Batch 73: Loss = 347.901, a = 60.009 b = 26.716, c = -0.243, Joint = -2.147, Marginal = -9.034, MI = -11.192\n",
      "Pre-Train Epoch 6, Batch 74: Loss = 124.717, a = 58.151 b = 34.499, c = -0.700, Joint = -0.972, Marginal = -4.752, MI = -12.283\n",
      "Pre-Train Epoch 6, Batch 75: Loss = 159.329, a = 61.100 b = 29.726, c = -0.243, Joint = -0.188, Marginal = -2.088, MI = -8.210\n",
      "Pre-Train Epoch 6, Batch 76: Loss = 313.466, a = 49.906 b = 35.235, c = -0.243, Joint = -2.761, Marginal = -1.642, MI = -12.593\n",
      "Pre-Train Epoch 6, Batch 77: Loss = 911.354, a = 63.932 b = 34.174, c = -0.242, Joint = -6.155, Marginal = -5.571, MI = -14.169\n",
      "Pre-Train Epoch 6, Batch 78: Loss = 121.636, a = 28.295 b = 42.401, c = -0.242, Joint = -2.923, Marginal = -2.922, MI = -10.724\n",
      "Pre-Train Epoch 6, Batch 79: Loss = 365.779, a = 52.588 b = 33.401, c = -0.242, Joint = -3.015, Marginal = -4.488, MI = -8.653\n",
      "Pre-Train Epoch 6, Batch 80: Loss = 338.366, a = 44.462 b = 39.729, c = -0.242, Joint = -3.205, Marginal = -3.170, MI = -14.036\n",
      "Pre-Train Epoch 6, Batch 81: Loss = 169.299, a = 51.244 b = 35.127, c = -5.698, Joint = -2.352, Marginal = -3.197, MI = -17.129\n",
      "Pre-Train Epoch 6, Batch 82: Loss = 734.992, a = 69.860 b = 37.463, c = -4.942, Joint = -3.536, Marginal = -18.288, MI = -13.817\n",
      "Pre-Train Epoch 6, Batch 83: Loss = 105.137, a = 43.609 b = 33.690, c = -0.241, Joint = -1.007, Marginal = -5.422, MI = -10.172\n",
      "Pre-Train Epoch 6, Batch 84: Loss = 204.087, a = 58.218 b = 31.893, c = -0.241, Joint = -0.955, Marginal = -13.200, MI = -11.329\n",
      "Pre-Train Epoch 6, Batch 85: Loss = 1285.055, a = 62.965 b = 34.757, c = -0.240, Joint = -7.020, Marginal = -14.125, MI = -16.422\n",
      "Pre-Train Epoch 6, Batch 86: Loss = 111.937, a = 43.323 b = 33.982, c = -0.240, Joint = -1.137, Marginal = -2.053, MI = -9.472\n",
      "Pre-Train Epoch 6, Batch 87: Loss = 335.677, a = 49.229 b = 30.843, c = -0.240, Joint = -1.539, Marginal = -5.177, MI = -11.225\n",
      "Pre-Train Epoch 6, Batch 88: Loss = 94.802, a = 30.555 b = 37.360, c = -0.240, Joint = -0.766, Marginal = -1.734, MI = -7.502\n",
      "Pre-Train Epoch 6, Batch 89: Loss = 115.279, a = 42.998 b = 35.549, c = -1.635, Joint = -0.216, Marginal = -2.446, MI = -9.578\n",
      "Pre-Train Epoch 6, Batch 90: Loss = 178.512, a = 70.835 b = 33.762, c = -0.239, Joint = -1.382, Marginal = -12.151, MI = -11.557\n",
      "Pre-Train Epoch 6, Batch 91: Loss = 101.411, a = 32.213 b = 35.607, c = -0.239, Joint = -1.074, Marginal = -2.258, MI = -9.703\n",
      "Pre-Train Epoch 6, Batch 92: Loss = 209.766, a = 88.607 b = 27.871, c = -0.239, Joint = -1.765, Marginal = -9.217, MI = -12.948\n",
      "Pre-Train Epoch 6, Batch 93: Loss = 769.921, a = 65.436 b = 44.649, c = -0.239, Joint = -6.774, Marginal = -7.048, MI = -13.300\n",
      "Pre-Train Epoch 6, Batch 94: Loss = 106.692, a = 45.320 b = 31.452, c = -0.238, Joint = 0.112, Marginal = -1.648, MI = -9.835\n",
      "Pre-Train Epoch 6, Batch 95: Loss = 561.169, a = 63.034 b = 40.528, c = -0.238, Joint = -4.666, Marginal = -5.900, MI = -12.014\n",
      "Pre-Train Epoch 6, Batch 96: Loss = 240.466, a = 42.663 b = 41.398, c = -0.238, Joint = -2.535, Marginal = -4.556, MI = -8.543\n",
      "Pre-Train Epoch 6, Batch 97: Loss = 106.991, a = 40.933 b = 36.827, c = -0.238, Joint = -0.922, Marginal = -3.263, MI = -14.276\n",
      "Pre-Train Epoch 6, Batch 98: Loss = 262.097, a = 69.117 b = 37.559, c = -6.775, Joint = -2.291, Marginal = -7.286, MI = -9.326\n",
      "Pre-Train Epoch 6, Batch 99: Loss = 303.658, a = 60.100 b = 37.822, c = -0.349, Joint = -2.031, Marginal = -2.921, MI = -11.803\n",
      "Pre-Train Epoch 6, Batch 100: Loss = 100.220, a = 33.196 b = 36.269, c = -0.237, Joint = -0.080, Marginal = -1.399, MI = -8.182\n",
      "Pre-Train Epoch 6, Batch 101: Loss = 811.306, a = 99.502 b = 36.319, c = -5.480, Joint = -5.010, Marginal = -14.534, MI = -10.768\n",
      "Pre-Train Epoch 6, Batch 102: Loss = 374.548, a = 54.002 b = 34.080, c = -0.237, Joint = -2.212, Marginal = -2.752, MI = -11.855\n",
      "Pre-Train Epoch 6, Batch 103: Loss = 97.855, a = 37.763 b = 35.664, c = -0.236, Joint = -1.016, Marginal = -4.923, MI = -7.268\n",
      "Pre-Train Epoch 6, Batch 104: Loss = 459.883, a = 59.391 b = 33.806, c = -0.236, Joint = -4.024, Marginal = -13.224, MI = -12.409\n",
      "Pre-Train Epoch 6, Batch 105: Loss = 326.545, a = 51.299 b = 32.061, c = -0.236, Joint = -3.111, Marginal = -4.089, MI = -9.893\n",
      "Pre-Train Epoch 6, Batch 106: Loss = 406.364, a = 72.252 b = 36.809, c = -0.236, Joint = -3.801, Marginal = -9.900, MI = -9.753\n",
      "Pre-Train Epoch 6, Batch 107: Loss = 127.915, a = 42.774 b = 34.946, c = -3.397, Joint = -0.819, Marginal = -6.525, MI = -9.404\n",
      "Pre-Train Epoch 6, Batch 108: Loss = 680.577, a = 31.277 b = 41.691, c = -0.235, Joint = -4.769, Marginal = -9.322, MI = -12.644\n",
      "Pre-Train Epoch 6, Batch 109: Loss = 1785.534, a = 44.195 b = 31.891, c = -1.885, Joint = -6.916, Marginal = -12.202, MI = -15.053\n",
      "Pre-Train Epoch 6, Batch 110: Loss = 1438.716, a = 62.293 b = 36.548, c = -0.235, Joint = -5.171, Marginal = -15.225, MI = -13.672\n",
      "Pre-Train Epoch 6, Batch 111: Loss = 3013.731, a = 75.576 b = 41.406, c = -0.235, Joint = -11.975, Marginal = -13.585, MI = -19.458\n",
      "Pre-Train Epoch 6, Batch 112: Loss = 211.539, a = 45.985 b = 31.077, c = -0.467, Joint = -1.765, Marginal = -3.376, MI = -9.266\n",
      "Pre-Train Epoch 6, Batch 113: Loss = 205.804, a = 59.513 b = 40.827, c = -0.234, Joint = -2.884, Marginal = -7.691, MI = -11.207\n",
      "Pre-Train Epoch 6, Batch 114: Loss = 133.272, a = 28.993 b = 34.840, c = -0.234, Joint = -2.001, Marginal = -3.187, MI = -9.955\n",
      "Pre-Train Epoch 6, Batch 115: Loss = 1120.442, a = 58.503 b = 32.575, c = -0.233, Joint = -4.847, Marginal = -11.716, MI = -12.150\n",
      "Pre-Train Epoch 6, Batch 116: Loss = 103.747, a = 34.309 b = 40.104, c = -0.233, Joint = -1.060, Marginal = -3.964, MI = -12.673\n",
      "Pre-Train Epoch 6, Batch 117: Loss = 353.671, a = 48.613 b = 35.405, c = -0.233, Joint = -3.321, Marginal = -2.395, MI = -12.652\n",
      "Pre-Train Epoch 6, Batch 118: Loss = 127.229, a = 37.619 b = 41.993, c = -0.233, Joint = -1.566, Marginal = -5.172, MI = -9.476\n",
      "Pre-Train Epoch 6, Batch 119: Loss = 112.148, a = 47.493 b = 33.696, c = -0.232, Joint = -1.412, Marginal = -6.019, MI = -8.786\n",
      "Pre-Train Epoch 6, Batch 120: Loss = 82.140, a = 28.890 b = 30.494, c = -6.944, Joint = -0.727, Marginal = -1.329, MI = -11.305\n",
      "Pre-Train Epoch 6, Batch 121: Loss = 82.234, a = 29.982 b = 29.663, c = -0.232, Joint = -0.224, Marginal = -1.013, MI = -7.692\n",
      "Pre-Train Epoch 6, Batch 122: Loss = 115.051, a = 50.621 b = 33.318, c = -5.616, Joint = -0.426, Marginal = -2.772, MI = -8.140\n",
      "Pre-Train Epoch 6, Batch 123: Loss = 139.087, a = 53.373 b = 34.716, c = -1.640, Joint = 0.688, Marginal = -4.104, MI = -7.810\n",
      "Pre-Train Epoch 6, Batch 124: Loss = 109.269, a = 44.535 b = 28.880, c = -0.231, Joint = 0.711, Marginal = -1.254, MI = -10.293\n",
      "Pre-Train Epoch 6, Batch 125: Loss = 728.601, a = 61.572 b = 28.848, c = -0.699, Joint = -5.518, Marginal = -7.883, MI = -11.579\n",
      "Pre-Train Epoch 6, Batch 126: Loss = 92.924, a = 30.617 b = 34.868, c = -0.231, Joint = -0.374, Marginal = -1.616, MI = -8.437\n",
      "Pre-Train Epoch 6, Batch 127: Loss = 113.235, a = 39.348 b = 41.653, c = -0.231, Joint = -1.579, Marginal = -3.110, MI = -8.159\n",
      "Pre-Train Epoch 6, Batch 128: Loss = 244.778, a = 35.584 b = 28.366, c = -1.188, Joint = -1.843, Marginal = -1.956, MI = -10.726\n",
      "Pre-Train Epoch 6, Batch 129: Loss = 110.221, a = 46.219 b = 34.772, c = -0.230, Joint = -1.285, Marginal = -8.330, MI = -8.590\n",
      "Pre-Train Epoch 6, Batch 130: Loss = 114.394, a = 36.533 b = 40.778, c = -0.230, Joint = -1.038, Marginal = -2.268, MI = -11.133\n",
      "Pre-Train Epoch 6, Batch 131: Loss = 136.812, a = 29.897 b = 34.969, c = -3.596, Joint = -1.471, Marginal = -2.831, MI = -7.714\n",
      "Pre-Train Epoch 6, Batch 132: Loss = 290.852, a = 34.239 b = 33.573, c = -0.230, Joint = -2.118, Marginal = -4.921, MI = -9.278\n",
      "Pre-Train Epoch 6, Batch 133: Loss = 1757.398, a = 57.112 b = 38.545, c = -0.230, Joint = -7.769, Marginal = -15.093, MI = -17.643\n",
      "Pre-Train Epoch 6, Batch 134: Loss = 233.899, a = 50.190 b = 25.727, c = -0.229, Joint = -0.704, Marginal = -5.239, MI = -9.644\n",
      "Pre-Train Epoch 6, Batch 135: Loss = 100.448, a = 37.725 b = 32.216, c = -0.229, Joint = -0.042, Marginal = -3.829, MI = -8.032\n",
      "Pre-Train Epoch 6, Batch 136: Loss = 116.041, a = 34.377 b = 46.111, c = -0.229, Joint = -2.074, Marginal = -2.149, MI = -10.314\n",
      "Pre-Train Epoch 6, Batch 137: Loss = 91.457, a = 35.704 b = 32.347, c = -0.229, Joint = -0.361, Marginal = -4.640, MI = -7.660\n",
      "Pre-Train Epoch 6, Batch 138: Loss = 188.172, a = 31.936 b = 32.912, c = -0.228, Joint = -2.300, Marginal = -3.195, MI = -9.854\n",
      "Pre-Train Epoch 6, Batch 139: Loss = 138.726, a = 41.444 b = 43.601, c = -2.381, Joint = -2.248, Marginal = -3.891, MI = -8.769\n",
      "Pre-Train Epoch 6, Batch 140: Loss = 2863.052, a = 45.943 b = 38.859, c = -3.254, Joint = -9.199, Marginal = -13.037, MI = -17.632\n",
      "Pre-Train Epoch 6, Batch 141: Loss = 133.956, a = 55.738 b = 36.314, c = -0.228, Joint = -0.141, Marginal = -4.805, MI = -10.375\n",
      "Pre-Train Epoch 6, Batch 142: Loss = 197.103, a = 52.070 b = 31.124, c = -0.228, Joint = -0.621, Marginal = -7.973, MI = -10.306\n",
      "Pre-Train Epoch 6, Batch 143: Loss = 1042.965, a = 51.330 b = 40.643, c = -0.227, Joint = -4.022, Marginal = -4.241, MI = -12.246\n",
      "Pre-Train Epoch 6, Batch 144: Loss = 175.156, a = 38.153 b = 40.409, c = -0.227, Joint = -2.099, Marginal = -4.024, MI = -7.097\n",
      "Pre-Train Epoch 6, Batch 145: Loss = 115.891, a = 40.179 b = 33.356, c = -0.227, Joint = -1.547, Marginal = -8.205, MI = -12.393\n",
      "Pre-Train Epoch 6, Batch 146: Loss = 99.372, a = 38.052 b = 36.187, c = -0.227, Joint = -0.549, Marginal = -2.850, MI = -9.924\n",
      "Pre-Train Epoch 6, Batch 147: Loss = 157.435, a = 55.935 b = 41.086, c = -0.226, Joint = -0.909, Marginal = -2.725, MI = -9.132\n",
      "Pre-Train Epoch 6, Batch 148: Loss = 154.936, a = 41.246 b = 33.474, c = -0.226, Joint = -1.747, Marginal = -5.599, MI = -9.289\n",
      "Pre-Train Epoch 6, Batch 149: Loss = 72.517, a = 24.715 b = 25.382, c = -4.426, Joint = -0.371, Marginal = -2.920, MI = -7.633\n",
      "Pre-Train Epoch 6, Batch 150: Loss = 121.676, a = 40.776 b = 35.857, c = -5.725, Joint = -1.232, Marginal = -5.373, MI = -11.652\n",
      "\n",
      "Start of epoch 7\n",
      "Pre-Train Epoch 7, Batch 0: Loss = 239.917, a = 37.401 b = 35.489, c = -0.225, Joint = -2.192, Marginal = -3.860, MI = -10.734\n",
      "Pre-Train Epoch 7, Batch 1: Loss = 2933.764, a = 60.109 b = 39.777, c = -0.225, Joint = -8.403, Marginal = -18.246, MI = -19.064\n",
      "Pre-Train Epoch 7, Batch 2: Loss = 109.986, a = 33.253 b = 40.227, c = -0.225, Joint = -0.889, Marginal = -2.258, MI = -10.629\n",
      "Pre-Train Epoch 7, Batch 3: Loss = 139.966, a = 28.882 b = 35.317, c = -0.225, Joint = -1.816, Marginal = -3.861, MI = -9.661\n",
      "Pre-Train Epoch 7, Batch 4: Loss = 84.787, a = 29.763 b = 28.998, c = -0.224, Joint = -0.236, Marginal = -1.154, MI = -7.889\n",
      "Pre-Train Epoch 7, Batch 5: Loss = 98.019, a = 37.731 b = 33.204, c = -0.224, Joint = -1.266, Marginal = -2.084, MI = -9.175\n",
      "Pre-Train Epoch 7, Batch 6: Loss = 113.790, a = 37.458 b = 34.860, c = -0.981, Joint = -0.246, Marginal = -1.191, MI = -12.659\n",
      "Pre-Train Epoch 7, Batch 7: Loss = 92.498, a = 23.579 b = 40.567, c = -0.224, Joint = -1.155, Marginal = -1.186, MI = -8.905\n",
      "Pre-Train Epoch 7, Batch 8: Loss = 99.029, a = 26.912 b = 39.213, c = -0.224, Joint = -1.153, Marginal = -1.776, MI = -9.075\n",
      "Pre-Train Epoch 7, Batch 9: Loss = 558.848, a = 70.927 b = 33.061, c = -0.223, Joint = -5.347, Marginal = -11.317, MI = -13.950\n",
      "Pre-Train Epoch 7, Batch 10: Loss = 124.545, a = 37.656 b = 42.756, c = -5.728, Joint = -1.816, Marginal = -7.092, MI = -9.131\n",
      "Pre-Train Epoch 7, Batch 11: Loss = 94.265, a = 31.412 b = 34.549, c = -0.223, Joint = -0.511, Marginal = -2.987, MI = -9.538\n",
      "Pre-Train Epoch 7, Batch 12: Loss = 197.130, a = 44.757 b = 27.456, c = -0.223, Joint = -1.961, Marginal = -6.855, MI = -10.306\n",
      "Pre-Train Epoch 7, Batch 13: Loss = 108.544, a = 40.016 b = 31.571, c = -0.223, Joint = -0.104, Marginal = -0.911, MI = -7.598\n",
      "Pre-Train Epoch 7, Batch 14: Loss = 952.017, a = 33.244 b = 29.230, c = -2.288, Joint = -4.146, Marginal = -18.064, MI = -12.977\n",
      "Pre-Train Epoch 7, Batch 15: Loss = 370.166, a = 35.943 b = 35.192, c = -0.222, Joint = -3.475, Marginal = -2.338, MI = -11.770\n",
      "Pre-Train Epoch 7, Batch 16: Loss = 1236.910, a = 41.030 b = 39.471, c = -0.222, Joint = -6.383, Marginal = -7.794, MI = -15.770\n",
      "Pre-Train Epoch 7, Batch 17: Loss = 757.728, a = 38.202 b = 39.899, c = -0.995, Joint = -5.080, Marginal = -15.921, MI = -12.518\n",
      "Pre-Train Epoch 7, Batch 18: Loss = 114.979, a = 40.763 b = 29.786, c = -0.221, Joint = -1.165, Marginal = -3.126, MI = -8.490\n",
      "Pre-Train Epoch 7, Batch 19: Loss = 119.857, a = 41.884 b = 30.643, c = -1.610, Joint = -0.331, Marginal = -2.210, MI = -9.921\n",
      "Pre-Train Epoch 7, Batch 20: Loss = 74.553, a = 22.327 b = 30.866, c = -0.221, Joint = -0.402, Marginal = -2.619, MI = -7.528\n",
      "Pre-Train Epoch 7, Batch 21: Loss = 163.241, a = 51.609 b = 33.045, c = -0.221, Joint = -2.073, Marginal = -7.830, MI = -10.203\n",
      "Pre-Train Epoch 7, Batch 22: Loss = 85.642, a = 33.588 b = 28.763, c = -5.333, Joint = -0.055, Marginal = -1.871, MI = -8.456\n",
      "Pre-Train Epoch 7, Batch 23: Loss = 197.752, a = 40.770 b = 32.521, c = -0.220, Joint = -2.386, Marginal = -4.073, MI = -9.808\n",
      "Pre-Train Epoch 7, Batch 24: Loss = 85.346, a = 32.322 b = 29.046, c = -0.220, Joint = 0.142, Marginal = -2.335, MI = -8.279\n",
      "Pre-Train Epoch 7, Batch 25: Loss = 1013.501, a = 45.880 b = 36.511, c = -0.220, Joint = -4.616, Marginal = -3.788, MI = -12.004\n",
      "Pre-Train Epoch 7, Batch 26: Loss = 149.782, a = 45.515 b = 41.443, c = -0.220, Joint = -1.579, Marginal = -2.807, MI = -9.100\n",
      "Pre-Train Epoch 7, Batch 27: Loss = 132.541, a = 42.002 b = 41.185, c = -0.220, Joint = -2.125, Marginal = -4.626, MI = -10.874\n",
      "Pre-Train Epoch 7, Batch 28: Loss = 230.219, a = 44.002 b = 27.712, c = -2.387, Joint = -1.398, Marginal = -4.911, MI = -7.601\n",
      "Pre-Train Epoch 7, Batch 29: Loss = 91.343, a = 25.653 b = 38.092, c = -0.219, Joint = -1.327, Marginal = -2.763, MI = -5.464\n",
      "Pre-Train Epoch 7, Batch 30: Loss = 205.266, a = 37.921 b = 43.872, c = -0.219, Joint = -3.486, Marginal = -3.410, MI = -10.661\n",
      "Pre-Train Epoch 7, Batch 31: Loss = 391.385, a = 44.700 b = 39.579, c = -10.059, Joint = -4.846, Marginal = -5.736, MI = -9.226\n",
      "Pre-Train Epoch 7, Batch 32: Loss = 89.871, a = 26.831 b = 34.992, c = -0.219, Joint = -1.355, Marginal = -4.378, MI = -9.747\n",
      "Pre-Train Epoch 7, Batch 33: Loss = 956.849, a = 38.831 b = 31.380, c = -0.218, Joint = -5.379, Marginal = -11.566, MI = -10.841\n",
      "Pre-Train Epoch 7, Batch 34: Loss = 97.751, a = 31.780 b = 35.557, c = -5.343, Joint = 0.029, Marginal = -2.590, MI = -4.576\n",
      "Pre-Train Epoch 7, Batch 35: Loss = 105.166, a = 34.964 b = 38.160, c = -0.218, Joint = -1.487, Marginal = -2.086, MI = -8.895\n",
      "Pre-Train Epoch 7, Batch 36: Loss = 87.361, a = 24.846 b = 38.345, c = -2.937, Joint = -1.076, Marginal = -1.302, MI = -7.990\n",
      "Pre-Train Epoch 7, Batch 37: Loss = 124.669, a = 35.786 b = 41.983, c = -0.218, Joint = -2.315, Marginal = -6.653, MI = -9.450\n",
      "Pre-Train Epoch 7, Batch 38: Loss = 68.466, a = 18.185 b = 33.383, c = -0.217, Joint = -0.849, Marginal = -2.606, MI = -8.830\n",
      "Pre-Train Epoch 7, Batch 39: Loss = 93.213, a = 28.688 b = 40.602, c = -0.217, Joint = -1.184, Marginal = -3.832, MI = -8.357\n",
      "Pre-Train Epoch 7, Batch 40: Loss = 13650.843, a = 73.701 b = 29.337, c = -0.217, Joint = -18.358, Marginal = -35.452, MI = -25.268\n",
      "Pre-Train Epoch 7, Batch 41: Loss = 108.517, a = 46.477 b = 37.865, c = -0.217, Joint = -0.777, Marginal = -5.350, MI = -6.193\n",
      "Pre-Train Epoch 7, Batch 42: Loss = 84.824, a = 27.178 b = 37.629, c = -0.280, Joint = -1.111, Marginal = -2.566, MI = -6.916\n",
      "Pre-Train Epoch 7, Batch 43: Loss = 142.517, a = 40.968 b = 39.809, c = -6.252, Joint = -2.451, Marginal = -3.773, MI = -7.304\n",
      "Pre-Train Epoch 7, Batch 44: Loss = 153.872, a = 45.630 b = 35.387, c = -0.216, Joint = -2.052, Marginal = -5.953, MI = -8.022\n",
      "Pre-Train Epoch 7, Batch 45: Loss = 117.060, a = 32.861 b = 35.423, c = -2.552, Joint = -1.047, Marginal = -0.805, MI = -8.747\n",
      "Pre-Train Epoch 7, Batch 46: Loss = 100.321, a = 37.727 b = 38.999, c = -0.215, Joint = -1.345, Marginal = -3.175, MI = -8.338\n",
      "Pre-Train Epoch 7, Batch 47: Loss = 3352.147, a = 46.898 b = 30.477, c = -0.215, Joint = -8.096, Marginal = -8.901, MI = -15.426\n",
      "Pre-Train Epoch 7, Batch 48: Loss = 221.230, a = 62.403 b = 34.661, c = -0.214, Joint = -2.777, Marginal = -8.333, MI = -10.694\n",
      "Pre-Train Epoch 7, Batch 49: Loss = 105.030, a = 32.131 b = 34.904, c = -0.214, Joint = -0.695, Marginal = -1.864, MI = -11.944\n",
      "Pre-Train Epoch 7, Batch 50: Loss = 147.808, a = 37.794 b = 28.852, c = -0.214, Joint = -2.519, Marginal = -8.118, MI = -7.715\n",
      "Pre-Train Epoch 7, Batch 51: Loss = 152.916, a = 51.061 b = 34.466, c = -1.615, Joint = -1.544, Marginal = -4.240, MI = -11.145\n",
      "Pre-Train Epoch 7, Batch 52: Loss = 75.289, a = 25.428 b = 30.394, c = -0.564, Joint = -0.350, Marginal = -0.084, MI = -7.496\n",
      "Pre-Train Epoch 7, Batch 53: Loss = 119.007, a = 45.294 b = 24.691, c = -2.814, Joint = 0.271, Marginal = -1.052, MI = -7.578\n",
      "Pre-Train Epoch 7, Batch 54: Loss = 107.524, a = 27.280 b = 39.753, c = -0.213, Joint = -2.005, Marginal = -4.981, MI = -7.672\n",
      "Pre-Train Epoch 7, Batch 55: Loss = 8498.510, a = 60.445 b = 36.407, c = -0.212, Joint = -13.047, Marginal = -21.484, MI = -18.546\n",
      "Pre-Train Epoch 7, Batch 56: Loss = 179.455, a = 35.965 b = 35.971, c = -2.533, Joint = -2.238, Marginal = -3.369, MI = -7.879\n",
      "Pre-Train Epoch 7, Batch 57: Loss = 87.463, a = 29.354 b = 34.410, c = -5.791, Joint = -0.572, Marginal = -2.277, MI = -7.191\n",
      "Pre-Train Epoch 7, Batch 58: Loss = 140.252, a = 31.892 b = 38.676, c = -4.268, Joint = -1.706, Marginal = -4.974, MI = -6.879\n",
      "Pre-Train Epoch 7, Batch 59: Loss = 89.181, a = 27.046 b = 31.676, c = -0.492, Joint = -0.280, Marginal = -1.306, MI = -7.252\n",
      "Pre-Train Epoch 7, Batch 60: Loss = 3258.448, a = 53.195 b = 33.984, c = -0.211, Joint = -9.239, Marginal = -14.677, MI = -15.904\n",
      "Pre-Train Epoch 7, Batch 61: Loss = 109.288, a = 40.393 b = 33.320, c = -0.210, Joint = -0.449, Marginal = -1.931, MI = -8.541\n",
      "Pre-Train Epoch 7, Batch 62: Loss = 490.006, a = 51.898 b = 31.753, c = -5.090, Joint = -4.248, Marginal = -8.041, MI = -12.179\n",
      "Pre-Train Epoch 7, Batch 63: Loss = 182.092, a = 44.773 b = 30.058, c = -0.210, Joint = -1.876, Marginal = -8.209, MI = -7.968\n",
      "Pre-Train Epoch 7, Batch 64: Loss = 92.250, a = 28.186 b = 39.736, c = -0.210, Joint = -1.146, Marginal = -4.093, MI = -5.350\n",
      "Pre-Train Epoch 7, Batch 65: Loss = 1474.358, a = 73.721 b = 28.443, c = -1.241, Joint = -6.437, Marginal = -10.422, MI = -11.992\n",
      "Pre-Train Epoch 7, Batch 66: Loss = 94.754, a = 28.236 b = 42.127, c = -0.209, Joint = -1.564, Marginal = -2.128, MI = -8.536\n",
      "Pre-Train Epoch 7, Batch 67: Loss = 117.256, a = 41.079 b = 45.397, c = -0.209, Joint = -1.876, Marginal = -6.136, MI = -6.759\n",
      "Pre-Train Epoch 7, Batch 68: Loss = 152.678, a = 51.583 b = 30.594, c = -0.208, Joint = -0.463, Marginal = -4.734, MI = -8.430\n",
      "Pre-Train Epoch 7, Batch 69: Loss = 204.940, a = 43.818 b = 32.459, c = -0.208, Joint = -2.318, Marginal = -7.206, MI = -9.739\n",
      "Pre-Train Epoch 7, Batch 70: Loss = 385.295, a = 33.133 b = 41.101, c = -2.671, Joint = -3.717, Marginal = -8.812, MI = -12.029\n",
      "Pre-Train Epoch 7, Batch 71: Loss = 1197.916, a = 57.787 b = 47.345, c = -0.208, Joint = -6.077, Marginal = -16.627, MI = -13.798\n",
      "Pre-Train Epoch 7, Batch 72: Loss = 386.829, a = 69.016 b = 28.695, c = -0.207, Joint = -2.859, Marginal = -9.095, MI = -10.654\n",
      "Pre-Train Epoch 7, Batch 73: Loss = 261.106, a = 45.642 b = 26.359, c = -0.207, Joint = -1.974, Marginal = -9.361, MI = -9.799\n",
      "Pre-Train Epoch 7, Batch 74: Loss = 102.925, a = 44.156 b = 33.987, c = -0.634, Joint = -0.785, Marginal = -3.955, MI = -10.708\n",
      "Pre-Train Epoch 7, Batch 75: Loss = 126.306, a = 46.052 b = 29.390, c = -0.207, Joint = -0.174, Marginal = 0.178, MI = -10.106\n",
      "Pre-Train Epoch 7, Batch 76: Loss = 248.433, a = 37.645 b = 34.796, c = -0.206, Joint = -2.488, Marginal = -2.983, MI = -8.661\n",
      "Pre-Train Epoch 7, Batch 77: Loss = 700.105, a = 49.950 b = 33.664, c = -0.206, Joint = -5.344, Marginal = -12.174, MI = -9.994\n",
      "Pre-Train Epoch 7, Batch 78: Loss = 98.818, a = 20.503 b = 41.798, c = -0.206, Joint = -2.586, Marginal = -2.921, MI = -9.951\n",
      "Pre-Train Epoch 7, Batch 79: Loss = 283.076, a = 39.523 b = 32.914, c = -0.206, Joint = -2.752, Marginal = -4.016, MI = -10.397\n",
      "Pre-Train Epoch 7, Batch 80: Loss = 273.086, a = 33.197 b = 39.129, c = -0.206, Joint = -2.845, Marginal = -4.677, MI = -7.136\n",
      "Pre-Train Epoch 7, Batch 81: Loss = 133.038, a = 39.257 b = 34.666, c = -5.512, Joint = -2.015, Marginal = -5.636, MI = -11.594\n",
      "Pre-Train Epoch 7, Batch 82: Loss = 559.586, a = 53.699 b = 36.852, c = -4.748, Joint = -3.110, Marginal = -12.552, MI = -12.231\n",
      "Pre-Train Epoch 7, Batch 83: Loss = 86.372, a = 32.905 b = 33.252, c = -0.205, Joint = -0.854, Marginal = -8.842, MI = -8.189\n",
      "Pre-Train Epoch 7, Batch 84: Loss = 164.004, a = 44.597 b = 31.483, c = -0.205, Joint = -0.796, Marginal = -4.639, MI = -8.906\n",
      "Pre-Train Epoch 7, Batch 85: Loss = 1001.095, a = 49.062 b = 34.293, c = -0.204, Joint = -6.285, Marginal = -17.410, MI = -13.930\n",
      "Pre-Train Epoch 7, Batch 86: Loss = 92.977, a = 32.451 b = 33.528, c = -0.204, Joint = -0.974, Marginal = -3.318, MI = -7.895\n",
      "Pre-Train Epoch 7, Batch 87: Loss = 258.809, a = 37.013 b = 30.612, c = -0.204, Joint = -1.425, Marginal = -2.822, MI = -10.460\n",
      "Pre-Train Epoch 7, Batch 88: Loss = 78.480, a = 21.933 b = 36.908, c = -0.204, Joint = -0.721, Marginal = -1.229, MI = -8.427\n",
      "Pre-Train Epoch 7, Batch 89: Loss = 95.232, a = 31.876 b = 35.080, c = -1.522, Joint = -0.199, Marginal = -3.270, MI = -7.168\n",
      "Pre-Train Epoch 7, Batch 90: Loss = 139.239, a = 54.691 b = 33.262, c = -0.203, Joint = -1.103, Marginal = -8.359, MI = -7.585\n",
      "Pre-Train Epoch 7, Batch 91: Loss = 83.027, a = 23.174 b = 35.145, c = -0.203, Joint = -1.030, Marginal = -2.297, MI = -10.338\n",
      "Pre-Train Epoch 7, Batch 92: Loss = 163.131, a = 69.407 b = 27.530, c = -0.203, Joint = -1.411, Marginal = -8.632, MI = -8.248\n",
      "Pre-Train Epoch 7, Batch 93: Loss = 599.458, a = 50.236 b = 43.849, c = -0.203, Joint = -5.902, Marginal = -8.579, MI = -16.300\n",
      "Pre-Train Epoch 7, Batch 94: Loss = 88.716, a = 33.916 b = 31.084, c = -0.203, Joint = 0.095, Marginal = -2.152, MI = -6.604\n",
      "Pre-Train Epoch 7, Batch 95: Loss = 440.328, a = 47.765 b = 39.816, c = -0.202, Joint = -4.129, Marginal = -5.042, MI = -12.514\n",
      "Pre-Train Epoch 7, Batch 96: Loss = 195.505, a = 31.731 b = 40.816, c = -0.202, Joint = -2.217, Marginal = -2.684, MI = -12.801\n",
      "Pre-Train Epoch 7, Batch 97: Loss = 89.144, a = 30.261 b = 36.350, c = -0.202, Joint = -0.792, Marginal = -1.056, MI = -7.705\n",
      "Pre-Train Epoch 7, Batch 98: Loss = 202.558, a = 52.616 b = 36.874, c = -6.567, Joint = -1.949, Marginal = -7.231, MI = -8.457\n",
      "Pre-Train Epoch 7, Batch 99: Loss = 238.349, a = 44.698 b = 37.141, c = -0.250, Joint = -1.836, Marginal = -11.246, MI = -9.940\n",
      "Pre-Train Epoch 7, Batch 100: Loss = 81.695, a = 23.615 b = 35.841, c = -0.201, Joint = -0.147, Marginal = -0.759, MI = -7.766\n",
      "Pre-Train Epoch 7, Batch 101: Loss = 648.174, a = 77.666 b = 35.587, c = -5.280, Joint = -4.244, Marginal = -16.847, MI = -12.326\n",
      "Pre-Train Epoch 7, Batch 102: Loss = 303.836, a = 40.244 b = 33.517, c = -0.201, Joint = -2.019, Marginal = -2.832, MI = -11.955\n",
      "Pre-Train Epoch 7, Batch 103: Loss = 80.532, a = 27.786 b = 35.161, c = -0.201, Joint = -0.925, Marginal = -5.499, MI = -5.718\n",
      "Pre-Train Epoch 7, Batch 104: Loss = 358.673, a = 45.427 b = 33.304, c = -0.201, Joint = -3.514, Marginal = -9.395, MI = -12.036\n",
      "Pre-Train Epoch 7, Batch 105: Loss = 259.041, a = 38.728 b = 31.619, c = -0.200, Joint = -2.698, Marginal = -4.855, MI = -10.859\n",
      "Pre-Train Epoch 7, Batch 106: Loss = 317.458, a = 56.039 b = 36.226, c = -0.200, Joint = -3.225, Marginal = -9.882, MI = -13.429\n",
      "Pre-Train Epoch 7, Batch 107: Loss = 103.735, a = 31.532 b = 34.444, c = -3.287, Joint = -0.811, Marginal = -5.949, MI = -8.533\n",
      "Pre-Train Epoch 7, Batch 108: Loss = 543.722, a = 23.286 b = 41.100, c = -0.200, Joint = -4.209, Marginal = -3.760, MI = -8.846\n",
      "Pre-Train Epoch 7, Batch 109: Loss = 1406.581, a = 34.243 b = 31.544, c = -1.785, Joint = -6.189, Marginal = -15.063, MI = -14.810\n",
      "Pre-Train Epoch 7, Batch 110: Loss = 1135.656, a = 47.745 b = 36.035, c = -0.199, Joint = -4.579, Marginal = -10.558, MI = -13.161\n",
      "Pre-Train Epoch 7, Batch 111: Loss = 2407.257, a = 59.039 b = 40.711, c = -0.199, Joint = -10.580, Marginal = -13.047, MI = -17.858\n",
      "Pre-Train Epoch 7, Batch 112: Loss = 165.884, a = 34.479 b = 30.682, c = -0.393, Joint = -1.559, Marginal = -9.130, MI = -8.093\n",
      "Pre-Train Epoch 7, Batch 113: Loss = 165.186, a = 45.429 b = 40.148, c = -0.199, Joint = -2.414, Marginal = -7.418, MI = -9.345\n",
      "Pre-Train Epoch 7, Batch 114: Loss = 109.395, a = 21.046 b = 34.469, c = -0.198, Joint = -1.811, Marginal = -1.631, MI = -7.823\n",
      "Pre-Train Epoch 7, Batch 115: Loss = 870.856, a = 45.210 b = 32.157, c = -0.198, Joint = -4.294, Marginal = -13.656, MI = -12.273\n",
      "Pre-Train Epoch 7, Batch 116: Loss = 87.260, a = 25.519 b = 39.660, c = -0.198, Joint = -0.854, Marginal = -1.251, MI = -8.277\n",
      "Pre-Train Epoch 7, Batch 117: Loss = 288.964, a = 36.116 b = 34.828, c = -0.198, Joint = -2.970, Marginal = -3.872, MI = -11.903\n",
      "Pre-Train Epoch 7, Batch 118: Loss = 103.501, a = 27.709 b = 41.373, c = -0.198, Joint = -1.325, Marginal = -3.226, MI = -7.646\n",
      "Pre-Train Epoch 7, Batch 119: Loss = 91.933, a = 35.923 b = 33.190, c = -0.197, Joint = -1.114, Marginal = -2.894, MI = -9.622\n",
      "Pre-Train Epoch 7, Batch 120: Loss = 69.395, a = 21.474 b = 30.264, c = -6.769, Joint = -0.594, Marginal = -0.820, MI = -7.255\n",
      "Pre-Train Epoch 7, Batch 121: Loss = 67.798, a = 21.508 b = 29.344, c = -0.197, Joint = -0.285, Marginal = -1.337, MI = -6.175\n",
      "Pre-Train Epoch 7, Batch 122: Loss = 94.631, a = 38.027 b = 32.876, c = -5.425, Joint = -0.328, Marginal = -1.958, MI = -6.567\n",
      "Pre-Train Epoch 7, Batch 123: Loss = 112.305, a = 39.288 b = 34.114, c = -1.565, Joint = 0.549, Marginal = -1.029, MI = -10.148\n",
      "Pre-Train Epoch 7, Batch 124: Loss = 88.915, a = 32.781 b = 28.594, c = -0.196, Joint = 0.566, Marginal = -0.727, MI = -6.748\n",
      "Pre-Train Epoch 7, Batch 125: Loss = 567.496, a = 47.798 b = 28.461, c = -0.684, Joint = -4.937, Marginal = -15.941, MI = -11.342\n",
      "Pre-Train Epoch 7, Batch 126: Loss = 76.880, a = 22.071 b = 34.485, c = -0.196, Joint = -0.379, Marginal = -1.666, MI = -6.815\n",
      "Pre-Train Epoch 7, Batch 127: Loss = 93.823, a = 29.292 b = 41.017, c = -0.196, Joint = -1.341, Marginal = -2.721, MI = -6.009\n",
      "Pre-Train Epoch 7, Batch 128: Loss = 194.881, a = 26.367 b = 28.082, c = -1.135, Joint = -1.727, Marginal = -2.735, MI = -8.220\n",
      "Pre-Train Epoch 7, Batch 129: Loss = 89.987, a = 34.845 b = 34.315, c = -0.195, Joint = -1.096, Marginal = -1.782, MI = -12.086\n",
      "Pre-Train Epoch 7, Batch 130: Loss = 95.223, a = 26.784 b = 40.231, c = -0.195, Joint = -0.841, Marginal = -1.461, MI = -7.135\n",
      "Pre-Train Epoch 7, Batch 131: Loss = 111.571, a = 21.652 b = 34.611, c = -3.497, Joint = -1.346, Marginal = -2.669, MI = -8.636\n",
      "Pre-Train Epoch 7, Batch 132: Loss = 230.665, a = 25.299 b = 33.268, c = -0.195, Joint = -1.951, Marginal = -2.575, MI = -8.199\n",
      "Pre-Train Epoch 7, Batch 133: Loss = 1388.801, a = 44.393 b = 37.953, c = -0.195, Joint = -6.912, Marginal = -9.882, MI = -11.190\n",
      "Pre-Train Epoch 7, Batch 134: Loss = 184.383, a = 38.022 b = 25.526, c = -0.194, Joint = -0.682, Marginal = -4.100, MI = -8.939\n",
      "Pre-Train Epoch 7, Batch 135: Loss = 83.384, a = 27.862 b = 31.863, c = -0.194, Joint = -0.096, Marginal = -2.922, MI = -9.445\n",
      "Pre-Train Epoch 7, Batch 136: Loss = 94.834, a = 24.666 b = 45.374, c = -0.194, Joint = -1.882, Marginal = -1.726, MI = -8.062\n",
      "Pre-Train Epoch 7, Batch 137: Loss = 76.180, a = 26.353 b = 31.966, c = -0.194, Joint = -0.319, Marginal = -2.564, MI = -7.610\n",
      "Pre-Train Epoch 7, Batch 138: Loss = 151.074, a = 23.651 b = 32.579, c = -0.194, Joint = -2.078, Marginal = -3.524, MI = -7.439\n",
      "Pre-Train Epoch 7, Batch 139: Loss = 112.306, a = 30.696 b = 42.914, c = -2.280, Joint = -1.913, Marginal = -3.559, MI = -9.040\n",
      "Pre-Train Epoch 7, Batch 140: Loss = 2292.207, a = 35.402 b = 38.309, c = -3.120, Joint = -8.273, Marginal = -22.334, MI = -12.429\n",
      "Pre-Train Epoch 7, Batch 141: Loss = 108.697, a = 41.386 b = 35.657, c = -0.193, Joint = -0.141, Marginal = -3.764, MI = -6.271\n",
      "Pre-Train Epoch 7, Batch 142: Loss = 154.171, a = 39.500 b = 30.752, c = -0.193, Joint = -0.514, Marginal = -3.474, MI = -8.049\n",
      "Pre-Train Epoch 7, Batch 143: Loss = 840.416, a = 38.018 b = 39.938, c = -0.193, Joint = -3.677, Marginal = -5.401, MI = -10.674\n",
      "Pre-Train Epoch 7, Batch 144: Loss = 143.369, a = 28.349 b = 39.879, c = -0.192, Joint = -1.873, Marginal = -6.264, MI = -6.058\n",
      "Pre-Train Epoch 7, Batch 145: Loss = 93.692, a = 29.901 b = 32.876, c = -0.192, Joint = -1.337, Marginal = -2.626, MI = -10.429\n",
      "Pre-Train Epoch 7, Batch 146: Loss = 82.678, a = 27.976 b = 35.704, c = -0.192, Joint = -0.479, Marginal = -1.881, MI = -9.121\n",
      "Pre-Train Epoch 7, Batch 147: Loss = 124.582, a = 41.699 b = 40.316, c = -0.192, Joint = -0.758, Marginal = -4.453, MI = -7.834\n",
      "Pre-Train Epoch 7, Batch 148: Loss = 124.845, a = 30.494 b = 32.991, c = -0.192, Joint = -1.572, Marginal = -4.895, MI = -7.556\n",
      "Pre-Train Epoch 7, Batch 149: Loss = 60.835, a = 18.044 b = 25.275, c = -4.299, Joint = -0.405, Marginal = -0.532, MI = -8.526\n",
      "Pre-Train Epoch 7, Batch 150: Loss = 97.431, a = 30.472 b = 35.397, c = -5.602, Joint = -1.065, Marginal = -7.514, MI = -8.448\n",
      "\n",
      "Start of epoch 8\n",
      "Pre-Train Epoch 8, Batch 0: Loss = 191.945, a = 27.373 b = 34.959, c = -0.191, Joint = -1.985, Marginal = -4.775, MI = -6.721\n",
      "Pre-Train Epoch 8, Batch 1: Loss = 2335.171, a = 46.278 b = 39.058, c = -0.191, Joint = -7.478, Marginal = -18.474, MI = -15.956\n",
      "Pre-Train Epoch 8, Batch 2: Loss = 90.396, a = 23.942 b = 39.655, c = -0.191, Joint = -0.830, Marginal = -2.851, MI = -5.741\n",
      "Pre-Train Epoch 8, Batch 3: Loss = 113.946, a = 21.254 b = 34.921, c = -0.190, Joint = -1.634, Marginal = -2.707, MI = -8.166\n",
      "Pre-Train Epoch 8, Batch 4: Loss = 69.610, a = 21.778 b = 28.775, c = -0.190, Joint = -0.269, Marginal = -1.091, MI = -8.168\n",
      "Pre-Train Epoch 8, Batch 5: Loss = 81.477, a = 28.323 b = 32.818, c = -0.190, Joint = -1.090, Marginal = -2.338, MI = -4.368\n",
      "Pre-Train Epoch 8, Batch 6: Loss = 91.598, a = 27.306 b = 34.421, c = -0.825, Joint = -0.258, Marginal = -1.248, MI = -8.032\n",
      "Pre-Train Epoch 8, Batch 7: Loss = 77.267, a = 16.722 b = 40.079, c = -0.190, Joint = -1.022, Marginal = -0.836, MI = -8.854\n",
      "Pre-Train Epoch 8, Batch 8: Loss = 82.188, a = 19.312 b = 38.686, c = -0.189, Joint = -1.038, Marginal = -1.587, MI = -6.687\n",
      "Pre-Train Epoch 8, Batch 9: Loss = 432.310, a = 55.754 b = 32.526, c = -0.189, Joint = -4.666, Marginal = -11.219, MI = -10.724\n",
      "Pre-Train Epoch 8, Batch 10: Loss = 101.449, a = 28.029 b = 42.050, c = -5.523, Joint = -1.589, Marginal = -5.523, MI = -4.979\n",
      "Pre-Train Epoch 8, Batch 11: Loss = 79.735, a = 23.359 b = 34.167, c = -0.189, Joint = -0.449, Marginal = -3.088, MI = -6.025\n",
      "Pre-Train Epoch 8, Batch 12: Loss = 153.099, a = 34.214 b = 27.218, c = -0.189, Joint = -1.717, Marginal = -4.050, MI = -8.443\n",
      "Pre-Train Epoch 8, Batch 13: Loss = 87.294, a = 29.232 b = 31.147, c = -0.188, Joint = -0.118, Marginal = -2.070, MI = -5.700\n",
      "Pre-Train Epoch 8, Batch 14: Loss = 752.896, a = 25.632 b = 28.969, c = -2.180, Joint = -3.687, Marginal = -5.841, MI = -8.588\n",
      "Pre-Train Epoch 8, Batch 15: Loss = 298.579, a = 26.749 b = 34.777, c = -0.188, Joint = -3.119, Marginal = -3.530, MI = -9.917\n",
      "Pre-Train Epoch 8, Batch 16: Loss = 1011.875, a = 31.161 b = 38.906, c = -0.188, Joint = -5.724, Marginal = -4.823, MI = -13.908\n",
      "Pre-Train Epoch 8, Batch 17: Loss = 594.451, a = 29.700 b = 39.430, c = -0.939, Joint = -4.410, Marginal = -8.996, MI = -6.788\n",
      "Pre-Train Epoch 8, Batch 18: Loss = 93.257, a = 31.452 b = 29.506, c = -0.187, Joint = -0.884, Marginal = -2.434, MI = -8.235\n",
      "Pre-Train Epoch 8, Batch 19: Loss = 95.201, a = 30.967 b = 30.244, c = -1.521, Joint = -0.359, Marginal = -6.647, MI = -6.920\n",
      "Pre-Train Epoch 8, Batch 20: Loss = 63.387, a = 16.197 b = 30.620, c = -0.187, Joint = -0.432, Marginal = -1.009, MI = -6.795\n",
      "Pre-Train Epoch 8, Batch 21: Loss = 130.449, a = 39.754 b = 32.568, c = -0.187, Joint = -1.752, Marginal = -7.133, MI = -9.097\n",
      "Pre-Train Epoch 8, Batch 22: Loss = 72.258, a = 25.143 b = 28.514, c = -5.153, Joint = -0.057, Marginal = -1.243, MI = -7.630\n",
      "Pre-Train Epoch 8, Batch 23: Loss = 159.465, a = 30.822 b = 32.129, c = -0.187, Joint = -2.138, Marginal = -3.872, MI = -9.460\n",
      "Pre-Train Epoch 8, Batch 24: Loss = 72.282, a = 24.169 b = 28.799, c = -0.186, Joint = 0.131, Marginal = -1.678, MI = -5.537\n",
      "Pre-Train Epoch 8, Batch 25: Loss = 822.832, a = 34.578 b = 35.953, c = -0.186, Joint = -4.158, Marginal = -3.400, MI = -11.545\n",
      "Pre-Train Epoch 8, Batch 26: Loss = 120.343, a = 33.414 b = 40.714, c = -0.186, Joint = -1.357, Marginal = -4.977, MI = -8.686\n",
      "Pre-Train Epoch 8, Batch 27: Loss = 108.369, a = 31.619 b = 40.547, c = -0.186, Joint = -1.809, Marginal = -4.707, MI = -6.025\n",
      "Pre-Train Epoch 8, Batch 28: Loss = 183.136, a = 33.317 b = 27.401, c = -2.352, Joint = -1.262, Marginal = -3.677, MI = -6.092\n",
      "Pre-Train Epoch 8, Batch 29: Loss = 76.139, a = 18.212 b = 37.592, c = -0.185, Joint = -1.219, Marginal = -2.680, MI = -4.600\n",
      "Pre-Train Epoch 8, Batch 30: Loss = 165.274, a = 28.469 b = 43.216, c = -0.185, Joint = -2.994, Marginal = -4.871, MI = -6.887\n",
      "Pre-Train Epoch 8, Batch 31: Loss = 305.074, a = 34.062 b = 38.955, c = -9.801, Joint = -4.164, Marginal = -3.920, MI = -9.299\n",
      "Pre-Train Epoch 8, Batch 32: Loss = 74.075, a = 19.742 b = 34.575, c = -0.185, Joint = -1.261, Marginal = -4.235, MI = -5.243\n",
      "Pre-Train Epoch 8, Batch 33: Loss = 757.003, a = 29.988 b = 30.998, c = -0.185, Joint = -4.822, Marginal = -6.150, MI = -8.527\n",
      "Pre-Train Epoch 8, Batch 34: Loss = 80.120, a = 22.435 b = 35.057, c = -5.180, Joint = -0.164, Marginal = -3.075, MI = -6.861\n",
      "Pre-Train Epoch 8, Batch 35: Loss = 87.050, a = 25.779 b = 37.605, c = -0.184, Joint = -1.291, Marginal = -2.800, MI = -7.606\n",
      "Pre-Train Epoch 8, Batch 36: Loss = 73.621, a = 17.722 b = 37.891, c = -2.811, Joint = -0.973, Marginal = -1.177, MI = -6.731\n",
      "Pre-Train Epoch 8, Batch 37: Loss = 102.963, a = 26.649 b = 41.378, c = -0.184, Joint = -1.986, Marginal = -3.928, MI = -9.447\n",
      "Pre-Train Epoch 8, Batch 38: Loss = 58.632, a = 12.798 b = 33.121, c = -0.184, Joint = -0.821, Marginal = -1.415, MI = -6.610\n",
      "Pre-Train Epoch 8, Batch 39: Loss = 78.724, a = 20.626 b = 40.037, c = -0.184, Joint = -1.089, Marginal = -3.096, MI = -4.578\n",
      "Pre-Train Epoch 8, Batch 40: Loss = 11123.459, a = 59.328 b = 28.811, c = -0.183, Joint = -16.566, Marginal = -40.287, MI = -21.883\n",
      "Pre-Train Epoch 8, Batch 41: Loss = 90.769, a = 35.298 b = 37.293, c = -0.183, Joint = -0.636, Marginal = -9.175, MI = -7.240\n",
      "Pre-Train Epoch 8, Batch 42: Loss = 72.160, a = 20.089 b = 37.138, c = -0.214, Joint = -0.970, Marginal = -2.176, MI = -4.758\n",
      "Pre-Train Epoch 8, Batch 43: Loss = 114.848, a = 31.266 b = 39.189, c = -6.060, Joint = -2.091, Marginal = -4.598, MI = -5.138\n",
      "Pre-Train Epoch 8, Batch 44: Loss = 120.833, a = 35.238 b = 34.862, c = -0.182, Joint = -1.751, Marginal = -6.308, MI = -8.890\n",
      "Pre-Train Epoch 8, Batch 45: Loss = 96.861, a = 24.547 b = 34.969, c = -2.455, Joint = -0.885, Marginal = -0.974, MI = -7.766\n",
      "Pre-Train Epoch 8, Batch 46: Loss = 84.749, a = 28.580 b = 38.418, c = -0.182, Joint = -1.109, Marginal = -2.197, MI = -12.862\n",
      "Pre-Train Epoch 8, Batch 47: Loss = 2749.629, a = 36.898 b = 30.085, c = -0.181, Joint = -7.269, Marginal = -10.697, MI = -13.863\n",
      "Pre-Train Epoch 8, Batch 48: Loss = 171.620, a = 49.101 b = 34.082, c = -0.181, Joint = -2.282, Marginal = -8.517, MI = -7.335\n",
      "Pre-Train Epoch 8, Batch 49: Loss = 88.994, a = 24.356 b = 34.515, c = -0.181, Joint = -0.577, Marginal = -3.974, MI = -6.556\n",
      "Pre-Train Epoch 8, Batch 50: Loss = 117.889, a = 29.496 b = 28.506, c = -0.180, Joint = -2.215, Marginal = -4.046, MI = -8.478\n",
      "Pre-Train Epoch 8, Batch 51: Loss = 122.141, a = 39.788 b = 33.948, c = -1.546, Joint = -1.240, Marginal = -8.386, MI = -8.030\n",
      "Pre-Train Epoch 8, Batch 52: Loss = 63.194, a = 18.609 b = 30.059, c = -0.493, Joint = -0.325, Marginal = -1.663, MI = -5.964\n",
      "Pre-Train Epoch 8, Batch 53: Loss = 95.693, a = 34.689 b = 24.453, c = -2.691, Joint = 0.287, Marginal = -0.817, MI = -6.721\n",
      "Pre-Train Epoch 8, Batch 54: Loss = 88.949, a = 20.302 b = 39.260, c = -0.180, Joint = -1.758, Marginal = -2.792, MI = -6.316\n",
      "Pre-Train Epoch 8, Batch 55: Loss = 6900.830, a = 48.827 b = 35.832, c = -0.179, Joint = -11.705, Marginal = -16.493, MI = -16.244\n",
      "Pre-Train Epoch 8, Batch 56: Loss = 143.852, a = 27.308 b = 35.524, c = -2.425, Joint = -1.973, Marginal = -3.234, MI = -6.292\n",
      "Pre-Train Epoch 8, Batch 57: Loss = 74.514, a = 22.139 b = 34.036, c = -5.603, Joint = -0.507, Marginal = -1.120, MI = -7.709\n",
      "Pre-Train Epoch 8, Batch 58: Loss = 114.786, a = 23.941 b = 38.084, c = -4.156, Joint = -1.457, Marginal = -2.843, MI = -5.464\n",
      "Pre-Train Epoch 8, Batch 59: Loss = 73.511, a = 19.603 b = 31.350, c = -0.400, Joint = -0.326, Marginal = -1.149, MI = -6.357\n",
      "Pre-Train Epoch 8, Batch 60: Loss = 2600.963, a = 43.052 b = 33.564, c = -0.178, Joint = -8.236, Marginal = -11.407, MI = -16.287\n",
      "Pre-Train Epoch 8, Batch 61: Loss = 90.374, a = 30.530 b = 32.866, c = -0.178, Joint = -0.385, Marginal = -2.663, MI = -9.196\n",
      "Pre-Train Epoch 8, Batch 62: Loss = 384.883, a = 40.886 b = 31.352, c = -4.934, Joint = -3.773, Marginal = -9.726, MI = -10.742\n",
      "Pre-Train Epoch 8, Batch 63: Loss = 141.739, a = 35.087 b = 29.710, c = -0.177, Joint = -1.654, Marginal = -6.882, MI = -6.190\n",
      "Pre-Train Epoch 8, Batch 64: Loss = 77.868, a = 20.831 b = 39.214, c = -0.177, Joint = -1.049, Marginal = -6.329, MI = -4.962\n",
      "Pre-Train Epoch 8, Batch 65: Loss = 1186.645, a = 59.854 b = 28.002, c = -1.159, Joint = -5.609, Marginal = -17.832, MI = -12.621\n",
      "Pre-Train Epoch 8, Batch 66: Loss = 80.744, a = 20.915 b = 41.558, c = -0.176, Joint = -1.336, Marginal = -2.092, MI = -7.789\n",
      "Pre-Train Epoch 8, Batch 67: Loss = 97.505, a = 31.251 b = 44.551, c = -0.176, Joint = -1.598, Marginal = -7.750, MI = -11.393\n",
      "Pre-Train Epoch 8, Batch 68: Loss = 121.240, a = 40.339 b = 30.175, c = -0.176, Joint = -0.372, Marginal = -6.117, MI = -6.167\n",
      "Pre-Train Epoch 8, Batch 69: Loss = 162.862, a = 34.390 b = 32.066, c = -0.176, Joint = -1.990, Marginal = -4.313, MI = -10.177\n",
      "Pre-Train Epoch 8, Batch 70: Loss = 315.136, a = 25.519 b = 40.547, c = -2.579, Joint = -3.351, Marginal = -6.509, MI = -5.138\n",
      "Pre-Train Epoch 8, Batch 71: Loss = 945.905, a = 45.980 b = 46.514, c = -0.175, Joint = -5.261, Marginal = -17.307, MI = -12.085\n",
      "Pre-Train Epoch 8, Batch 72: Loss = 302.931, a = 55.499 b = 28.315, c = -0.175, Joint = -2.387, Marginal = -6.865, MI = -9.081\n",
      "Pre-Train Epoch 8, Batch 73: Loss = 200.457, a = 35.865 b = 26.058, c = -0.175, Joint = -1.790, Marginal = -4.304, MI = -8.871\n",
      "Pre-Train Epoch 8, Batch 74: Loss = 88.050, a = 34.481 b = 33.545, c = -0.576, Joint = -0.594, Marginal = -5.707, MI = -7.878\n",
      "Pre-Train Epoch 8, Batch 75: Loss = 103.372, a = 35.636 b = 29.106, c = -0.174, Joint = -0.133, Marginal = 0.153, MI = -7.884\n",
      "Pre-Train Epoch 8, Batch 76: Loss = 201.263, a = 29.184 b = 34.416, c = -0.174, Joint = -2.215, Marginal = -4.679, MI = -7.074\n",
      "Pre-Train Epoch 8, Batch 77: Loss = 547.349, a = 40.348 b = 33.224, c = -0.174, Joint = -4.622, Marginal = -13.330, MI = -9.404\n",
      "Pre-Train Epoch 8, Batch 78: Loss = 82.905, a = 15.114 b = 41.269, c = -0.174, Joint = -2.279, Marginal = -2.913, MI = -7.932\n",
      "Pre-Train Epoch 8, Batch 79: Loss = 223.805, a = 30.556 b = 32.495, c = -0.173, Joint = -2.502, Marginal = -6.744, MI = -7.323\n",
      "Pre-Train Epoch 8, Batch 80: Loss = 224.920, a = 25.478 b = 38.605, c = -0.173, Joint = -2.495, Marginal = -3.258, MI = -9.533\n",
      "Pre-Train Epoch 8, Batch 81: Loss = 108.073, a = 31.005 b = 34.267, c = -5.347, Joint = -1.689, Marginal = -5.871, MI = -5.623\n",
      "Pre-Train Epoch 8, Batch 82: Loss = 434.150, a = 42.623 b = 36.323, c = -4.576, Joint = -2.710, Marginal = -14.455, MI = -6.422\n",
      "Pre-Train Epoch 8, Batch 83: Loss = 73.548, a = 25.571 b = 32.873, c = -0.173, Joint = -0.679, Marginal = -7.082, MI = -3.605\n",
      "Pre-Train Epoch 8, Batch 84: Loss = 135.334, a = 35.244 b = 31.131, c = -0.172, Joint = -0.622, Marginal = -1.498, MI = -10.942\n",
      "Pre-Train Epoch 8, Batch 85: Loss = 792.893, a = 39.599 b = 33.894, c = -0.172, Joint = -5.621, Marginal = -10.368, MI = -11.225\n",
      "Pre-Train Epoch 8, Batch 86: Loss = 79.647, a = 24.968 b = 33.134, c = -0.172, Joint = -0.797, Marginal = -3.188, MI = -6.746\n",
      "Pre-Train Epoch 8, Batch 87: Loss = 204.020, a = 28.702 b = 30.419, c = -0.172, Joint = -1.278, Marginal = -2.482, MI = -8.463\n",
      "Pre-Train Epoch 8, Batch 88: Loss = 67.201, a = 16.054 b = 36.514, c = -0.172, Joint = -0.639, Marginal = -0.831, MI = -6.306\n",
      "Pre-Train Epoch 8, Batch 89: Loss = 81.075, a = 24.156 b = 34.674, c = -1.422, Joint = -0.153, Marginal = -2.105, MI = -8.148\n",
      "Pre-Train Epoch 8, Batch 90: Loss = 112.778, a = 43.457 b = 32.832, c = -0.171, Joint = -0.849, Marginal = -5.874, MI = -5.818\n",
      "Pre-Train Epoch 8, Batch 91: Loss = 70.317, a = 16.915 b = 34.743, c = -0.171, Joint = -0.973, Marginal = -1.540, MI = -7.308\n",
      "Pre-Train Epoch 8, Batch 92: Loss = 131.671, a = 56.075 b = 27.244, c = -0.171, Joint = -1.091, Marginal = -10.260, MI = -10.245\n",
      "Pre-Train Epoch 8, Batch 93: Loss = 474.916, a = 39.549 b = 43.151, c = -0.171, Joint = -5.153, Marginal = -6.410, MI = -9.062\n",
      "Pre-Train Epoch 8, Batch 94: Loss = 76.115, a = 25.922 b = 30.767, c = -0.171, Joint = 0.087, Marginal = -0.664, MI = -8.131\n",
      "Pre-Train Epoch 8, Batch 95: Loss = 352.071, a = 37.192 b = 39.197, c = -0.170, Joint = -3.620, Marginal = -6.673, MI = -8.319\n",
      "Pre-Train Epoch 8, Batch 96: Loss = 162.495, a = 24.080 b = 40.308, c = -0.170, Joint = -1.929, Marginal = -3.387, MI = -7.929\n",
      "Pre-Train Epoch 8, Batch 97: Loss = 76.632, a = 22.778 b = 35.934, c = -0.170, Joint = -0.670, Marginal = -2.117, MI = -5.594\n",
      "Pre-Train Epoch 8, Batch 98: Loss = 160.979, a = 41.087 b = 36.281, c = -6.382, Joint = -1.658, Marginal = -5.315, MI = -7.908\n",
      "Pre-Train Epoch 8, Batch 99: Loss = 191.427, a = 34.014 b = 36.550, c = -0.170, Joint = -1.640, Marginal = -4.590, MI = -11.320\n",
      "Pre-Train Epoch 8, Batch 100: Loss = 68.834, a = 17.018 b = 35.470, c = -0.170, Joint = -0.166, Marginal = -0.541, MI = -6.828\n",
      "Pre-Train Epoch 8, Batch 101: Loss = 528.092, a = 62.376 b = 34.956, c = -5.103, Joint = -3.597, Marginal = -9.626, MI = -13.444\n",
      "Pre-Train Epoch 8, Batch 102: Loss = 251.254, a = 30.720 b = 33.029, c = -0.169, Joint = -1.826, Marginal = -4.241, MI = -7.526\n",
      "Pre-Train Epoch 8, Batch 103: Loss = 68.664, a = 20.903 b = 34.722, c = -0.169, Joint = -0.810, Marginal = -5.077, MI = -5.754\n",
      "Pre-Train Epoch 8, Batch 104: Loss = 285.157, a = 35.672 b = 32.868, c = -0.169, Joint = -3.062, Marginal = -5.244, MI = -8.785\n",
      "Pre-Train Epoch 8, Batch 105: Loss = 209.895, a = 29.971 b = 31.237, c = -0.169, Joint = -2.341, Marginal = -5.967, MI = -8.537\n",
      "Pre-Train Epoch 8, Batch 106: Loss = 253.854, a = 44.739 b = 35.722, c = -0.168, Joint = -2.719, Marginal = -7.412, MI = -11.203\n",
      "Pre-Train Epoch 8, Batch 107: Loss = 86.755, a = 23.840 b = 34.009, c = -3.189, Joint = -0.749, Marginal = -2.019, MI = -6.929\n",
      "Pre-Train Epoch 8, Batch 108: Loss = 441.583, a = 17.726 b = 40.581, c = -0.168, Joint = -3.711, Marginal = -2.455, MI = -9.926\n",
      "Pre-Train Epoch 8, Batch 109: Loss = 1125.175, a = 27.397 b = 31.245, c = -1.696, Joint = -5.522, Marginal = -14.583, MI = -7.611\n",
      "Pre-Train Epoch 8, Batch 110: Loss = 910.738, a = 37.611 b = 35.592, c = -0.168, Joint = -4.051, Marginal = -4.341, MI = -11.432\n",
      "Pre-Train Epoch 8, Batch 111: Loss = 1951.170, a = 47.443 b = 40.107, c = -0.168, Joint = -9.383, Marginal = -9.382, MI = -16.905\n",
      "Pre-Train Epoch 8, Batch 112: Loss = 133.475, a = 26.551 b = 30.342, c = -0.328, Joint = -1.338, Marginal = -1.266, MI = -6.597\n",
      "Pre-Train Epoch 8, Batch 113: Loss = 136.253, a = 35.537 b = 39.556, c = -0.167, Joint = -1.991, Marginal = -2.326, MI = -14.504\n",
      "Pre-Train Epoch 8, Batch 114: Loss = 92.245, a = 15.516 b = 34.146, c = -0.167, Joint = -1.623, Marginal = -1.130, MI = -8.464\n",
      "Pre-Train Epoch 8, Batch 115: Loss = 688.105, a = 35.993 b = 31.799, c = -0.167, Joint = -3.790, Marginal = -9.120, MI = -10.853\n",
      "Pre-Train Epoch 8, Batch 116: Loss = 75.525, a = 19.356 b = 39.272, c = -0.166, Joint = -0.647, Marginal = -1.590, MI = -8.165\n",
      "Pre-Train Epoch 8, Batch 117: Loss = 240.577, a = 27.334 b = 34.325, c = -0.166, Joint = -2.664, Marginal = -2.853, MI = -5.700\n",
      "Pre-Train Epoch 8, Batch 118: Loss = 86.784, a = 20.791 b = 40.829, c = -0.166, Joint = -1.090, Marginal = -2.136, MI = -5.912\n",
      "Pre-Train Epoch 8, Batch 119: Loss = 77.890, a = 27.754 b = 32.750, c = -0.166, Joint = -0.857, Marginal = -5.017, MI = -6.816\n",
      "Pre-Train Epoch 8, Batch 120: Loss = 60.299, a = 16.210 b = 30.066, c = -6.614, Joint = -0.476, Marginal = -1.677, MI = -7.201\n",
      "Pre-Train Epoch 8, Batch 121: Loss = 57.785, a = 15.652 b = 29.068, c = -0.165, Joint = -0.316, Marginal = -1.662, MI = -6.717\n",
      "Pre-Train Epoch 8, Batch 122: Loss = 80.328, a = 29.145 b = 32.494, c = -5.255, Joint = -0.246, Marginal = -1.253, MI = -7.801\n",
      "Pre-Train Epoch 8, Batch 123: Loss = 93.297, a = 29.363 b = 33.591, c = -1.498, Joint = 0.448, Marginal = -2.622, MI = -6.094\n",
      "Pre-Train Epoch 8, Batch 124: Loss = 74.596, a = 24.622 b = 28.350, c = -0.165, Joint = 0.479, Marginal = -0.312, MI = -5.367\n",
      "Pre-Train Epoch 8, Batch 125: Loss = 450.006, a = 38.112 b = 28.131, c = -0.670, Joint = -4.445, Marginal = -6.621, MI = -9.234\n",
      "Pre-Train Epoch 8, Batch 126: Loss = 65.659, a = 16.104 b = 34.152, c = -0.165, Joint = -0.380, Marginal = -2.272, MI = -7.834\n",
      "Pre-Train Epoch 8, Batch 127: Loss = 80.183, a = 22.174 b = 40.460, c = -0.164, Joint = -1.114, Marginal = -1.739, MI = -6.124\n",
      "Pre-Train Epoch 8, Batch 128: Loss = 158.448, a = 19.940 b = 27.838, c = -1.087, Joint = -1.608, Marginal = -3.230, MI = -8.271\n",
      "Pre-Train Epoch 8, Batch 129: Loss = 76.172, a = 26.815 b = 33.918, c = -0.164, Joint = -0.914, Marginal = -2.841, MI = -2.511\n",
      "Pre-Train Epoch 8, Batch 130: Loss = 81.477, a = 19.962 b = 39.751, c = -0.164, Joint = -0.649, Marginal = -2.312, MI = -6.718\n",
      "Pre-Train Epoch 8, Batch 131: Loss = 93.455, a = 15.915 b = 34.301, c = -3.408, Joint = -1.220, Marginal = -1.009, MI = -4.821\n",
      "Pre-Train Epoch 8, Batch 132: Loss = 186.716, a = 19.119 b = 33.004, c = -0.164, Joint = -1.788, Marginal = -1.216, MI = -5.839\n",
      "Pre-Train Epoch 8, Batch 133: Loss = 1114.508, a = 35.558 b = 37.437, c = -0.163, Joint = -6.160, Marginal = -14.182, MI = -8.953\n",
      "Pre-Train Epoch 8, Batch 134: Loss = 148.745, a = 29.550 b = 25.360, c = -0.163, Joint = -0.632, Marginal = -4.903, MI = -8.161\n",
      "Pre-Train Epoch 8, Batch 135: Loss = 71.375, a = 20.985 b = 31.557, c = -0.163, Joint = -0.115, Marginal = -0.830, MI = -7.691\n",
      "Pre-Train Epoch 8, Batch 136: Loss = 80.181, a = 17.879 b = 44.726, c = -0.163, Joint = -1.675, Marginal = -1.756, MI = -7.054\n",
      "Pre-Train Epoch 8, Batch 137: Loss = 65.545, a = 19.832 b = 31.636, c = -0.163, Joint = -0.275, Marginal = -3.301, MI = -5.431\n",
      "Pre-Train Epoch 8, Batch 138: Loss = 124.275, a = 17.904 b = 32.292, c = -0.163, Joint = -1.869, Marginal = -2.969, MI = -5.119\n",
      "Pre-Train Epoch 8, Batch 139: Loss = 93.710, a = 23.133 b = 42.310, c = -2.189, Joint = -1.602, Marginal = -2.911, MI = -5.317\n",
      "Pre-Train Epoch 8, Batch 140: Loss = 1861.882, a = 28.115 b = 37.828, c = -3.002, Joint = -7.465, Marginal = -11.980, MI = -13.307\n",
      "Pre-Train Epoch 8, Batch 141: Loss = 91.142, a = 31.305 b = 35.086, c = -0.162, Joint = -0.131, Marginal = -3.313, MI = -7.117\n",
      "Pre-Train Epoch 8, Batch 142: Loss = 123.819, a = 30.665 b = 30.433, c = -0.162, Joint = -0.419, Marginal = -1.141, MI = -9.793\n",
      "Pre-Train Epoch 8, Batch 143: Loss = 687.595, a = 28.830 b = 39.323, c = -0.162, Joint = -3.357, Marginal = -5.895, MI = -9.768\n",
      "Pre-Train Epoch 8, Batch 144: Loss = 120.372, a = 21.555 b = 39.415, c = -0.161, Joint = -1.650, Marginal = -4.254, MI = -4.683\n",
      "Pre-Train Epoch 8, Batch 145: Loss = 78.319, a = 22.725 b = 32.458, c = -0.161, Joint = -1.160, Marginal = -4.997, MI = -3.657\n",
      "Pre-Train Epoch 8, Batch 146: Loss = 70.989, a = 20.867 b = 35.282, c = -0.161, Joint = -0.415, Marginal = -1.894, MI = -6.308\n",
      "Pre-Train Epoch 8, Batch 147: Loss = 101.773, a = 31.528 b = 39.644, c = -0.161, Joint = -0.625, Marginal = -5.866, MI = -6.161\n",
      "Pre-Train Epoch 8, Batch 148: Loss = 103.299, a = 22.950 b = 32.570, c = -0.161, Joint = -1.414, Marginal = -4.226, MI = -5.120\n",
      "Pre-Train Epoch 8, Batch 149: Loss = 52.698, a = 13.447 b = 25.188, c = -4.185, Joint = -0.401, Marginal = -2.342, MI = -5.002\n",
      "Pre-Train Epoch 8, Batch 150: Loss = 80.700, a = 23.214 b = 34.998, c = -5.492, Joint = -0.909, Marginal = -3.838, MI = -6.766\n",
      "\n",
      "Start of epoch 9\n",
      "Pre-Train Epoch 9, Batch 0: Loss = 156.905, a = 20.282 b = 34.497, c = -0.160, Joint = -1.807, Marginal = -0.968, MI = -8.528\n",
      "Pre-Train Epoch 9, Batch 1: Loss = 1886.025, a = 36.521 b = 38.429, c = -0.160, Joint = -6.673, Marginal = -16.262, MI = -10.566\n",
      "Pre-Train Epoch 9, Batch 2: Loss = 76.604, a = 17.398 b = 39.153, c = -0.160, Joint = -0.742, Marginal = -1.136, MI = -5.395\n",
      "Pre-Train Epoch 9, Batch 3: Loss = 95.286, a = 15.894 b = 34.575, c = -0.160, Joint = -1.457, Marginal = -3.747, MI = -4.686\n",
      "Pre-Train Epoch 9, Batch 4: Loss = 59.136, a = 16.184 b = 28.585, c = -0.159, Joint = -0.287, Marginal = -0.829, MI = -6.038\n",
      "Pre-Train Epoch 9, Batch 5: Loss = 70.047, a = 21.679 b = 32.484, c = -0.159, Joint = -0.934, Marginal = -1.162, MI = -6.156\n",
      "Pre-Train Epoch 9, Batch 6: Loss = 76.058, a = 20.215 b = 34.041, c = -0.687, Joint = -0.239, Marginal = -1.311, MI = -5.496\n",
      "Pre-Train Epoch 9, Batch 7: Loss = 66.446, a = 11.872 b = 39.650, c = -0.159, Joint = -0.891, Marginal = -1.003, MI = -7.097\n",
      "Pre-Train Epoch 9, Batch 8: Loss = 70.253, a = 13.937 b = 38.224, c = -0.159, Joint = -0.928, Marginal = -0.941, MI = -7.216\n",
      "Pre-Train Epoch 9, Batch 9: Loss = 341.020, a = 45.022 b = 32.065, c = -0.159, Joint = -4.080, Marginal = -14.789, MI = -10.876\n",
      "Pre-Train Epoch 9, Batch 10: Loss = 85.491, a = 21.337 b = 41.430, c = -5.340, Joint = -1.358, Marginal = -5.028, MI = -3.097\n",
      "Pre-Train Epoch 9, Batch 11: Loss = 69.401, a = 17.711 b = 33.835, c = -0.158, Joint = -0.375, Marginal = -2.869, MI = -5.207\n",
      "Pre-Train Epoch 9, Batch 12: Loss = 122.101, a = 26.816 b = 27.019, c = -0.158, Joint = -1.486, Marginal = -6.236, MI = -7.574\n",
      "Pre-Train Epoch 9, Batch 13: Loss = 72.444, a = 21.713 b = 30.781, c = -0.158, Joint = -0.106, Marginal = -1.169, MI = -4.913\n",
      "Pre-Train Epoch 9, Batch 14: Loss = 604.602, a = 20.271 b = 28.744, c = -2.083, Joint = -3.297, Marginal = -10.999, MI = -9.991\n",
      "Pre-Train Epoch 9, Batch 15: Loss = 245.126, a = 20.274 b = 34.417, c = -0.158, Joint = -2.800, Marginal = -1.393, MI = -9.279\n",
      "Pre-Train Epoch 9, Batch 16: Loss = 839.602, a = 24.226 b = 38.411, c = -0.157, Joint = -5.144, Marginal = -6.612, MI = -7.305\n",
      "Pre-Train Epoch 9, Batch 17: Loss = 474.301, a = 23.661 b = 39.020, c = -0.887, Joint = -3.850, Marginal = -12.158, MI = -5.596\n",
      "Pre-Train Epoch 9, Batch 18: Loss = 78.109, a = 24.787 b = 29.265, c = -0.157, Joint = -0.649, Marginal = -0.723, MI = -7.553\n",
      "Pre-Train Epoch 9, Batch 19: Loss = 78.121, a = 23.350 b = 29.901, c = -1.442, Joint = -0.368, Marginal = -0.129, MI = -7.148\n",
      "Pre-Train Epoch 9, Batch 20: Loss = 55.514, a = 11.971 b = 30.407, c = -0.157, Joint = -0.424, Marginal = -1.248, MI = -5.957\n",
      "Pre-Train Epoch 9, Batch 21: Loss = 107.386, a = 31.293 b = 32.154, c = -0.156, Joint = -1.493, Marginal = -4.421, MI = -9.433\n",
      "Pre-Train Epoch 9, Batch 22: Loss = 62.836, a = 19.131 b = 28.301, c = -4.993, Joint = -0.072, Marginal = -3.127, MI = -5.957\n",
      "Pre-Train Epoch 9, Batch 23: Loss = 131.887, a = 23.826 b = 31.790, c = -0.156, Joint = -1.919, Marginal = -3.140, MI = -6.679\n",
      "Pre-Train Epoch 9, Batch 24: Loss = 62.992, a = 18.389 b = 28.587, c = -0.156, Joint = 0.139, Marginal = -1.029, MI = -6.246\n",
      "Pre-Train Epoch 9, Batch 25: Loss = 678.089, a = 26.627 b = 35.467, c = -0.156, Joint = -3.770, Marginal = -6.049, MI = -9.363\n",
      "Pre-Train Epoch 9, Batch 26: Loss = 99.590, a = 24.776 b = 40.076, c = -0.156, Joint = -1.164, Marginal = -2.341, MI = -5.713\n",
      "Pre-Train Epoch 9, Batch 27: Loss = 91.419, a = 24.391 b = 39.987, c = -0.156, Joint = -1.508, Marginal = -7.519, MI = -5.390\n",
      "Pre-Train Epoch 9, Batch 28: Loss = 148.909, a = 25.771 b = 27.134, c = -2.319, Joint = -1.147, Marginal = -1.695, MI = -5.612\n",
      "Pre-Train Epoch 9, Batch 29: Loss = 65.530, a = 13.074 b = 37.153, c = -0.155, Joint = -1.113, Marginal = -1.432, MI = -6.297\n",
      "Pre-Train Epoch 9, Batch 30: Loss = 136.431, a = 21.857 b = 42.639, c = -0.155, Joint = -2.562, Marginal = -3.951, MI = -8.327\n",
      "Pre-Train Epoch 9, Batch 31: Loss = 242.554, a = 26.606 b = 38.408, c = -9.572, Joint = -3.567, Marginal = -8.456, MI = -8.611\n",
      "Pre-Train Epoch 9, Batch 32: Loss = 63.325, a = 14.865 b = 34.210, c = -0.155, Joint = -1.143, Marginal = -3.185, MI = -4.687\n",
      "Pre-Train Epoch 9, Batch 33: Loss = 608.078, a = 23.833 b = 30.665, c = -0.155, Joint = -4.312, Marginal = -8.837, MI = -7.741\n",
      "Pre-Train Epoch 9, Batch 34: Loss = 67.866, a = 16.081 b = 34.621, c = -5.034, Joint = -0.293, Marginal = -1.394, MI = -4.503\n",
      "Pre-Train Epoch 9, Batch 35: Loss = 74.334, a = 19.410 b = 37.118, c = -0.154, Joint = -1.099, Marginal = -3.479, MI = -4.567\n",
      "Pre-Train Epoch 9, Batch 36: Loss = 63.913, a = 12.778 b = 37.492, c = -2.699, Joint = -0.864, Marginal = -1.416, MI = -4.953\n",
      "Pre-Train Epoch 9, Batch 37: Loss = 87.712, a = 20.315 b = 40.847, c = -0.154, Joint = -1.683, Marginal = -5.543, MI = -6.354\n",
      "Pre-Train Epoch 9, Batch 38: Loss = 51.818, a = 9.201 b = 32.893, c = -0.154, Joint = -0.749, Marginal = -1.940, MI = -4.997\n",
      "Pre-Train Epoch 9, Batch 39: Loss = 68.649, a = 15.138 b = 39.541, c = -0.154, Joint = -0.959, Marginal = -3.300, MI = -6.229\n",
      "Pre-Train Epoch 9, Batch 40: Loss = 9187.993, a = 49.400 b = 28.356, c = -0.153, Joint = -15.030, Marginal = -25.360, MI = -17.992\n",
      "Pre-Train Epoch 9, Batch 41: Loss = 78.436, a = 27.265 b = 36.795, c = -0.153, Joint = -0.493, Marginal = -7.418, MI = -3.860\n",
      "Pre-Train Epoch 9, Batch 42: Loss = 63.177, a = 14.925 b = 36.706, c = -0.154, Joint = -0.852, Marginal = -2.275, MI = -6.017\n",
      "Pre-Train Epoch 9, Batch 43: Loss = 95.524, a = 24.028 b = 38.644, c = -5.889, Joint = -1.801, Marginal = -2.387, MI = -7.164\n",
      "Pre-Train Epoch 9, Batch 44: Loss = 97.835, a = 27.270 b = 34.403, c = -0.152, Joint = -1.525, Marginal = -1.945, MI = -7.870\n",
      "Pre-Train Epoch 9, Batch 45: Loss = 81.964, a = 18.223 b = 34.572, c = -2.370, Joint = -0.774, Marginal = -0.835, MI = -5.631\n",
      "Pre-Train Epoch 9, Batch 46: Loss = 73.493, a = 21.514 b = 37.908, c = -0.152, Joint = -0.941, Marginal = -3.462, MI = -4.370\n",
      "Pre-Train Epoch 9, Batch 47: Loss = 2287.050, a = 29.097 b = 29.744, c = -0.151, Joint = -6.613, Marginal = -10.739, MI = -13.893\n",
      "Pre-Train Epoch 9, Batch 48: Loss = 136.693, a = 38.724 b = 33.579, c = -0.151, Joint = -1.914, Marginal = -8.056, MI = -6.308\n",
      "Pre-Train Epoch 9, Batch 49: Loss = 76.822, a = 18.311 b = 34.175, c = -0.151, Joint = -0.497, Marginal = -1.888, MI = -6.862\n",
      "Pre-Train Epoch 9, Batch 50: Loss = 96.682, a = 22.956 b = 28.207, c = -0.151, Joint = -1.962, Marginal = -5.348, MI = -3.638\n",
      "Pre-Train Epoch 9, Batch 51: Loss = 100.030, a = 30.874 b = 33.496, c = -1.485, Joint = -1.026, Marginal = -3.581, MI = -7.858\n",
      "Pre-Train Epoch 9, Batch 52: Loss = 54.265, a = 13.359 b = 29.768, c = -0.429, Joint = -0.331, Marginal = -1.673, MI = -5.194\n",
      "Pre-Train Epoch 9, Batch 53: Loss = 78.207, a = 26.322 b = 24.253, c = -2.580, Joint = 0.257, Marginal = -2.444, MI = -5.911\n",
      "Pre-Train Epoch 9, Batch 54: Loss = 75.535, a = 14.908 b = 38.826, c = -0.150, Joint = -1.538, Marginal = -4.083, MI = -6.045\n",
      "Pre-Train Epoch 9, Batch 55: Loss = 5685.247, a = 39.542 b = 35.329, c = -0.150, Joint = -10.578, Marginal = -19.982, MI = -15.163\n",
      "Pre-Train Epoch 9, Batch 56: Loss = 117.878, a = 20.574 b = 35.132, c = -2.330, Joint = -1.778, Marginal = -3.967, MI = -5.796\n",
      "Pre-Train Epoch 9, Batch 57: Loss = 64.693, a = 16.687 b = 33.709, c = -5.436, Joint = -0.419, Marginal = -4.802, MI = -3.692\n",
      "Pre-Train Epoch 9, Batch 58: Loss = 96.012, a = 17.822 b = 37.563, c = -4.055, Joint = -1.268, Marginal = -3.608, MI = -5.293\n",
      "Pre-Train Epoch 9, Batch 59: Loss = 61.894, a = 13.943 b = 31.067, c = -0.318, Joint = -0.372, Marginal = -1.098, MI = -5.061\n",
      "Pre-Train Epoch 9, Batch 60: Loss = 2108.128, a = 35.045 b = 33.198, c = -0.148, Joint = -7.373, Marginal = -12.804, MI = -12.616\n",
      "Pre-Train Epoch 9, Batch 61: Loss = 76.032, a = 22.859 b = 32.469, c = -0.148, Joint = -0.344, Marginal = -2.103, MI = -5.535\n",
      "Pre-Train Epoch 9, Batch 62: Loss = 308.083, a = 32.218 b = 31.004, c = -4.795, Joint = -3.381, Marginal = -8.264, MI = -10.714\n",
      "Pre-Train Epoch 9, Batch 63: Loss = 113.094, a = 27.500 b = 29.409, c = -0.148, Joint = -1.482, Marginal = -5.440, MI = -8.767\n",
      "Pre-Train Epoch 9, Batch 64: Loss = 67.229, a = 15.334 b = 38.752, c = -0.147, Joint = -0.932, Marginal = -2.050, MI = -6.208\n",
      "Pre-Train Epoch 9, Batch 65: Loss = 969.514, a = 48.643 b = 27.621, c = -1.086, Joint = -4.927, Marginal = -14.236, MI = -12.214\n",
      "Pre-Train Epoch 9, Batch 66: Loss = 70.312, a = 15.371 b = 41.054, c = -0.147, Joint = -1.132, Marginal = -2.264, MI = -9.443\n",
      "Pre-Train Epoch 9, Batch 67: Loss = 83.303, a = 23.637 b = 43.801, c = -0.147, Joint = -1.377, Marginal = -4.789, MI = -5.583\n",
      "Pre-Train Epoch 9, Batch 68: Loss = 98.344, a = 31.478 b = 29.811, c = -0.146, Joint = -0.319, Marginal = -8.239, MI = -5.680\n",
      "Pre-Train Epoch 9, Batch 69: Loss = 132.257, a = 26.991 b = 31.722, c = -0.146, Joint = -1.710, Marginal = -5.431, MI = -6.964\n",
      "Pre-Train Epoch 9, Batch 70: Loss = 262.372, a = 19.723 b = 40.057, c = -2.497, Joint = -3.017, Marginal = -5.146, MI = -4.837\n",
      "Pre-Train Epoch 9, Batch 71: Loss = 759.473, a = 36.610 b = 45.779, c = -0.146, Joint = -4.593, Marginal = -9.284, MI = -10.104\n",
      "Pre-Train Epoch 9, Batch 72: Loss = 241.607, a = 44.626 b = 27.988, c = -0.146, Joint = -2.018, Marginal = -9.094, MI = -11.217\n",
      "Pre-Train Epoch 9, Batch 73: Loss = 157.214, a = 28.259 b = 25.799, c = -0.145, Joint = -1.609, Marginal = -4.262, MI = -7.371\n",
      "Pre-Train Epoch 9, Batch 74: Loss = 76.614, a = 26.831 b = 33.158, c = -0.523, Joint = -0.458, Marginal = -0.781, MI = -7.925\n",
      "Pre-Train Epoch 9, Batch 75: Loss = 86.063, a = 27.411 b = 28.863, c = -0.145, Joint = -0.121, Marginal = -0.854, MI = -8.635\n",
      "Pre-Train Epoch 9, Batch 76: Loss = 166.087, a = 22.534 b = 34.084, c = -0.145, Joint = -2.003, Marginal = -1.540, MI = -8.784\n",
      "Pre-Train Epoch 9, Batch 77: Loss = 435.534, a = 32.751 b = 32.837, c = -0.145, Joint = -4.000, Marginal = -12.089, MI = -7.983\n",
      "Pre-Train Epoch 9, Batch 78: Loss = 71.630, a = 11.101 b = 40.797, c = -0.144, Joint = -2.002, Marginal = -2.878, MI = -3.454\n",
      "Pre-Train Epoch 9, Batch 79: Loss = 180.629, a = 23.607 b = 32.128, c = -0.144, Joint = -2.314, Marginal = -6.426, MI = -8.801\n",
      "Pre-Train Epoch 9, Batch 80: Loss = 188.666, a = 19.504 b = 38.142, c = -0.144, Joint = -2.203, Marginal = -2.584, MI = -8.977\n",
      "Pre-Train Epoch 9, Batch 81: Loss = 90.110, a = 24.521 b = 33.916, c = -5.199, Joint = -1.428, Marginal = -4.351, MI = -8.457\n",
      "Pre-Train Epoch 9, Batch 82: Loss = 342.855, a = 33.931 b = 35.860, c = -4.422, Joint = -2.388, Marginal = -7.115, MI = -8.805\n",
      "Pre-Train Epoch 9, Batch 83: Loss = 63.970, a = 19.885 b = 32.539, c = -0.144, Joint = -0.531, Marginal = -1.699, MI = -5.673\n",
      "Pre-Train Epoch 9, Batch 84: Loss = 113.559, a = 27.894 b = 30.824, c = -0.143, Joint = -0.498, Marginal = -5.017, MI = -6.595\n",
      "Pre-Train Epoch 9, Batch 85: Loss = 638.662, a = 32.282 b = 33.544, c = -0.143, Joint = -5.039, Marginal = -7.304, MI = -7.087\n",
      "Pre-Train Epoch 9, Batch 86: Loss = 69.356, a = 19.184 b = 32.787, c = -0.143, Joint = -0.654, Marginal = -2.278, MI = -6.089\n",
      "Pre-Train Epoch 9, Batch 87: Loss = 163.960, a = 22.351 b = 30.254, c = -0.143, Joint = -1.138, Marginal = -3.499, MI = -8.528\n",
      "Pre-Train Epoch 9, Batch 88: Loss = 58.895, a = 11.683 b = 36.165, c = -0.143, Joint = -0.559, Marginal = -1.590, MI = -4.444\n",
      "Pre-Train Epoch 9, Batch 89: Loss = 70.051, a = 18.238 b = 34.317, c = -1.332, Joint = -0.127, Marginal = -1.754, MI = -5.454\n",
      "Pre-Train Epoch 9, Batch 90: Loss = 93.774, a = 34.599 b = 32.457, c = -0.142, Joint = -0.666, Marginal = -7.493, MI = -4.649\n",
      "Pre-Train Epoch 9, Batch 91: Loss = 60.985, a = 12.303 b = 34.388, c = -0.142, Joint = -0.892, Marginal = -1.260, MI = -4.273\n",
      "Pre-Train Epoch 9, Batch 92: Loss = 108.899, a = 45.483 b = 27.001, c = -0.142, Joint = -0.829, Marginal = -4.149, MI = -6.677\n",
      "Pre-Train Epoch 9, Batch 93: Loss = 382.966, a = 31.240 b = 42.532, c = -0.142, Joint = -4.530, Marginal = -5.737, MI = -11.312\n",
      "Pre-Train Epoch 9, Batch 94: Loss = 66.205, a = 19.738 b = 30.490, c = -0.142, Joint = 0.062, Marginal = -1.979, MI = -7.677\n",
      "Pre-Train Epoch 9, Batch 95: Loss = 286.501, a = 28.959 b = 38.650, c = -0.142, Joint = -3.199, Marginal = -6.548, MI = -8.932\n",
      "Pre-Train Epoch 9, Batch 96: Loss = 137.525, a = 18.266 b = 39.858, c = -0.142, Joint = -1.680, Marginal = -2.428, MI = -7.286\n",
      "Pre-Train Epoch 9, Batch 97: Loss = 67.088, a = 17.073 b = 35.568, c = -0.141, Joint = -0.572, Marginal = -1.987, MI = -5.237\n",
      "Pre-Train Epoch 9, Batch 98: Loss = 130.966, a = 32.080 b = 35.759, c = -6.217, Joint = -1.460, Marginal = -4.215, MI = -5.189\n",
      "Pre-Train Epoch 9, Batch 99: Loss = 156.619, a = 25.788 b = 36.029, c = -0.141, Joint = -1.497, Marginal = -2.692, MI = -8.201\n",
      "Pre-Train Epoch 9, Batch 100: Loss = 59.266, a = 12.063 b = 35.142, c = -0.141, Joint = -0.180, Marginal = -0.840, MI = -5.999\n",
      "Pre-Train Epoch 9, Batch 101: Loss = 437.580, a = 50.194 b = 34.403, c = -4.944, Joint = -3.106, Marginal = -10.122, MI = -11.265\n",
      "Pre-Train Epoch 9, Batch 102: Loss = 211.124, a = 23.337 b = 32.600, c = -0.141, Joint = -1.679, Marginal = -3.271, MI = -8.477\n",
      "Pre-Train Epoch 9, Batch 103: Loss = 59.898, a = 15.704 b = 34.333, c = -0.140, Joint = -0.712, Marginal = -4.387, MI = -5.917\n",
      "Pre-Train Epoch 9, Batch 104: Loss = 230.791, a = 28.004 b = 32.485, c = -0.140, Joint = -2.700, Marginal = -8.816, MI = -9.985\n",
      "Pre-Train Epoch 9, Batch 105: Loss = 173.294, a = 23.164 b = 30.901, c = -0.140, Joint = -2.068, Marginal = -4.597, MI = -6.526\n",
      "Pre-Train Epoch 9, Batch 106: Loss = 207.174, a = 35.842 b = 35.280, c = -0.140, Joint = -2.314, Marginal = -9.783, MI = -7.351\n",
      "Pre-Train Epoch 9, Batch 107: Loss = 74.130, a = 18.031 b = 33.626, c = -3.102, Joint = -0.688, Marginal = -3.140, MI = -5.237\n",
      "Pre-Train Epoch 9, Batch 108: Loss = 364.510, a = 13.530 b = 40.118, c = -0.140, Joint = -3.292, Marginal = -5.888, MI = -6.143\n",
      "Pre-Train Epoch 9, Batch 109: Loss = 913.420, a = 22.003 b = 30.982, c = -1.615, Joint = -4.980, Marginal = -6.750, MI = -10.430\n",
      "Pre-Train Epoch 9, Batch 110: Loss = 741.205, a = 29.731 b = 35.203, c = -0.139, Joint = -3.613, Marginal = -11.079, MI = -11.786\n",
      "Pre-Train Epoch 9, Batch 111: Loss = 1603.794, a = 38.297 b = 39.574, c = -0.139, Joint = -8.377, Marginal = -11.623, MI = -13.564\n",
      "Pre-Train Epoch 9, Batch 112: Loss = 109.720, a = 20.467 b = 30.044, c = -0.268, Joint = -1.156, Marginal = -6.495, MI = -6.380\n",
      "Pre-Train Epoch 9, Batch 113: Loss = 114.824, a = 27.883 b = 39.034, c = -0.139, Joint = -1.646, Marginal = -2.713, MI = -7.270\n",
      "Pre-Train Epoch 9, Batch 114: Loss = 79.539, a = 11.389 b = 33.861, c = -0.139, Joint = -1.462, Marginal = -1.036, MI = -5.570\n",
      "Pre-Train Epoch 9, Batch 115: Loss = 552.033, a = 28.898 b = 31.485, c = -0.138, Joint = -3.346, Marginal = -7.936, MI = -11.621\n",
      "Pre-Train Epoch 9, Batch 116: Loss = 66.554, a = 14.653 b = 38.928, c = -0.138, Joint = -0.469, Marginal = -1.521, MI = -6.473\n",
      "Pre-Train Epoch 9, Batch 117: Loss = 203.590, a = 20.677 b = 33.880, c = -0.138, Joint = -2.406, Marginal = -3.758, MI = -10.515\n",
      "Pre-Train Epoch 9, Batch 118: Loss = 74.534, a = 15.657 b = 40.345, c = -0.138, Joint = -0.871, Marginal = -3.335, MI = -5.065\n",
      "Pre-Train Epoch 9, Batch 119: Loss = 67.531, a = 21.549 b = 32.361, c = -0.138, Joint = -0.632, Marginal = -2.271, MI = -3.534\n",
      "Pre-Train Epoch 9, Batch 120: Loss = 53.281, a = 12.166 b = 29.892, c = -6.474, Joint = -0.383, Marginal = -1.466, MI = -5.406\n",
      "Pre-Train Epoch 9, Batch 121: Loss = 50.363, a = 11.302 b = 28.825, c = -0.137, Joint = -0.345, Marginal = -0.954, MI = -5.611\n",
      "Pre-Train Epoch 9, Batch 122: Loss = 69.493, a = 22.305 b = 32.158, c = -5.103, Joint = -0.199, Marginal = -1.834, MI = -3.784\n",
      "Pre-Train Epoch 9, Batch 123: Loss = 78.770, a = 21.885 b = 33.131, c = -1.438, Joint = 0.370, Marginal = -1.988, MI = -3.490\n",
      "Pre-Train Epoch 9, Batch 124: Loss = 63.676, a = 18.436 b = 28.141, c = -0.137, Joint = 0.404, Marginal = -0.495, MI = -4.494\n",
      "Pre-Train Epoch 9, Batch 125: Loss = 363.102, a = 30.635 b = 27.843, c = -0.657, Joint = -4.037, Marginal = -9.752, MI = -7.469\n",
      "Pre-Train Epoch 9, Batch 126: Loss = 57.395, a = 11.671 b = 33.858, c = -0.137, Joint = -0.388, Marginal = -0.855, MI = -4.094\n",
      "Pre-Train Epoch 9, Batch 127: Loss = 70.032, a = 16.792 b = 39.963, c = -0.136, Joint = -0.919, Marginal = -2.724, MI = -4.726\n",
      "Pre-Train Epoch 9, Batch 128: Loss = 131.244, a = 15.078 b = 27.626, c = -1.044, Joint = -1.517, Marginal = -2.357, MI = -5.208\n",
      "Pre-Train Epoch 9, Batch 129: Loss = 66.148, a = 20.694 b = 33.569, c = -0.136, Joint = -0.761, Marginal = -3.418, MI = -5.855\n",
      "Pre-Train Epoch 9, Batch 130: Loss = 71.052, a = 14.817 b = 39.325, c = -0.136, Joint = -0.477, Marginal = -2.080, MI = -6.533\n",
      "Pre-Train Epoch 9, Batch 131: Loss = 80.060, a = 11.640 b = 34.028, c = -3.328, Joint = -1.108, Marginal = -1.993, MI = -4.968\n",
      "Pre-Train Epoch 9, Batch 132: Loss = 154.016, a = 14.509 b = 32.773, c = -0.136, Joint = -1.634, Marginal = -2.979, MI = -4.553\n",
      "Pre-Train Epoch 9, Batch 133: Loss = 907.353, a = 28.820 b = 36.980, c = -0.136, Joint = -5.495, Marginal = -11.097, MI = -8.129\n",
      "Pre-Train Epoch 9, Batch 134: Loss = 122.241, a = 23.042 b = 25.219, c = -0.135, Joint = -0.604, Marginal = -1.692, MI = -8.797\n",
      "Pre-Train Epoch 9, Batch 135: Loss = 62.209, a = 15.822 b = 31.289, c = -0.135, Joint = -0.122, Marginal = -1.825, MI = -6.878\n",
      "Pre-Train Epoch 9, Batch 136: Loss = 69.694, a = 12.908 b = 44.149, c = -0.135, Joint = -1.471, Marginal = -1.493, MI = -6.541\n",
      "Pre-Train Epoch 9, Batch 137: Loss = 57.597, a = 14.975 b = 31.346, c = -0.135, Joint = -0.214, Marginal = -2.146, MI = -5.038\n",
      "Pre-Train Epoch 9, Batch 138: Loss = 104.532, a = 13.655 b = 32.038, c = -0.135, Joint = -1.663, Marginal = -1.180, MI = -5.641\n",
      "Pre-Train Epoch 9, Batch 139: Loss = 80.228, a = 17.439 b = 41.773, c = -2.108, Joint = -1.346, Marginal = -4.268, MI = -4.582\n",
      "Pre-Train Epoch 9, Batch 140: Loss = 1532.777, a = 22.681 b = 37.401, c = -2.896, Joint = -6.743, Marginal = -9.851, MI = -8.778\n",
      "Pre-Train Epoch 9, Batch 141: Loss = 78.130, a = 23.703 b = 34.581, c = -0.134, Joint = -0.125, Marginal = -2.447, MI = -4.143\n",
      "Pre-Train Epoch 9, Batch 142: Loss = 101.490, a = 23.923 b = 30.154, c = -0.134, Joint = -0.336, Marginal = -7.189, MI = -7.828\n",
      "Pre-Train Epoch 9, Batch 143: Loss = 570.377, a = 22.036 b = 38.777, c = -0.134, Joint = -3.057, Marginal = -4.511, MI = -7.304\n",
      "Pre-Train Epoch 9, Batch 144: Loss = 103.177, a = 16.536 b = 39.003, c = -0.134, Joint = -1.443, Marginal = -3.352, MI = -4.760\n",
      "Pre-Train Epoch 9, Batch 145: Loss = 67.228, a = 17.343 b = 32.087, c = -0.134, Joint = -1.002, Marginal = -6.428, MI = -1.142\n",
      "Pre-Train Epoch 9, Batch 146: Loss = 62.327, a = 15.599 b = 34.909, c = -0.133, Joint = -0.353, Marginal = -2.388, MI = -4.444\n",
      "Pre-Train Epoch 9, Batch 147: Loss = 85.214, a = 23.874 b = 39.048, c = -0.133, Joint = -0.503, Marginal = -3.622, MI = -5.541\n",
      "Pre-Train Epoch 9, Batch 148: Loss = 87.423, a = 17.315 b = 32.200, c = -0.133, Joint = -1.279, Marginal = -1.936, MI = -7.798\n",
      "Pre-Train Epoch 9, Batch 149: Loss = 46.554, a = 10.054 b = 25.116, c = -4.082, Joint = -0.372, Marginal = -0.701, MI = -6.431\n",
      "Pre-Train Epoch 9, Batch 150: Loss = 68.669, a = 17.745 b = 34.645, c = -5.393, Joint = -0.791, Marginal = -4.672, MI = -5.360\n",
      "Epoch 1/1000\n",
      "10/10 - 0s - loss: 5.9601 - MI: -5.9606e+00\n",
      "Epoch 2/1000\n",
      "10/10 - 0s - loss: 3.5493 - MI: -3.5218e+00\n",
      "Epoch 3/1000\n",
      "10/10 - 0s - loss: 2.3844 - MI: -2.3751e+00\n",
      "Epoch 4/1000\n",
      "10/10 - 0s - loss: 1.8422 - MI: -1.8272e+00\n",
      "Epoch 5/1000\n",
      "10/10 - 0s - loss: 1.5524 - MI: -1.5617e+00\n",
      "Epoch 6/1000\n",
      "10/10 - 0s - loss: 1.3850 - MI: -1.3910e+00\n",
      "Epoch 7/1000\n",
      "10/10 - 0s - loss: 1.2063 - MI: -1.2255e+00\n",
      "Epoch 8/1000\n",
      "10/10 - 0s - loss: 1.0727 - MI: -1.0638e+00\n",
      "Epoch 9/1000\n",
      "10/10 - 0s - loss: 0.9748 - MI: -9.7679e-01\n",
      "Epoch 10/1000\n",
      "10/10 - 0s - loss: 0.8981 - MI: -8.9536e-01\n",
      "Epoch 11/1000\n",
      "10/10 - 0s - loss: 0.7895 - MI: -7.8271e-01\n",
      "Epoch 12/1000\n",
      "10/10 - 0s - loss: 0.7423 - MI: -7.3634e-01\n",
      "Epoch 13/1000\n",
      "10/10 - 0s - loss: 0.6943 - MI: -6.8854e-01\n",
      "Epoch 14/1000\n",
      "10/10 - 0s - loss: 0.6462 - MI: -6.4532e-01\n",
      "Epoch 15/1000\n",
      "10/10 - 0s - loss: 0.5651 - MI: -5.6565e-01\n",
      "Epoch 16/1000\n",
      "10/10 - 0s - loss: 0.5344 - MI: -5.3606e-01\n",
      "Epoch 17/1000\n",
      "10/10 - 0s - loss: 0.5120 - MI: -5.1051e-01\n",
      "Epoch 18/1000\n",
      "10/10 - 0s - loss: 0.4602 - MI: -4.6117e-01\n",
      "Epoch 19/1000\n",
      "10/10 - 0s - loss: 0.4329 - MI: -4.3217e-01\n",
      "Epoch 20/1000\n",
      "10/10 - 0s - loss: 0.4225 - MI: -4.2009e-01\n",
      "Epoch 21/1000\n",
      "10/10 - 0s - loss: 0.3842 - MI: -3.8215e-01\n",
      "Epoch 22/1000\n",
      "10/10 - 0s - loss: 0.3624 - MI: -3.5730e-01\n",
      "Epoch 23/1000\n",
      "10/10 - 0s - loss: 0.3510 - MI: -3.4694e-01\n",
      "Epoch 24/1000\n",
      "10/10 - 0s - loss: 0.3308 - MI: -3.2983e-01\n",
      "Epoch 25/1000\n",
      "10/10 - 0s - loss: 0.3069 - MI: -3.0447e-01\n",
      "Epoch 26/1000\n",
      "10/10 - 0s - loss: 0.2766 - MI: -2.7456e-01\n",
      "Epoch 27/1000\n",
      "10/10 - 0s - loss: 0.2696 - MI: -2.6788e-01\n",
      "Epoch 28/1000\n",
      "10/10 - 0s - loss: 0.2476 - MI: -2.4907e-01\n",
      "Epoch 29/1000\n",
      "10/10 - 0s - loss: 0.2332 - MI: -2.3563e-01\n",
      "Epoch 30/1000\n",
      "10/10 - 0s - loss: 0.2244 - MI: -2.2688e-01\n",
      "Epoch 31/1000\n",
      "10/10 - 0s - loss: 0.2118 - MI: -2.0986e-01\n",
      "Epoch 32/1000\n",
      "10/10 - 0s - loss: 0.1922 - MI: -1.9039e-01\n",
      "Epoch 33/1000\n",
      "10/10 - 0s - loss: 0.1846 - MI: -1.8371e-01\n",
      "Epoch 34/1000\n",
      "10/10 - 0s - loss: 0.1823 - MI: -1.8140e-01\n",
      "Epoch 35/1000\n",
      "10/10 - 0s - loss: 0.1673 - MI: -1.6519e-01\n",
      "Epoch 36/1000\n",
      "10/10 - 0s - loss: 0.1589 - MI: -1.6136e-01\n",
      "Epoch 37/1000\n",
      "10/10 - 0s - loss: 0.1557 - MI: -1.5285e-01\n",
      "Epoch 38/1000\n",
      "10/10 - 0s - loss: 0.1428 - MI: -1.4329e-01\n",
      "Epoch 39/1000\n",
      "10/10 - 0s - loss: 0.1353 - MI: -1.3312e-01\n",
      "Epoch 40/1000\n",
      "10/10 - 0s - loss: 0.1291 - MI: -1.2768e-01\n",
      "Epoch 41/1000\n",
      "10/10 - 0s - loss: 0.1267 - MI: -1.2660e-01\n",
      "Epoch 42/1000\n",
      "10/10 - 0s - loss: 0.1090 - MI: -1.0863e-01\n",
      "Epoch 43/1000\n",
      "10/10 - 0s - loss: 0.1003 - MI: -1.0317e-01\n",
      "Epoch 44/1000\n",
      "10/10 - 0s - loss: 0.0958 - MI: -9.3969e-02\n",
      "Epoch 45/1000\n",
      "10/10 - 0s - loss: 0.1107 - MI: -1.0845e-01\n",
      "Epoch 46/1000\n",
      "10/10 - 0s - loss: 0.0935 - MI: -9.2470e-02\n",
      "Epoch 47/1000\n",
      "10/10 - 0s - loss: 0.0792 - MI: -7.8765e-02\n",
      "Epoch 48/1000\n",
      "10/10 - 0s - loss: 0.0800 - MI: -7.9572e-02\n",
      "Epoch 49/1000\n",
      "10/10 - 0s - loss: 0.0752 - MI: -7.3842e-02\n",
      "Epoch 50/1000\n",
      "10/10 - 0s - loss: 0.0727 - MI: -7.1894e-02\n",
      "Epoch 51/1000\n",
      "10/10 - 0s - loss: 0.0686 - MI: -6.9173e-02\n",
      "Epoch 52/1000\n",
      "10/10 - 0s - loss: 0.0572 - MI: -5.5939e-02\n",
      "Epoch 53/1000\n",
      "10/10 - 0s - loss: 0.0645 - MI: -6.5665e-02\n",
      "Epoch 54/1000\n",
      "10/10 - 0s - loss: 0.0387 - MI: -3.8948e-02\n",
      "Epoch 55/1000\n",
      "10/10 - 0s - loss: 0.0458 - MI: -4.5500e-02\n",
      "Epoch 56/1000\n",
      "10/10 - 0s - loss: 0.0444 - MI: -4.3401e-02\n",
      "Epoch 57/1000\n",
      "10/10 - 0s - loss: 0.0488 - MI: -4.7967e-02\n",
      "Epoch 58/1000\n",
      "10/10 - 0s - loss: 0.0370 - MI: -3.6361e-02\n",
      "Epoch 59/1000\n",
      "10/10 - 0s - loss: 0.0344 - MI: -3.3170e-02\n",
      "Epoch 60/1000\n",
      "10/10 - 0s - loss: 0.0286 - MI: -2.8617e-02\n",
      "Epoch 61/1000\n",
      "10/10 - 0s - loss: 0.0278 - MI: -2.7633e-02\n",
      "Epoch 62/1000\n",
      "10/10 - 0s - loss: 0.0337 - MI: -3.3198e-02\n",
      "Epoch 63/1000\n",
      "10/10 - 0s - loss: 0.0299 - MI: -2.9821e-02\n",
      "Epoch 64/1000\n",
      "10/10 - 0s - loss: 0.0177 - MI: -1.7950e-02\n",
      "Epoch 65/1000\n",
      "10/10 - 0s - loss: 0.0243 - MI: -2.5205e-02\n",
      "Epoch 66/1000\n",
      "10/10 - 0s - loss: 0.0144 - MI: -1.3961e-02\n",
      "Epoch 67/1000\n",
      "10/10 - 0s - loss: 0.0104 - MI: -1.0946e-02\n",
      "Epoch 68/1000\n",
      "10/10 - 0s - loss: 0.0064 - MI: -6.4044e-03\n",
      "Epoch 69/1000\n",
      "10/10 - 0s - loss: 0.0199 - MI: -1.8434e-02\n",
      "Epoch 70/1000\n",
      "10/10 - 0s - loss: 0.0142 - MI: -1.4499e-02\n",
      "Epoch 71/1000\n",
      "10/10 - 0s - loss: 0.0021 - MI: -2.9718e-03\n",
      "Epoch 72/1000\n",
      "10/10 - 0s - loss: -7.4791e-04 - MI: 0.0019\n",
      "Epoch 73/1000\n",
      "10/10 - 0s - loss: 0.0109 - MI: -1.1237e-02\n",
      "Epoch 74/1000\n",
      "10/10 - 0s - loss: 6.9425e-04 - MI: 2.3026e-04\n",
      "Epoch 75/1000\n",
      "10/10 - 0s - loss: -4.3939e-03 - MI: 0.0044\n",
      "Epoch 76/1000\n",
      "10/10 - 0s - loss: 0.0018 - MI: -2.6383e-03\n",
      "Epoch 77/1000\n",
      "10/10 - 0s - loss: 4.3455e-04 - MI: -5.6863e-04\n",
      "Epoch 78/1000\n",
      "10/10 - 0s - loss: -6.4743e-03 - MI: 0.0062\n",
      "Epoch 79/1000\n",
      "10/10 - 0s - loss: -3.9761e-03 - MI: 0.0030\n",
      "Epoch 80/1000\n",
      "10/10 - 0s - loss: -7.8815e-03 - MI: 0.0086\n",
      "Epoch 81/1000\n",
      "10/10 - 0s - loss: -3.7572e-03 - MI: 0.0040\n",
      "Epoch 82/1000\n",
      "10/10 - 0s - loss: -2.0188e-02 - MI: 0.0208\n",
      "Epoch 83/1000\n",
      "10/10 - 0s - loss: -1.2668e-02 - MI: 0.0130\n",
      "Epoch 84/1000\n",
      "10/10 - 0s - loss: -1.9033e-02 - MI: 0.0183\n",
      "Epoch 85/1000\n",
      "10/10 - 0s - loss: -1.0596e-02 - MI: 0.0117\n",
      "Epoch 86/1000\n",
      "10/10 - 0s - loss: -1.2379e-02 - MI: 0.0125\n",
      "Epoch 87/1000\n",
      "10/10 - 0s - loss: -1.7416e-02 - MI: 0.0170\n",
      "Epoch 88/1000\n",
      "10/10 - 0s - loss: -2.4678e-02 - MI: 0.0243\n",
      "Epoch 89/1000\n",
      "10/10 - 0s - loss: -2.1504e-02 - MI: 0.0219\n",
      "Epoch 90/1000\n",
      "10/10 - 0s - loss: -1.3149e-02 - MI: 0.0124\n",
      "Epoch 91/1000\n",
      "10/10 - 0s - loss: -1.2790e-02 - MI: 0.0127\n",
      "Epoch 92/1000\n",
      "10/10 - 0s - loss: -2.8448e-02 - MI: 0.0265\n",
      "Epoch 93/1000\n",
      "10/10 - 0s - loss: -2.2467e-02 - MI: 0.0235\n",
      "Epoch 94/1000\n",
      "10/10 - 0s - loss: -2.0529e-02 - MI: 0.0216\n",
      "Epoch 95/1000\n",
      "10/10 - 0s - loss: -2.8821e-02 - MI: 0.0272\n",
      "Epoch 96/1000\n",
      "10/10 - 0s - loss: -1.9966e-02 - MI: 0.0193\n",
      "Epoch 97/1000\n",
      "10/10 - 0s - loss: -2.2192e-02 - MI: 0.0226\n",
      "Epoch 98/1000\n",
      "10/10 - 0s - loss: -2.1960e-02 - MI: 0.0219\n",
      "Epoch 99/1000\n",
      "10/10 - 0s - loss: -2.4309e-02 - MI: 0.0235\n",
      "Epoch 100/1000\n",
      "10/10 - 0s - loss: -2.9187e-02 - MI: 0.0296\n",
      "Epoch 101/1000\n",
      "10/10 - 0s - loss: -2.1890e-02 - MI: 0.0208\n",
      "Epoch 102/1000\n",
      "10/10 - 0s - loss: -2.4288e-02 - MI: 0.0240\n",
      "Epoch 103/1000\n",
      "10/10 - 0s - loss: -3.2457e-02 - MI: 0.0322\n",
      "Epoch 104/1000\n",
      "10/10 - 0s - loss: -2.1328e-02 - MI: 0.0215\n",
      "Epoch 105/1000\n",
      "10/10 - 0s - loss: -3.1023e-02 - MI: 0.0307\n",
      "Epoch 106/1000\n",
      "10/10 - 0s - loss: -3.0196e-02 - MI: 0.0310\n",
      "Epoch 107/1000\n",
      "10/10 - 0s - loss: -3.2438e-02 - MI: 0.0326\n",
      "Epoch 108/1000\n",
      "10/10 - 0s - loss: -3.2412e-02 - MI: 0.0320\n",
      "Epoch 109/1000\n",
      "10/10 - 0s - loss: -2.2241e-02 - MI: 0.0218\n",
      "Epoch 110/1000\n",
      "10/10 - 0s - loss: -3.6439e-02 - MI: 0.0368\n",
      "Epoch 111/1000\n",
      "10/10 - 0s - loss: -2.4248e-02 - MI: 0.0241\n",
      "Epoch 112/1000\n",
      "10/10 - 0s - loss: -3.2371e-02 - MI: 0.0327\n",
      "Epoch 113/1000\n",
      "10/10 - 0s - loss: -3.0464e-02 - MI: 0.0308\n",
      "Epoch 114/1000\n",
      "10/10 - 0s - loss: -3.4054e-02 - MI: 0.0340\n",
      "Epoch 115/1000\n",
      "10/10 - 0s - loss: -3.2436e-02 - MI: 0.0326\n",
      "Epoch 116/1000\n",
      "10/10 - 0s - loss: -3.2740e-02 - MI: 0.0324\n",
      "Epoch 117/1000\n",
      "10/10 - 0s - loss: -3.4139e-02 - MI: 0.0337\n",
      "Epoch 118/1000\n",
      "10/10 - 0s - loss: -3.3032e-02 - MI: 0.0334\n",
      "Epoch 119/1000\n",
      "10/10 - 0s - loss: -3.3761e-02 - MI: 0.0341\n",
      "Epoch 120/1000\n",
      "10/10 - 0s - loss: -3.1273e-02 - MI: 0.0311\n",
      "Epoch 121/1000\n",
      "10/10 - 0s - loss: -3.7806e-02 - MI: 0.0384\n",
      "Epoch 122/1000\n",
      "10/10 - 0s - loss: -3.4253e-02 - MI: 0.0337\n",
      "Epoch 123/1000\n",
      "10/10 - 0s - loss: -3.9510e-02 - MI: 0.0395\n",
      "Epoch 124/1000\n",
      "10/10 - 0s - loss: -3.5868e-02 - MI: 0.0352\n",
      "Epoch 125/1000\n",
      "10/10 - 0s - loss: -3.9265e-02 - MI: 0.0393\n",
      "Epoch 126/1000\n",
      "10/10 - 0s - loss: -4.1452e-02 - MI: 0.0414\n",
      "Epoch 127/1000\n",
      "10/10 - 0s - loss: -3.8619e-02 - MI: 0.0384\n",
      "Epoch 128/1000\n",
      "10/10 - 0s - loss: -3.3993e-02 - MI: 0.0338\n",
      "Epoch 129/1000\n",
      "10/10 - 0s - loss: -3.9624e-02 - MI: 0.0394\n",
      "Epoch 130/1000\n",
      "10/10 - 0s - loss: -4.0392e-02 - MI: 0.0399\n",
      "Epoch 131/1000\n",
      "10/10 - 0s - loss: -4.1933e-02 - MI: 0.0425\n",
      "Epoch 132/1000\n",
      "10/10 - 0s - loss: -4.0959e-02 - MI: 0.0410\n",
      "Epoch 133/1000\n",
      "10/10 - 0s - loss: -3.8060e-02 - MI: 0.0384\n",
      "Epoch 134/1000\n",
      "10/10 - 0s - loss: -3.7745e-02 - MI: 0.0384\n",
      "Epoch 135/1000\n",
      "10/10 - 0s - loss: -4.7414e-02 - MI: 0.0479\n",
      "Epoch 136/1000\n",
      "10/10 - 0s - loss: -3.9206e-02 - MI: 0.0387\n",
      "Epoch 137/1000\n",
      "10/10 - 0s - loss: -4.4322e-02 - MI: 0.0447\n",
      "Epoch 138/1000\n",
      "10/10 - 0s - loss: -3.4970e-02 - MI: 0.0358\n",
      "Epoch 139/1000\n",
      "10/10 - 0s - loss: -3.8018e-02 - MI: 0.0378\n",
      "Epoch 140/1000\n",
      "10/10 - 0s - loss: -4.3367e-02 - MI: 0.0429\n",
      "Epoch 141/1000\n",
      "10/10 - 0s - loss: -4.1785e-02 - MI: 0.0423\n",
      "Epoch 142/1000\n",
      "10/10 - 0s - loss: -4.4962e-02 - MI: 0.0453\n",
      "Epoch 143/1000\n",
      "10/10 - 0s - loss: -4.2362e-02 - MI: 0.0423\n",
      "Epoch 144/1000\n",
      "10/10 - 0s - loss: -4.0897e-02 - MI: 0.0407\n",
      "Epoch 145/1000\n",
      "10/10 - 0s - loss: -4.0333e-02 - MI: 0.0407\n",
      "Epoch 146/1000\n",
      "10/10 - 0s - loss: -4.2215e-02 - MI: 0.0420\n",
      "Epoch 147/1000\n",
      "10/10 - 0s - loss: -4.5857e-02 - MI: 0.0461\n",
      "Epoch 148/1000\n",
      "10/10 - 0s - loss: -4.4015e-02 - MI: 0.0435\n",
      "Epoch 149/1000\n",
      "10/10 - 0s - loss: -4.4696e-02 - MI: 0.0453\n",
      "Epoch 150/1000\n",
      "10/10 - 0s - loss: -5.0640e-02 - MI: 0.0503\n",
      "Epoch 151/1000\n",
      "10/10 - 0s - loss: -3.8439e-02 - MI: 0.0384\n",
      "Epoch 152/1000\n",
      "10/10 - 0s - loss: -3.6929e-02 - MI: 0.0366\n",
      "Epoch 153/1000\n",
      "10/10 - 0s - loss: -4.2741e-02 - MI: 0.0431\n",
      "Epoch 154/1000\n",
      "10/10 - 0s - loss: -3.9667e-02 - MI: 0.0403\n",
      "Epoch 155/1000\n",
      "10/10 - 0s - loss: -4.0951e-02 - MI: 0.0406\n",
      "Epoch 156/1000\n",
      "10/10 - 0s - loss: -3.9846e-02 - MI: 0.0407\n",
      "Epoch 157/1000\n",
      "10/10 - 0s - loss: -4.8449e-02 - MI: 0.0482\n",
      "Epoch 158/1000\n",
      "10/10 - 0s - loss: -4.1598e-02 - MI: 0.0417\n",
      "Epoch 159/1000\n",
      "10/10 - 0s - loss: -4.8319e-02 - MI: 0.0479\n",
      "Epoch 160/1000\n",
      "10/10 - 0s - loss: -4.4408e-02 - MI: 0.0453\n",
      "Epoch 161/1000\n",
      "10/10 - 0s - loss: -5.1098e-02 - MI: 0.0514\n",
      "Epoch 162/1000\n",
      "10/10 - 0s - loss: -4.4428e-02 - MI: 0.0442\n",
      "Epoch 163/1000\n",
      "10/10 - 0s - loss: -4.0085e-02 - MI: 0.0400\n",
      "Epoch 164/1000\n",
      "10/10 - 0s - loss: -5.1658e-02 - MI: 0.0514\n",
      "Epoch 165/1000\n",
      "10/10 - 0s - loss: -4.9997e-02 - MI: 0.0501\n",
      "Epoch 166/1000\n",
      "10/10 - 0s - loss: -4.2195e-02 - MI: 0.0420\n",
      "Epoch 167/1000\n",
      "10/10 - 0s - loss: -4.8646e-02 - MI: 0.0483\n",
      "Epoch 168/1000\n",
      "10/10 - 0s - loss: -4.7250e-02 - MI: 0.0473\n",
      "Epoch 169/1000\n",
      "10/10 - 0s - loss: -4.8599e-02 - MI: 0.0491\n",
      "Epoch 170/1000\n",
      "10/10 - 0s - loss: -4.7090e-02 - MI: 0.0470\n",
      "Epoch 171/1000\n",
      "10/10 - 0s - loss: -5.0623e-02 - MI: 0.0507\n",
      "Epoch 172/1000\n",
      "10/10 - 0s - loss: -4.7035e-02 - MI: 0.0477\n",
      "Epoch 173/1000\n",
      "10/10 - 0s - loss: -5.4935e-02 - MI: 0.0554\n",
      "Epoch 174/1000\n",
      "10/10 - 0s - loss: -4.9595e-02 - MI: 0.0496\n",
      "Epoch 175/1000\n",
      "10/10 - 0s - loss: -5.0848e-02 - MI: 0.0509\n",
      "Epoch 176/1000\n",
      "10/10 - 0s - loss: -4.7918e-02 - MI: 0.0476\n",
      "Epoch 177/1000\n",
      "10/10 - 0s - loss: -5.0207e-02 - MI: 0.0492\n",
      "Epoch 178/1000\n",
      "10/10 - 0s - loss: -4.4502e-02 - MI: 0.0441\n",
      "Epoch 179/1000\n",
      "10/10 - 0s - loss: -4.6758e-02 - MI: 0.0469\n",
      "Epoch 180/1000\n",
      "10/10 - 0s - loss: -4.3398e-02 - MI: 0.0433\n",
      "Epoch 181/1000\n",
      "10/10 - 0s - loss: -4.6718e-02 - MI: 0.0462\n",
      "Epoch 182/1000\n",
      "10/10 - 0s - loss: -4.3250e-02 - MI: 0.0439\n",
      "Epoch 183/1000\n",
      "10/10 - 0s - loss: -4.8801e-02 - MI: 0.0479\n",
      "Epoch 184/1000\n",
      "10/10 - 0s - loss: -4.9648e-02 - MI: 0.0490\n",
      "Epoch 185/1000\n",
      "10/10 - 0s - loss: -4.8860e-02 - MI: 0.0485\n",
      "Epoch 186/1000\n",
      "10/10 - 0s - loss: -4.8264e-02 - MI: 0.0480\n",
      "Epoch 187/1000\n",
      "10/10 - 0s - loss: -4.2755e-02 - MI: 0.0427\n",
      "Epoch 188/1000\n",
      "10/10 - 0s - loss: -4.9067e-02 - MI: 0.0485\n",
      "Epoch 189/1000\n",
      "10/10 - 0s - loss: -4.9138e-02 - MI: 0.0492\n",
      "Epoch 190/1000\n",
      "10/10 - 0s - loss: -4.5934e-02 - MI: 0.0460\n",
      "Epoch 191/1000\n",
      "10/10 - 0s - loss: -4.8197e-02 - MI: 0.0479\n",
      "Epoch 192/1000\n",
      "10/10 - 0s - loss: -4.6372e-02 - MI: 0.0472\n",
      "Epoch 193/1000\n",
      "10/10 - 0s - loss: -4.4253e-02 - MI: 0.0449\n",
      "Epoch 194/1000\n",
      "10/10 - 0s - loss: -5.0367e-02 - MI: 0.0505\n",
      "Epoch 195/1000\n",
      "10/10 - 0s - loss: -5.0786e-02 - MI: 0.0502\n",
      "Epoch 196/1000\n",
      "10/10 - 0s - loss: -4.6974e-02 - MI: 0.0478\n",
      "Epoch 197/1000\n",
      "10/10 - 0s - loss: -4.4492e-02 - MI: 0.0441\n",
      "Epoch 198/1000\n",
      "10/10 - 0s - loss: -4.9981e-02 - MI: 0.0502\n",
      "Epoch 199/1000\n",
      "10/10 - 0s - loss: -5.2351e-02 - MI: 0.0524\n",
      "Epoch 200/1000\n",
      "10/10 - 0s - loss: -4.4765e-02 - MI: 0.0442\n",
      "Epoch 201/1000\n",
      "10/10 - 0s - loss: -4.8171e-02 - MI: 0.0485\n",
      "Epoch 202/1000\n",
      "10/10 - 0s - loss: -4.8794e-02 - MI: 0.0486\n",
      "Epoch 203/1000\n",
      "10/10 - 0s - loss: -4.6532e-02 - MI: 0.0468\n",
      "Epoch 204/1000\n",
      "10/10 - 0s - loss: -5.0631e-02 - MI: 0.0512\n",
      "Epoch 205/1000\n",
      "10/10 - 0s - loss: -4.3915e-02 - MI: 0.0433\n",
      "Epoch 206/1000\n",
      "10/10 - 0s - loss: -4.4500e-02 - MI: 0.0443\n",
      "Epoch 207/1000\n",
      "10/10 - 0s - loss: -5.1019e-02 - MI: 0.0511\n",
      "Epoch 208/1000\n",
      "10/10 - 0s - loss: -4.6612e-02 - MI: 0.0475\n",
      "Epoch 209/1000\n",
      "10/10 - 0s - loss: -4.9814e-02 - MI: 0.0502\n",
      "Epoch 210/1000\n",
      "10/10 - 0s - loss: -4.3978e-02 - MI: 0.0445\n",
      "Epoch 211/1000\n",
      "10/10 - 0s - loss: -4.8784e-02 - MI: 0.0489\n",
      "Epoch 212/1000\n",
      "10/10 - 0s - loss: -4.9058e-02 - MI: 0.0492\n",
      "Epoch 213/1000\n",
      "10/10 - 0s - loss: -4.7889e-02 - MI: 0.0480\n",
      "Epoch 214/1000\n",
      "10/10 - 0s - loss: -4.9113e-02 - MI: 0.0490\n",
      "Epoch 215/1000\n",
      "10/10 - 0s - loss: -4.9289e-02 - MI: 0.0488\n",
      "Epoch 216/1000\n",
      "10/10 - 0s - loss: -4.3154e-02 - MI: 0.0432\n",
      "Epoch 217/1000\n",
      "10/10 - 0s - loss: -4.8873e-02 - MI: 0.0489\n",
      "Epoch 218/1000\n",
      "10/10 - 0s - loss: -4.6884e-02 - MI: 0.0475\n",
      "Epoch 219/1000\n",
      "10/10 - 0s - loss: -4.7405e-02 - MI: 0.0478\n",
      "Epoch 220/1000\n",
      "10/10 - 0s - loss: -4.8746e-02 - MI: 0.0490\n",
      "Epoch 221/1000\n",
      "10/10 - 0s - loss: -4.6167e-02 - MI: 0.0464\n",
      "Epoch 222/1000\n",
      "10/10 - 0s - loss: -5.4207e-02 - MI: 0.0544\n",
      "Epoch 223/1000\n",
      "10/10 - 0s - loss: -4.4911e-02 - MI: 0.0452\n",
      "Epoch 224/1000\n",
      "10/10 - 0s - loss: -4.6459e-02 - MI: 0.0468\n",
      "Epoch 225/1000\n",
      "10/10 - 0s - loss: -5.2335e-02 - MI: 0.0530\n",
      "Epoch 226/1000\n",
      "10/10 - 0s - loss: -4.4421e-02 - MI: 0.0442\n",
      "Epoch 227/1000\n",
      "10/10 - 0s - loss: -4.9170e-02 - MI: 0.0493\n",
      "Epoch 228/1000\n",
      "10/10 - 0s - loss: -4.6667e-02 - MI: 0.0469\n",
      "Epoch 229/1000\n",
      "10/10 - 0s - loss: -5.3644e-02 - MI: 0.0532\n",
      "Epoch 230/1000\n",
      "10/10 - 0s - loss: -4.6790e-02 - MI: 0.0472\n",
      "Epoch 231/1000\n",
      "10/10 - 0s - loss: -5.4037e-02 - MI: 0.0541\n",
      "Epoch 232/1000\n",
      "10/10 - 0s - loss: -4.5968e-02 - MI: 0.0456\n",
      "Epoch 233/1000\n",
      "10/10 - 0s - loss: -4.9931e-02 - MI: 0.0497\n",
      "Epoch 234/1000\n",
      "10/10 - 0s - loss: -4.6827e-02 - MI: 0.0465\n",
      "Epoch 235/1000\n",
      "10/10 - 0s - loss: -4.9774e-02 - MI: 0.0493\n",
      "Epoch 236/1000\n",
      "10/10 - 0s - loss: -4.7185e-02 - MI: 0.0475\n",
      "Epoch 237/1000\n",
      "10/10 - 0s - loss: -4.6119e-02 - MI: 0.0462\n",
      "Epoch 238/1000\n",
      "10/10 - 0s - loss: -5.2211e-02 - MI: 0.0521\n",
      "Epoch 239/1000\n",
      "10/10 - 0s - loss: -4.8725e-02 - MI: 0.0487\n",
      "Epoch 240/1000\n",
      "10/10 - 0s - loss: -4.9075e-02 - MI: 0.0488\n",
      "Epoch 241/1000\n",
      "10/10 - 0s - loss: -5.2445e-02 - MI: 0.0522\n",
      "Epoch 242/1000\n",
      "10/10 - 0s - loss: -4.6003e-02 - MI: 0.0463\n",
      "Epoch 243/1000\n",
      "10/10 - 0s - loss: -4.7141e-02 - MI: 0.0473\n",
      "Epoch 244/1000\n",
      "10/10 - 0s - loss: -4.3520e-02 - MI: 0.0431\n",
      "Epoch 245/1000\n",
      "10/10 - 0s - loss: -4.4202e-02 - MI: 0.0445\n",
      "Epoch 246/1000\n",
      "10/10 - 0s - loss: -4.3816e-02 - MI: 0.0441\n",
      "Epoch 247/1000\n",
      "10/10 - 0s - loss: -4.7365e-02 - MI: 0.0472\n",
      "Epoch 248/1000\n",
      "10/10 - 0s - loss: -4.6190e-02 - MI: 0.0464\n",
      "Epoch 249/1000\n",
      "10/10 - 0s - loss: -5.0467e-02 - MI: 0.0499\n",
      "Epoch 250/1000\n",
      "10/10 - 0s - loss: -4.6304e-02 - MI: 0.0464\n",
      "Epoch 251/1000\n",
      "10/10 - 0s - loss: -4.5521e-02 - MI: 0.0450\n",
      "Epoch 252/1000\n",
      "10/10 - 0s - loss: -4.4567e-02 - MI: 0.0448\n",
      "Epoch 253/1000\n",
      "10/10 - 0s - loss: -5.0836e-02 - MI: 0.0514\n",
      "Epoch 254/1000\n",
      "10/10 - 0s - loss: -4.5047e-02 - MI: 0.0447\n",
      "Epoch 255/1000\n",
      "10/10 - 0s - loss: -4.8276e-02 - MI: 0.0482\n",
      "Epoch 256/1000\n",
      "10/10 - 0s - loss: -4.7903e-02 - MI: 0.0479\n",
      "Epoch 257/1000\n",
      "10/10 - 0s - loss: -5.0446e-02 - MI: 0.0507\n",
      "Epoch 258/1000\n",
      "10/10 - 0s - loss: -4.6380e-02 - MI: 0.0461\n",
      "Epoch 259/1000\n",
      "10/10 - 0s - loss: -5.0980e-02 - MI: 0.0508\n",
      "Epoch 260/1000\n",
      "10/10 - 0s - loss: -4.7896e-02 - MI: 0.0474\n",
      "Epoch 261/1000\n",
      "10/10 - 0s - loss: -5.0915e-02 - MI: 0.0498\n",
      "Epoch 262/1000\n",
      "10/10 - 0s - loss: -4.9326e-02 - MI: 0.0493\n",
      "Epoch 263/1000\n",
      "10/10 - 0s - loss: -5.2231e-02 - MI: 0.0517\n",
      "Epoch 264/1000\n",
      "10/10 - 0s - loss: -4.7312e-02 - MI: 0.0476\n",
      "Epoch 265/1000\n",
      "10/10 - 0s - loss: -4.9878e-02 - MI: 0.0499\n",
      "Epoch 266/1000\n",
      "10/10 - 0s - loss: -4.9042e-02 - MI: 0.0489\n",
      "Epoch 267/1000\n",
      "10/10 - 0s - loss: -4.7497e-02 - MI: 0.0477\n",
      "Epoch 268/1000\n",
      "10/10 - 0s - loss: -5.0626e-02 - MI: 0.0507\n",
      "Epoch 269/1000\n",
      "10/10 - 0s - loss: -5.2287e-02 - MI: 0.0515\n",
      "Epoch 270/1000\n",
      "10/10 - 0s - loss: -4.4706e-02 - MI: 0.0448\n",
      "Epoch 271/1000\n",
      "10/10 - 0s - loss: -4.3546e-02 - MI: 0.0429\n",
      "Epoch 272/1000\n",
      "10/10 - 0s - loss: -5.1509e-02 - MI: 0.0520\n",
      "Epoch 273/1000\n",
      "10/10 - 0s - loss: -4.6034e-02 - MI: 0.0461\n",
      "Epoch 274/1000\n",
      "10/10 - 0s - loss: -4.5831e-02 - MI: 0.0458\n",
      "Epoch 275/1000\n",
      "10/10 - 0s - loss: -5.1951e-02 - MI: 0.0519\n",
      "Epoch 276/1000\n",
      "10/10 - 0s - loss: -4.4218e-02 - MI: 0.0443\n",
      "Epoch 277/1000\n",
      "10/10 - 0s - loss: -5.0749e-02 - MI: 0.0510\n",
      "Epoch 278/1000\n",
      "10/10 - 0s - loss: -5.1807e-02 - MI: 0.0524\n",
      "Epoch 279/1000\n",
      "10/10 - 0s - loss: -4.7877e-02 - MI: 0.0479\n",
      "Epoch 280/1000\n",
      "10/10 - 0s - loss: -5.3304e-02 - MI: 0.0532\n",
      "Epoch 281/1000\n",
      "10/10 - 0s - loss: -4.9923e-02 - MI: 0.0503\n",
      "Epoch 282/1000\n",
      "10/10 - 0s - loss: -4.5045e-02 - MI: 0.0454\n",
      "Epoch 283/1000\n",
      "10/10 - 0s - loss: -4.6549e-02 - MI: 0.0464\n",
      "Epoch 284/1000\n",
      "10/10 - 0s - loss: -4.9068e-02 - MI: 0.0490\n",
      "Epoch 285/1000\n",
      "10/10 - 0s - loss: -4.2222e-02 - MI: 0.0424\n",
      "Epoch 286/1000\n",
      "10/10 - 0s - loss: -4.5241e-02 - MI: 0.0457\n",
      "Epoch 287/1000\n",
      "10/10 - 0s - loss: -5.3221e-02 - MI: 0.0533\n",
      "Epoch 288/1000\n",
      "10/10 - 0s - loss: -4.7119e-02 - MI: 0.0475\n",
      "Epoch 289/1000\n",
      "10/10 - 0s - loss: -4.7766e-02 - MI: 0.0478\n",
      "Epoch 290/1000\n",
      "10/10 - 0s - loss: -4.6795e-02 - MI: 0.0467\n",
      "Epoch 291/1000\n",
      "10/10 - 0s - loss: -4.4243e-02 - MI: 0.0444\n",
      "Epoch 292/1000\n",
      "10/10 - 0s - loss: -4.4520e-02 - MI: 0.0445\n",
      "Epoch 293/1000\n",
      "10/10 - 0s - loss: -4.6340e-02 - MI: 0.0462\n",
      "Epoch 294/1000\n",
      "10/10 - 0s - loss: -4.5479e-02 - MI: 0.0446\n",
      "Epoch 295/1000\n",
      "10/10 - 0s - loss: -4.8859e-02 - MI: 0.0491\n",
      "Epoch 296/1000\n",
      "10/10 - 0s - loss: -4.8860e-02 - MI: 0.0490\n",
      "Epoch 297/1000\n",
      "10/10 - 0s - loss: -4.7844e-02 - MI: 0.0488\n",
      "Epoch 298/1000\n",
      "10/10 - 0s - loss: -5.3568e-02 - MI: 0.0531\n",
      "Epoch 299/1000\n",
      "10/10 - 0s - loss: -4.6319e-02 - MI: 0.0463\n",
      "Epoch 300/1000\n",
      "10/10 - 0s - loss: -4.7523e-02 - MI: 0.0474\n",
      "Epoch 301/1000\n",
      "10/10 - 0s - loss: -4.5264e-02 - MI: 0.0454\n",
      "Epoch 302/1000\n",
      "10/10 - 0s - loss: -4.9794e-02 - MI: 0.0500\n",
      "Epoch 303/1000\n",
      "10/10 - 0s - loss: -5.2379e-02 - MI: 0.0526\n",
      "Epoch 304/1000\n",
      "10/10 - 0s - loss: -5.1535e-02 - MI: 0.0516\n",
      "Epoch 305/1000\n",
      "10/10 - 0s - loss: -4.8280e-02 - MI: 0.0484\n",
      "Epoch 306/1000\n",
      "10/10 - 0s - loss: -4.8821e-02 - MI: 0.0490\n",
      "Epoch 307/1000\n",
      "10/10 - 0s - loss: -4.3112e-02 - MI: 0.0423\n",
      "Epoch 308/1000\n",
      "10/10 - 0s - loss: -4.4778e-02 - MI: 0.0448\n",
      "Epoch 309/1000\n",
      "10/10 - 0s - loss: -4.8660e-02 - MI: 0.0491\n",
      "Epoch 310/1000\n",
      "10/10 - 0s - loss: -5.2627e-02 - MI: 0.0520\n",
      "Epoch 311/1000\n",
      "10/10 - 0s - loss: -5.0028e-02 - MI: 0.0498\n",
      "Epoch 312/1000\n",
      "10/10 - 0s - loss: -4.0812e-02 - MI: 0.0416\n",
      "Epoch 313/1000\n",
      "10/10 - 0s - loss: -4.6372e-02 - MI: 0.0467\n",
      "Epoch 314/1000\n",
      "10/10 - 0s - loss: -5.1073e-02 - MI: 0.0509\n",
      "Epoch 315/1000\n",
      "10/10 - 0s - loss: -4.6668e-02 - MI: 0.0461\n",
      "Epoch 316/1000\n",
      "10/10 - 0s - loss: -4.5889e-02 - MI: 0.0458\n",
      "Epoch 317/1000\n",
      "10/10 - 0s - loss: -5.0080e-02 - MI: 0.0504\n",
      "Epoch 318/1000\n",
      "10/10 - 0s - loss: -5.3901e-02 - MI: 0.0540\n",
      "Epoch 319/1000\n",
      "10/10 - 0s - loss: -4.8596e-02 - MI: 0.0494\n",
      "Epoch 320/1000\n",
      "10/10 - 0s - loss: -4.9759e-02 - MI: 0.0492\n",
      "Epoch 321/1000\n",
      "10/10 - 0s - loss: -5.0473e-02 - MI: 0.0506\n",
      "Epoch 322/1000\n",
      "10/10 - 0s - loss: -4.6152e-02 - MI: 0.0464\n",
      "Epoch 323/1000\n",
      "10/10 - 0s - loss: -4.8023e-02 - MI: 0.0485\n",
      "Epoch 324/1000\n",
      "10/10 - 0s - loss: -4.8471e-02 - MI: 0.0478\n",
      "Epoch 325/1000\n",
      "10/10 - 0s - loss: -5.2587e-02 - MI: 0.0525\n",
      "Epoch 326/1000\n",
      "10/10 - 0s - loss: -4.6336e-02 - MI: 0.0463\n",
      "Epoch 327/1000\n",
      "10/10 - 0s - loss: -4.2275e-02 - MI: 0.0423\n",
      "Epoch 328/1000\n",
      "10/10 - 0s - loss: -4.9429e-02 - MI: 0.0498\n",
      "Epoch 329/1000\n",
      "10/10 - 0s - loss: -4.3772e-02 - MI: 0.0434\n",
      "Epoch 330/1000\n",
      "10/10 - 0s - loss: -4.7609e-02 - MI: 0.0481\n",
      "Epoch 331/1000\n",
      "10/10 - 0s - loss: -5.2083e-02 - MI: 0.0522\n",
      "Epoch 332/1000\n",
      "10/10 - 0s - loss: -4.7624e-02 - MI: 0.0473\n",
      "Epoch 333/1000\n",
      "10/10 - 0s - loss: -4.8905e-02 - MI: 0.0490\n",
      "Epoch 334/1000\n",
      "10/10 - 0s - loss: -4.8279e-02 - MI: 0.0482\n",
      "Epoch 335/1000\n",
      "10/10 - 0s - loss: -5.0939e-02 - MI: 0.0506\n",
      "Epoch 336/1000\n",
      "10/10 - 0s - loss: -4.7284e-02 - MI: 0.0468\n",
      "Epoch 337/1000\n",
      "10/10 - 0s - loss: -4.5464e-02 - MI: 0.0451\n",
      "Epoch 338/1000\n",
      "10/10 - 0s - loss: -4.9035e-02 - MI: 0.0495\n",
      "Epoch 339/1000\n",
      "10/10 - 0s - loss: -4.5861e-02 - MI: 0.0464\n",
      "Epoch 340/1000\n",
      "10/10 - 0s - loss: -4.9846e-02 - MI: 0.0500\n",
      "Epoch 341/1000\n",
      "10/10 - 0s - loss: -5.2721e-02 - MI: 0.0519\n",
      "Epoch 342/1000\n",
      "10/10 - 0s - loss: -4.9371e-02 - MI: 0.0489\n",
      "Epoch 343/1000\n",
      "10/10 - 0s - loss: -4.9437e-02 - MI: 0.0489\n",
      "Epoch 344/1000\n",
      "10/10 - 0s - loss: -5.3088e-02 - MI: 0.0534\n",
      "Epoch 345/1000\n",
      "10/10 - 0s - loss: -4.7552e-02 - MI: 0.0477\n",
      "Epoch 346/1000\n",
      "10/10 - 0s - loss: -5.4388e-02 - MI: 0.0549\n",
      "Epoch 347/1000\n",
      "10/10 - 0s - loss: -5.2603e-02 - MI: 0.0531\n",
      "Epoch 348/1000\n",
      "10/10 - 0s - loss: -5.1425e-02 - MI: 0.0511\n",
      "Epoch 349/1000\n",
      "10/10 - 0s - loss: -4.8121e-02 - MI: 0.0490\n",
      "Epoch 350/1000\n",
      "10/10 - 0s - loss: -4.4181e-02 - MI: 0.0440\n",
      "Epoch 351/1000\n",
      "10/10 - 0s - loss: -4.7595e-02 - MI: 0.0472\n",
      "Epoch 352/1000\n",
      "10/10 - 0s - loss: -4.7576e-02 - MI: 0.0479\n",
      "Epoch 353/1000\n",
      "10/10 - 0s - loss: -4.5170e-02 - MI: 0.0446\n",
      "Epoch 354/1000\n",
      "10/10 - 0s - loss: -4.7717e-02 - MI: 0.0477\n",
      "Epoch 355/1000\n",
      "10/10 - 0s - loss: -4.7301e-02 - MI: 0.0477\n",
      "Epoch 356/1000\n",
      "10/10 - 0s - loss: -4.8991e-02 - MI: 0.0494\n",
      "Epoch 357/1000\n",
      "10/10 - 0s - loss: -5.3399e-02 - MI: 0.0536\n",
      "Epoch 358/1000\n",
      "10/10 - 0s - loss: -4.4736e-02 - MI: 0.0450\n",
      "Epoch 359/1000\n",
      "10/10 - 0s - loss: -5.0422e-02 - MI: 0.0501\n",
      "Epoch 360/1000\n",
      "10/10 - 0s - loss: -4.2806e-02 - MI: 0.0429\n",
      "Epoch 361/1000\n",
      "10/10 - 0s - loss: -4.8303e-02 - MI: 0.0485\n",
      "Epoch 362/1000\n",
      "10/10 - 0s - loss: -5.2889e-02 - MI: 0.0529\n",
      "Epoch 363/1000\n",
      "10/10 - 0s - loss: -4.4980e-02 - MI: 0.0447\n",
      "Epoch 364/1000\n",
      "10/10 - 0s - loss: -4.4512e-02 - MI: 0.0447\n",
      "Epoch 365/1000\n",
      "10/10 - 0s - loss: -5.3124e-02 - MI: 0.0531\n",
      "Epoch 366/1000\n",
      "10/10 - 0s - loss: -5.0127e-02 - MI: 0.0502\n",
      "Epoch 367/1000\n",
      "10/10 - 0s - loss: -5.2265e-02 - MI: 0.0526\n",
      "Epoch 368/1000\n",
      "10/10 - 0s - loss: -4.8108e-02 - MI: 0.0479\n",
      "Epoch 369/1000\n",
      "10/10 - 0s - loss: -4.8471e-02 - MI: 0.0481\n",
      "Epoch 370/1000\n",
      "10/10 - 0s - loss: -4.5477e-02 - MI: 0.0455\n",
      "Epoch 371/1000\n",
      "10/10 - 0s - loss: -4.8067e-02 - MI: 0.0476\n",
      "Epoch 372/1000\n",
      "10/10 - 0s - loss: -4.7894e-02 - MI: 0.0476\n",
      "Epoch 373/1000\n",
      "10/10 - 0s - loss: -4.5765e-02 - MI: 0.0454\n",
      "Epoch 374/1000\n",
      "10/10 - 0s - loss: -4.8092e-02 - MI: 0.0477\n",
      "Epoch 375/1000\n",
      "10/10 - 0s - loss: -4.7581e-02 - MI: 0.0480\n",
      "Epoch 376/1000\n",
      "10/10 - 0s - loss: -5.0062e-02 - MI: 0.0494\n",
      "Epoch 377/1000\n",
      "10/10 - 0s - loss: -4.7118e-02 - MI: 0.0473\n",
      "Epoch 378/1000\n",
      "10/10 - 0s - loss: -4.6248e-02 - MI: 0.0466\n",
      "Epoch 379/1000\n",
      "10/10 - 0s - loss: -4.8769e-02 - MI: 0.0486\n",
      "Epoch 380/1000\n",
      "10/10 - 0s - loss: -4.3995e-02 - MI: 0.0435\n",
      "Epoch 381/1000\n",
      "10/10 - 0s - loss: -4.6660e-02 - MI: 0.0468\n",
      "Epoch 382/1000\n",
      "10/10 - 0s - loss: -4.6890e-02 - MI: 0.0470\n",
      "Epoch 383/1000\n",
      "10/10 - 0s - loss: -4.5399e-02 - MI: 0.0455\n",
      "Epoch 384/1000\n",
      "10/10 - 0s - loss: -4.7709e-02 - MI: 0.0481\n",
      "Epoch 385/1000\n",
      "10/10 - 0s - loss: -4.8693e-02 - MI: 0.0483\n",
      "Epoch 386/1000\n",
      "10/10 - 0s - loss: -4.8676e-02 - MI: 0.0488\n",
      "Epoch 387/1000\n",
      "10/10 - 0s - loss: -4.9464e-02 - MI: 0.0490\n",
      "Epoch 388/1000\n",
      "10/10 - 0s - loss: -4.6145e-02 - MI: 0.0456\n",
      "Epoch 389/1000\n",
      "10/10 - 0s - loss: -4.7272e-02 - MI: 0.0471\n",
      "Epoch 390/1000\n",
      "10/10 - 0s - loss: -4.3004e-02 - MI: 0.0427\n",
      "Epoch 391/1000\n",
      "10/10 - 0s - loss: -4.8719e-02 - MI: 0.0482\n",
      "Epoch 392/1000\n",
      "10/10 - 0s - loss: -4.3835e-02 - MI: 0.0438\n",
      "Epoch 393/1000\n",
      "10/10 - 0s - loss: -5.1009e-02 - MI: 0.0509\n",
      "Epoch 394/1000\n",
      "10/10 - 0s - loss: -5.2431e-02 - MI: 0.0525\n",
      "Epoch 395/1000\n",
      "10/10 - 0s - loss: -4.4997e-02 - MI: 0.0447\n",
      "Epoch 396/1000\n",
      "10/10 - 0s - loss: -5.0833e-02 - MI: 0.0510\n",
      "Epoch 397/1000\n",
      "10/10 - 0s - loss: -4.7688e-02 - MI: 0.0478\n",
      "Epoch 398/1000\n",
      "10/10 - 0s - loss: -4.5728e-02 - MI: 0.0456\n",
      "Epoch 399/1000\n",
      "10/10 - 0s - loss: -4.9723e-02 - MI: 0.0496\n",
      "Epoch 400/1000\n",
      "10/10 - 0s - loss: -4.6249e-02 - MI: 0.0451\n",
      "Epoch 401/1000\n",
      "10/10 - 0s - loss: -4.4576e-02 - MI: 0.0441\n",
      "Epoch 402/1000\n",
      "10/10 - 0s - loss: -5.2247e-02 - MI: 0.0522\n",
      "Epoch 403/1000\n",
      "10/10 - 0s - loss: -4.4823e-02 - MI: 0.0452\n",
      "Epoch 404/1000\n",
      "10/10 - 0s - loss: -4.6113e-02 - MI: 0.0458\n",
      "Epoch 405/1000\n",
      "10/10 - 0s - loss: -5.1422e-02 - MI: 0.0517\n",
      "Epoch 406/1000\n",
      "10/10 - 0s - loss: -5.1521e-02 - MI: 0.0521\n",
      "Epoch 407/1000\n",
      "10/10 - 0s - loss: -4.7387e-02 - MI: 0.0476\n",
      "Epoch 408/1000\n",
      "10/10 - 0s - loss: -4.8537e-02 - MI: 0.0483\n",
      "Epoch 409/1000\n",
      "10/10 - 0s - loss: -4.9219e-02 - MI: 0.0495\n",
      "Epoch 410/1000\n",
      "10/10 - 0s - loss: -4.5767e-02 - MI: 0.0454\n",
      "Epoch 411/1000\n",
      "10/10 - 0s - loss: -4.7233e-02 - MI: 0.0472\n",
      "Epoch 412/1000\n",
      "10/10 - 0s - loss: -4.6215e-02 - MI: 0.0463\n",
      "Epoch 413/1000\n",
      "10/10 - 0s - loss: -4.7520e-02 - MI: 0.0476\n",
      "Epoch 414/1000\n",
      "10/10 - 0s - loss: -5.0769e-02 - MI: 0.0507\n",
      "Epoch 415/1000\n",
      "10/10 - 0s - loss: -4.9768e-02 - MI: 0.0496\n",
      "Epoch 416/1000\n",
      "10/10 - 0s - loss: -4.3133e-02 - MI: 0.0439\n",
      "Epoch 417/1000\n",
      "10/10 - 0s - loss: -5.1781e-02 - MI: 0.0524\n",
      "Epoch 418/1000\n",
      "10/10 - 0s - loss: -5.2727e-02 - MI: 0.0526\n",
      "Epoch 419/1000\n",
      "10/10 - 0s - loss: -5.3950e-02 - MI: 0.0545\n",
      "Epoch 420/1000\n",
      "10/10 - 0s - loss: -4.6562e-02 - MI: 0.0475\n",
      "Epoch 421/1000\n",
      "10/10 - 0s - loss: -4.8028e-02 - MI: 0.0488\n",
      "Epoch 422/1000\n",
      "10/10 - 0s - loss: -4.8990e-02 - MI: 0.0495\n",
      "Epoch 423/1000\n",
      "10/10 - 0s - loss: -4.4699e-02 - MI: 0.0452\n",
      "Epoch 424/1000\n",
      "10/10 - 0s - loss: -5.2716e-02 - MI: 0.0532\n",
      "Epoch 425/1000\n",
      "10/10 - 0s - loss: -5.4010e-02 - MI: 0.0542\n",
      "Epoch 426/1000\n",
      "10/10 - 0s - loss: -4.8690e-02 - MI: 0.0489\n",
      "Epoch 427/1000\n",
      "10/10 - 0s - loss: -4.5804e-02 - MI: 0.0462\n",
      "Epoch 428/1000\n",
      "10/10 - 0s - loss: -4.4179e-02 - MI: 0.0439\n",
      "Epoch 429/1000\n",
      "10/10 - 0s - loss: -5.2985e-02 - MI: 0.0530\n",
      "Epoch 430/1000\n",
      "10/10 - 0s - loss: -4.8408e-02 - MI: 0.0488\n",
      "Epoch 431/1000\n",
      "10/10 - 0s - loss: -4.9417e-02 - MI: 0.0497\n",
      "Epoch 432/1000\n",
      "10/10 - 0s - loss: -4.4811e-02 - MI: 0.0450\n",
      "Epoch 433/1000\n",
      "10/10 - 0s - loss: -4.8380e-02 - MI: 0.0483\n",
      "Epoch 434/1000\n",
      "10/10 - 0s - loss: -5.1463e-02 - MI: 0.0510\n",
      "Epoch 435/1000\n",
      "10/10 - 0s - loss: -4.9769e-02 - MI: 0.0500\n",
      "Epoch 436/1000\n",
      "10/10 - 0s - loss: -5.0905e-02 - MI: 0.0503\n",
      "Epoch 437/1000\n",
      "10/10 - 0s - loss: -4.9187e-02 - MI: 0.0490\n",
      "Epoch 438/1000\n",
      "10/10 - 0s - loss: -4.5251e-02 - MI: 0.0455\n",
      "Epoch 439/1000\n",
      "10/10 - 0s - loss: -4.3708e-02 - MI: 0.0442\n",
      "Epoch 440/1000\n",
      "10/10 - 0s - loss: -5.2733e-02 - MI: 0.0533\n",
      "Epoch 441/1000\n",
      "10/10 - 0s - loss: -4.6861e-02 - MI: 0.0467\n",
      "Epoch 442/1000\n",
      "10/10 - 0s - loss: -4.6621e-02 - MI: 0.0468\n",
      "Epoch 443/1000\n",
      "10/10 - 0s - loss: -4.3706e-02 - MI: 0.0439\n",
      "Epoch 444/1000\n",
      "10/10 - 0s - loss: -4.5289e-02 - MI: 0.0458\n",
      "Epoch 445/1000\n",
      "10/10 - 0s - loss: -5.2112e-02 - MI: 0.0518\n",
      "Epoch 446/1000\n",
      "10/10 - 0s - loss: -4.7121e-02 - MI: 0.0471\n",
      "Epoch 447/1000\n",
      "10/10 - 0s - loss: -4.4392e-02 - MI: 0.0441\n",
      "Epoch 448/1000\n",
      "10/10 - 0s - loss: -4.4273e-02 - MI: 0.0447\n",
      "Epoch 449/1000\n",
      "10/10 - 0s - loss: -5.4027e-02 - MI: 0.0536\n",
      "Epoch 450/1000\n",
      "10/10 - 0s - loss: -4.9165e-02 - MI: 0.0491\n",
      "Epoch 451/1000\n",
      "10/10 - 0s - loss: -5.1645e-02 - MI: 0.0515\n",
      "Epoch 452/1000\n",
      "10/10 - 0s - loss: -5.2161e-02 - MI: 0.0527\n",
      "Epoch 453/1000\n",
      "10/10 - 0s - loss: -5.3017e-02 - MI: 0.0530\n",
      "Epoch 454/1000\n",
      "10/10 - 0s - loss: -4.9415e-02 - MI: 0.0490\n",
      "Epoch 455/1000\n",
      "10/10 - 0s - loss: -5.0794e-02 - MI: 0.0507\n",
      "Epoch 456/1000\n",
      "10/10 - 0s - loss: -4.2324e-02 - MI: 0.0416\n",
      "Epoch 457/1000\n",
      "10/10 - 0s - loss: -4.8829e-02 - MI: 0.0492\n",
      "Epoch 458/1000\n",
      "10/10 - 0s - loss: -4.5279e-02 - MI: 0.0452\n",
      "Epoch 459/1000\n",
      "10/10 - 0s - loss: -4.8646e-02 - MI: 0.0487\n",
      "Epoch 460/1000\n",
      "10/10 - 0s - loss: -4.3209e-02 - MI: 0.0433\n",
      "Epoch 461/1000\n",
      "10/10 - 0s - loss: -4.9781e-02 - MI: 0.0498\n",
      "Epoch 462/1000\n",
      "10/10 - 0s - loss: -4.8891e-02 - MI: 0.0489\n",
      "Epoch 463/1000\n",
      "10/10 - 0s - loss: -4.9010e-02 - MI: 0.0490\n",
      "Epoch 464/1000\n",
      "10/10 - 0s - loss: -5.3425e-02 - MI: 0.0535\n",
      "Epoch 465/1000\n",
      "10/10 - 0s - loss: -4.6805e-02 - MI: 0.0469\n",
      "Epoch 466/1000\n",
      "10/10 - 0s - loss: -4.4075e-02 - MI: 0.0434\n",
      "Epoch 467/1000\n",
      "10/10 - 0s - loss: -4.8878e-02 - MI: 0.0490\n",
      "Epoch 468/1000\n",
      "10/10 - 0s - loss: -4.7685e-02 - MI: 0.0481\n",
      "Epoch 469/1000\n",
      "10/10 - 0s - loss: -4.0680e-02 - MI: 0.0407\n",
      "Epoch 470/1000\n",
      "10/10 - 0s - loss: -4.7396e-02 - MI: 0.0478\n",
      "Epoch 471/1000\n",
      "10/10 - 0s - loss: -5.0881e-02 - MI: 0.0509\n",
      "Epoch 472/1000\n",
      "10/10 - 0s - loss: -4.4029e-02 - MI: 0.0442\n",
      "Epoch 473/1000\n",
      "10/10 - 0s - loss: -5.0170e-02 - MI: 0.0500\n",
      "Epoch 474/1000\n",
      "10/10 - 0s - loss: -5.3958e-02 - MI: 0.0538\n",
      "Epoch 475/1000\n",
      "10/10 - 0s - loss: -4.8505e-02 - MI: 0.0486\n",
      "Epoch 476/1000\n",
      "10/10 - 0s - loss: -4.7336e-02 - MI: 0.0475\n",
      "Epoch 477/1000\n",
      "10/10 - 0s - loss: -3.9631e-02 - MI: 0.0392\n",
      "Epoch 478/1000\n",
      "10/10 - 0s - loss: -4.5929e-02 - MI: 0.0463\n",
      "Epoch 479/1000\n",
      "10/10 - 0s - loss: -4.6956e-02 - MI: 0.0474\n",
      "Epoch 480/1000\n",
      "10/10 - 0s - loss: -4.9168e-02 - MI: 0.0488\n",
      "Epoch 481/1000\n",
      "10/10 - 0s - loss: -4.6967e-02 - MI: 0.0471\n",
      "Epoch 482/1000\n",
      "10/10 - 0s - loss: -4.6542e-02 - MI: 0.0462\n",
      "Epoch 483/1000\n",
      "10/10 - 0s - loss: -4.5462e-02 - MI: 0.0456\n",
      "Epoch 484/1000\n",
      "10/10 - 0s - loss: -4.6784e-02 - MI: 0.0467\n",
      "Epoch 485/1000\n",
      "10/10 - 0s - loss: -4.9729e-02 - MI: 0.0497\n",
      "Epoch 486/1000\n",
      "10/10 - 0s - loss: -5.1653e-02 - MI: 0.0520\n",
      "Epoch 487/1000\n",
      "10/10 - 0s - loss: -5.1653e-02 - MI: 0.0514\n",
      "Epoch 488/1000\n",
      "10/10 - 0s - loss: -4.7408e-02 - MI: 0.0478\n",
      "Epoch 489/1000\n",
      "10/10 - 0s - loss: -4.7981e-02 - MI: 0.0478\n",
      "Epoch 490/1000\n",
      "10/10 - 0s - loss: -5.3903e-02 - MI: 0.0540\n",
      "Epoch 491/1000\n",
      "10/10 - 0s - loss: -4.8083e-02 - MI: 0.0484\n",
      "Epoch 492/1000\n",
      "10/10 - 0s - loss: -4.9634e-02 - MI: 0.0498\n",
      "Epoch 493/1000\n",
      "10/10 - 0s - loss: -4.9619e-02 - MI: 0.0494\n",
      "Epoch 494/1000\n",
      "10/10 - 0s - loss: -5.3999e-02 - MI: 0.0542\n",
      "Epoch 495/1000\n",
      "10/10 - 0s - loss: -4.7473e-02 - MI: 0.0475\n",
      "Epoch 496/1000\n",
      "10/10 - 0s - loss: -4.8949e-02 - MI: 0.0491\n",
      "Epoch 497/1000\n",
      "10/10 - 0s - loss: -3.8161e-02 - MI: 0.0383\n",
      "Epoch 498/1000\n",
      "10/10 - 0s - loss: -4.9773e-02 - MI: 0.0497\n",
      "Epoch 499/1000\n",
      "10/10 - 0s - loss: -4.4084e-02 - MI: 0.0435\n",
      "Epoch 500/1000\n",
      "10/10 - 0s - loss: -4.3797e-02 - MI: 0.0439\n",
      "Epoch 501/1000\n",
      "10/10 - 0s - loss: -4.8015e-02 - MI: 0.0482\n",
      "Epoch 502/1000\n",
      "10/10 - 0s - loss: -4.7910e-02 - MI: 0.0479\n",
      "Epoch 503/1000\n",
      "10/10 - 0s - loss: -5.0441e-02 - MI: 0.0508\n",
      "Epoch 504/1000\n",
      "10/10 - 0s - loss: -5.0324e-02 - MI: 0.0508\n",
      "Epoch 505/1000\n",
      "10/10 - 0s - loss: -4.9804e-02 - MI: 0.0496\n",
      "Epoch 506/1000\n",
      "10/10 - 0s - loss: -5.1278e-02 - MI: 0.0516\n",
      "Epoch 507/1000\n",
      "10/10 - 0s - loss: -5.0015e-02 - MI: 0.0504\n",
      "Epoch 508/1000\n",
      "10/10 - 0s - loss: -4.8719e-02 - MI: 0.0490\n",
      "Epoch 509/1000\n",
      "10/10 - 0s - loss: -4.6735e-02 - MI: 0.0466\n",
      "Epoch 510/1000\n",
      "10/10 - 0s - loss: -4.8916e-02 - MI: 0.0486\n",
      "Epoch 511/1000\n",
      "10/10 - 0s - loss: -4.5277e-02 - MI: 0.0458\n",
      "Epoch 512/1000\n",
      "10/10 - 0s - loss: -4.9221e-02 - MI: 0.0489\n",
      "Epoch 513/1000\n",
      "10/10 - 0s - loss: -4.5969e-02 - MI: 0.0456\n",
      "Epoch 514/1000\n",
      "10/10 - 0s - loss: -5.0262e-02 - MI: 0.0504\n",
      "Epoch 515/1000\n",
      "10/10 - 0s - loss: -5.0237e-02 - MI: 0.0498\n",
      "Epoch 516/1000\n",
      "10/10 - 0s - loss: -5.0678e-02 - MI: 0.0505\n",
      "Epoch 517/1000\n",
      "10/10 - 0s - loss: -5.0569e-02 - MI: 0.0507\n",
      "Epoch 518/1000\n",
      "10/10 - 0s - loss: -4.7405e-02 - MI: 0.0477\n",
      "Epoch 519/1000\n",
      "10/10 - 0s - loss: -4.3780e-02 - MI: 0.0435\n",
      "Epoch 520/1000\n",
      "10/10 - 0s - loss: -5.0201e-02 - MI: 0.0505\n",
      "Epoch 521/1000\n",
      "10/10 - 0s - loss: -4.9741e-02 - MI: 0.0488\n",
      "Epoch 522/1000\n",
      "10/10 - 0s - loss: -4.7950e-02 - MI: 0.0481\n",
      "Epoch 523/1000\n",
      "10/10 - 0s - loss: -5.1232e-02 - MI: 0.0514\n",
      "Epoch 524/1000\n",
      "10/10 - 0s - loss: -4.8526e-02 - MI: 0.0485\n",
      "Epoch 525/1000\n",
      "10/10 - 0s - loss: -5.4214e-02 - MI: 0.0547\n",
      "Epoch 526/1000\n",
      "10/10 - 0s - loss: -4.8315e-02 - MI: 0.0486\n",
      "Epoch 527/1000\n",
      "10/10 - 0s - loss: -3.7632e-02 - MI: 0.0382\n",
      "Epoch 528/1000\n",
      "10/10 - 0s - loss: -4.5673e-02 - MI: 0.0456\n",
      "Epoch 529/1000\n",
      "10/10 - 0s - loss: -4.8243e-02 - MI: 0.0483\n",
      "Epoch 530/1000\n",
      "10/10 - 0s - loss: -4.5486e-02 - MI: 0.0455\n",
      "Epoch 531/1000\n",
      "10/10 - 0s - loss: -5.2409e-02 - MI: 0.0526\n",
      "Epoch 532/1000\n",
      "10/10 - 0s - loss: -5.4392e-02 - MI: 0.0549\n",
      "Epoch 533/1000\n",
      "10/10 - 0s - loss: -5.1465e-02 - MI: 0.0514\n",
      "Epoch 534/1000\n",
      "10/10 - 0s - loss: -5.4534e-02 - MI: 0.0552\n",
      "Epoch 535/1000\n",
      "10/10 - 0s - loss: -4.9416e-02 - MI: 0.0503\n",
      "Epoch 536/1000\n",
      "10/10 - 0s - loss: -4.5141e-02 - MI: 0.0453\n",
      "Epoch 537/1000\n",
      "10/10 - 0s - loss: -4.3440e-02 - MI: 0.0435\n",
      "Epoch 538/1000\n",
      "10/10 - 0s - loss: -4.6826e-02 - MI: 0.0466\n",
      "Epoch 539/1000\n",
      "10/10 - 0s - loss: -4.7474e-02 - MI: 0.0472\n",
      "Epoch 540/1000\n",
      "10/10 - 0s - loss: -5.0009e-02 - MI: 0.0497\n",
      "Epoch 541/1000\n",
      "10/10 - 0s - loss: -4.8997e-02 - MI: 0.0489\n",
      "Epoch 542/1000\n",
      "10/10 - 0s - loss: -5.4537e-02 - MI: 0.0550\n",
      "Epoch 543/1000\n",
      "10/10 - 0s - loss: -5.5689e-02 - MI: 0.0556\n",
      "Epoch 544/1000\n",
      "10/10 - 0s - loss: -5.2354e-02 - MI: 0.0520\n",
      "Epoch 545/1000\n",
      "10/10 - 0s - loss: -5.1379e-02 - MI: 0.0517\n",
      "Epoch 546/1000\n",
      "10/10 - 0s - loss: -4.9731e-02 - MI: 0.0500\n",
      "Epoch 547/1000\n",
      "10/10 - 0s - loss: -5.1970e-02 - MI: 0.0520\n",
      "Epoch 548/1000\n",
      "10/10 - 0s - loss: -4.6709e-02 - MI: 0.0467\n",
      "Epoch 549/1000\n",
      "10/10 - 0s - loss: -5.1525e-02 - MI: 0.0521\n",
      "Epoch 550/1000\n",
      "10/10 - 0s - loss: -5.0813e-02 - MI: 0.0508\n",
      "Epoch 551/1000\n",
      "10/10 - 0s - loss: -5.5654e-02 - MI: 0.0558\n",
      "Epoch 552/1000\n",
      "10/10 - 0s - loss: -4.6194e-02 - MI: 0.0464\n",
      "Epoch 553/1000\n",
      "10/10 - 0s - loss: -5.3539e-02 - MI: 0.0538\n",
      "Epoch 554/1000\n",
      "10/10 - 0s - loss: -4.9887e-02 - MI: 0.0495\n",
      "Epoch 555/1000\n",
      "10/10 - 0s - loss: -4.9329e-02 - MI: 0.0496\n",
      "Epoch 556/1000\n",
      "10/10 - 0s - loss: -4.8810e-02 - MI: 0.0485\n",
      "Epoch 557/1000\n",
      "10/10 - 0s - loss: -5.1029e-02 - MI: 0.0511\n",
      "Epoch 558/1000\n",
      "10/10 - 0s - loss: -4.6403e-02 - MI: 0.0469\n",
      "Epoch 559/1000\n",
      "10/10 - 0s - loss: -5.1582e-02 - MI: 0.0518\n",
      "Epoch 560/1000\n",
      "10/10 - 0s - loss: -4.6354e-02 - MI: 0.0462\n",
      "Epoch 561/1000\n",
      "10/10 - 0s - loss: -4.9673e-02 - MI: 0.0500\n",
      "Epoch 562/1000\n",
      "10/10 - 0s - loss: -4.6389e-02 - MI: 0.0468\n",
      "Epoch 563/1000\n",
      "10/10 - 0s - loss: -4.6939e-02 - MI: 0.0469\n",
      "Epoch 564/1000\n",
      "10/10 - 0s - loss: -4.9970e-02 - MI: 0.0498\n",
      "Epoch 565/1000\n",
      "10/10 - 0s - loss: -4.9691e-02 - MI: 0.0501\n",
      "Epoch 566/1000\n",
      "10/10 - 0s - loss: -5.1540e-02 - MI: 0.0517\n",
      "Epoch 567/1000\n",
      "10/10 - 0s - loss: -5.0923e-02 - MI: 0.0507\n",
      "Epoch 568/1000\n",
      "10/10 - 0s - loss: -4.3523e-02 - MI: 0.0433\n",
      "Epoch 569/1000\n",
      "10/10 - 0s - loss: -5.4393e-02 - MI: 0.0544\n",
      "Epoch 570/1000\n",
      "10/10 - 0s - loss: -5.3767e-02 - MI: 0.0539\n",
      "Epoch 571/1000\n",
      "10/10 - 0s - loss: -5.1831e-02 - MI: 0.0516\n",
      "Epoch 572/1000\n",
      "10/10 - 0s - loss: -4.8608e-02 - MI: 0.0484\n",
      "Epoch 573/1000\n",
      "10/10 - 0s - loss: -5.1076e-02 - MI: 0.0514\n",
      "Epoch 574/1000\n",
      "10/10 - 0s - loss: -5.0334e-02 - MI: 0.0504\n",
      "Epoch 575/1000\n",
      "10/10 - 0s - loss: -4.4466e-02 - MI: 0.0451\n",
      "Epoch 576/1000\n",
      "10/10 - 0s - loss: -4.8728e-02 - MI: 0.0491\n",
      "Epoch 577/1000\n",
      "10/10 - 0s - loss: -5.2843e-02 - MI: 0.0523\n",
      "Epoch 578/1000\n",
      "10/10 - 0s - loss: -5.3114e-02 - MI: 0.0534\n",
      "Epoch 579/1000\n",
      "10/10 - 0s - loss: -4.8383e-02 - MI: 0.0482\n",
      "Epoch 580/1000\n",
      "10/10 - 0s - loss: -4.2328e-02 - MI: 0.0424\n",
      "Epoch 581/1000\n",
      "10/10 - 0s - loss: -4.9374e-02 - MI: 0.0494\n",
      "Epoch 582/1000\n",
      "10/10 - 0s - loss: -4.5734e-02 - MI: 0.0453\n",
      "Epoch 583/1000\n",
      "10/10 - 0s - loss: -5.3075e-02 - MI: 0.0532\n",
      "Epoch 584/1000\n",
      "10/10 - 0s - loss: -4.7618e-02 - MI: 0.0477\n",
      "Epoch 585/1000\n",
      "10/10 - 0s - loss: -5.0966e-02 - MI: 0.0508\n",
      "Epoch 586/1000\n",
      "10/10 - 0s - loss: -4.9228e-02 - MI: 0.0487\n",
      "Epoch 587/1000\n",
      "10/10 - 0s - loss: -5.0489e-02 - MI: 0.0505\n",
      "Epoch 588/1000\n",
      "10/10 - 0s - loss: -5.0975e-02 - MI: 0.0508\n",
      "Epoch 589/1000\n",
      "10/10 - 0s - loss: -4.7558e-02 - MI: 0.0475\n",
      "Epoch 590/1000\n",
      "10/10 - 0s - loss: -4.9008e-02 - MI: 0.0490\n",
      "Epoch 591/1000\n",
      "10/10 - 0s - loss: -5.0455e-02 - MI: 0.0506\n",
      "Epoch 592/1000\n",
      "10/10 - 0s - loss: -4.2478e-02 - MI: 0.0425\n",
      "Epoch 593/1000\n",
      "10/10 - 0s - loss: -4.4249e-02 - MI: 0.0438\n",
      "Epoch 594/1000\n",
      "10/10 - 0s - loss: -4.3718e-02 - MI: 0.0438\n",
      "Epoch 595/1000\n",
      "10/10 - 0s - loss: -4.8978e-02 - MI: 0.0496\n",
      "Epoch 596/1000\n",
      "10/10 - 0s - loss: -5.0834e-02 - MI: 0.0508\n",
      "Epoch 597/1000\n",
      "10/10 - 0s - loss: -5.1739e-02 - MI: 0.0515\n",
      "Epoch 598/1000\n",
      "10/10 - 0s - loss: -5.4957e-02 - MI: 0.0550\n",
      "Epoch 599/1000\n",
      "10/10 - 0s - loss: -4.7180e-02 - MI: 0.0473\n",
      "Epoch 600/1000\n",
      "10/10 - 0s - loss: -4.9307e-02 - MI: 0.0498\n",
      "Epoch 601/1000\n",
      "10/10 - 0s - loss: -5.1136e-02 - MI: 0.0515\n",
      "Epoch 602/1000\n",
      "10/10 - 0s - loss: -5.0802e-02 - MI: 0.0505\n",
      "Epoch 603/1000\n",
      "10/10 - 0s - loss: -5.3026e-02 - MI: 0.0528\n",
      "Epoch 604/1000\n",
      "10/10 - 0s - loss: -4.4811e-02 - MI: 0.0447\n",
      "Epoch 605/1000\n",
      "10/10 - 0s - loss: -5.0367e-02 - MI: 0.0505\n",
      "Epoch 606/1000\n",
      "10/10 - 0s - loss: -5.0648e-02 - MI: 0.0503\n",
      "Epoch 607/1000\n",
      "10/10 - 0s - loss: -5.3260e-02 - MI: 0.0533\n",
      "Epoch 608/1000\n",
      "10/10 - 0s - loss: -5.6727e-02 - MI: 0.0567\n",
      "Epoch 609/1000\n",
      "10/10 - 0s - loss: -4.9552e-02 - MI: 0.0495\n",
      "Epoch 610/1000\n",
      "10/10 - 0s - loss: -4.4205e-02 - MI: 0.0442\n",
      "Epoch 611/1000\n",
      "10/10 - 0s - loss: -4.8467e-02 - MI: 0.0482\n",
      "Epoch 612/1000\n",
      "10/10 - 0s - loss: -4.8931e-02 - MI: 0.0493\n",
      "Epoch 613/1000\n",
      "10/10 - 0s - loss: -5.4262e-02 - MI: 0.0544\n",
      "Epoch 614/1000\n",
      "10/10 - 0s - loss: -5.1564e-02 - MI: 0.0521\n",
      "Epoch 615/1000\n",
      "10/10 - 0s - loss: -4.6543e-02 - MI: 0.0462\n",
      "Epoch 616/1000\n",
      "10/10 - 0s - loss: -5.2398e-02 - MI: 0.0521\n",
      "Epoch 617/1000\n",
      "10/10 - 0s - loss: -4.5366e-02 - MI: 0.0456\n",
      "Epoch 618/1000\n",
      "10/10 - 0s - loss: -5.2274e-02 - MI: 0.0526\n",
      "Epoch 619/1000\n",
      "10/10 - 0s - loss: -4.9614e-02 - MI: 0.0493\n",
      "Epoch 620/1000\n",
      "10/10 - 0s - loss: -4.7265e-02 - MI: 0.0468\n",
      "Epoch 621/1000\n",
      "10/10 - 0s - loss: -5.5769e-02 - MI: 0.0557\n",
      "Epoch 622/1000\n",
      "10/10 - 0s - loss: -5.0115e-02 - MI: 0.0494\n",
      "Epoch 623/1000\n",
      "10/10 - 0s - loss: -4.9394e-02 - MI: 0.0497\n",
      "Epoch 624/1000\n",
      "10/10 - 0s - loss: -4.7507e-02 - MI: 0.0470\n",
      "Epoch 625/1000\n",
      "10/10 - 0s - loss: -5.4449e-02 - MI: 0.0548\n",
      "Epoch 626/1000\n",
      "10/10 - 0s - loss: -4.5991e-02 - MI: 0.0460\n",
      "Epoch 627/1000\n",
      "10/10 - 0s - loss: -5.7954e-02 - MI: 0.0579\n",
      "Epoch 628/1000\n",
      "10/10 - 0s - loss: -5.4303e-02 - MI: 0.0547\n",
      "Epoch 629/1000\n",
      "10/10 - 0s - loss: -4.6210e-02 - MI: 0.0462\n",
      "Epoch 630/1000\n",
      "10/10 - 0s - loss: -4.8271e-02 - MI: 0.0481\n",
      "Epoch 631/1000\n",
      "10/10 - 0s - loss: -4.5849e-02 - MI: 0.0458\n",
      "Epoch 632/1000\n",
      "10/10 - 0s - loss: -4.8329e-02 - MI: 0.0484\n",
      "Epoch 633/1000\n",
      "10/10 - 0s - loss: -4.6415e-02 - MI: 0.0462\n",
      "Epoch 634/1000\n",
      "10/10 - 0s - loss: -4.6090e-02 - MI: 0.0455\n",
      "Epoch 635/1000\n",
      "10/10 - 0s - loss: -4.5655e-02 - MI: 0.0463\n",
      "Epoch 636/1000\n",
      "10/10 - 0s - loss: -4.6694e-02 - MI: 0.0460\n",
      "Epoch 637/1000\n",
      "10/10 - 0s - loss: -5.1580e-02 - MI: 0.0508\n",
      "Epoch 638/1000\n",
      "10/10 - 0s - loss: -4.9749e-02 - MI: 0.0497\n",
      "Epoch 639/1000\n",
      "10/10 - 0s - loss: -5.1850e-02 - MI: 0.0520\n",
      "Epoch 640/1000\n",
      "10/10 - 0s - loss: -5.0065e-02 - MI: 0.0500\n",
      "Epoch 641/1000\n",
      "10/10 - 0s - loss: -4.8356e-02 - MI: 0.0487\n",
      "Epoch 642/1000\n",
      "10/10 - 0s - loss: -5.1373e-02 - MI: 0.0514\n",
      "Epoch 643/1000\n",
      "10/10 - 0s - loss: -4.8618e-02 - MI: 0.0482\n",
      "Epoch 644/1000\n",
      "10/10 - 0s - loss: -4.8711e-02 - MI: 0.0483\n",
      "Epoch 645/1000\n",
      "10/10 - 0s - loss: -5.3754e-02 - MI: 0.0537\n",
      "Epoch 646/1000\n",
      "10/10 - 0s - loss: -5.2360e-02 - MI: 0.0523\n",
      "Epoch 647/1000\n",
      "10/10 - 0s - loss: -5.4742e-02 - MI: 0.0554\n",
      "Epoch 648/1000\n",
      "10/10 - 0s - loss: -5.2153e-02 - MI: 0.0525\n",
      "Epoch 649/1000\n",
      "10/10 - 0s - loss: -5.2419e-02 - MI: 0.0534\n",
      "Epoch 650/1000\n",
      "10/10 - 0s - loss: -5.3294e-02 - MI: 0.0535\n",
      "Epoch 651/1000\n",
      "10/10 - 0s - loss: -5.0189e-02 - MI: 0.0503\n",
      "Epoch 652/1000\n",
      "10/10 - 0s - loss: -5.0279e-02 - MI: 0.0509\n",
      "Epoch 653/1000\n",
      "10/10 - 0s - loss: -4.9186e-02 - MI: 0.0498\n",
      "Epoch 654/1000\n",
      "10/10 - 0s - loss: -4.5366e-02 - MI: 0.0451\n",
      "Epoch 655/1000\n",
      "10/10 - 0s - loss: -4.9438e-02 - MI: 0.0495\n",
      "Epoch 656/1000\n",
      "10/10 - 0s - loss: -5.5562e-02 - MI: 0.0551\n",
      "Epoch 657/1000\n",
      "10/10 - 0s - loss: -5.1344e-02 - MI: 0.0514\n",
      "Epoch 658/1000\n",
      "10/10 - 0s - loss: -4.7161e-02 - MI: 0.0469\n",
      "Epoch 659/1000\n",
      "10/10 - 0s - loss: -4.9267e-02 - MI: 0.0485\n",
      "Epoch 660/1000\n",
      "10/10 - 0s - loss: -5.0484e-02 - MI: 0.0511\n",
      "Epoch 661/1000\n",
      "10/10 - 0s - loss: -5.0987e-02 - MI: 0.0511\n",
      "Epoch 662/1000\n",
      "10/10 - 0s - loss: -5.1855e-02 - MI: 0.0517\n",
      "Epoch 663/1000\n",
      "10/10 - 0s - loss: -5.4071e-02 - MI: 0.0545\n",
      "Epoch 664/1000\n",
      "10/10 - 0s - loss: -4.9983e-02 - MI: 0.0500\n",
      "Epoch 665/1000\n",
      "10/10 - 0s - loss: -4.4399e-02 - MI: 0.0436\n",
      "Epoch 666/1000\n",
      "10/10 - 0s - loss: -5.3319e-02 - MI: 0.0532\n",
      "Epoch 667/1000\n",
      "10/10 - 0s - loss: -5.5635e-02 - MI: 0.0558\n",
      "Epoch 668/1000\n",
      "10/10 - 0s - loss: -5.0821e-02 - MI: 0.0507\n",
      "Epoch 669/1000\n",
      "10/10 - 0s - loss: -4.9190e-02 - MI: 0.0497\n",
      "Epoch 670/1000\n",
      "10/10 - 0s - loss: -5.0412e-02 - MI: 0.0497\n",
      "Epoch 671/1000\n",
      "10/10 - 0s - loss: -4.8975e-02 - MI: 0.0486\n",
      "Epoch 672/1000\n",
      "10/10 - 0s - loss: -4.5777e-02 - MI: 0.0458\n",
      "Epoch 673/1000\n",
      "10/10 - 0s - loss: -5.1665e-02 - MI: 0.0516\n",
      "Epoch 674/1000\n",
      "10/10 - 0s - loss: -5.0712e-02 - MI: 0.0497\n",
      "Epoch 675/1000\n",
      "10/10 - 0s - loss: -4.7525e-02 - MI: 0.0473\n",
      "Epoch 676/1000\n",
      "10/10 - 0s - loss: -5.2667e-02 - MI: 0.0521\n",
      "Epoch 677/1000\n",
      "10/10 - 0s - loss: -5.2495e-02 - MI: 0.0523\n",
      "Epoch 678/1000\n",
      "10/10 - 0s - loss: -4.6582e-02 - MI: 0.0466\n",
      "Epoch 679/1000\n",
      "10/10 - 0s - loss: -5.1185e-02 - MI: 0.0513\n",
      "Epoch 680/1000\n",
      "10/10 - 0s - loss: -4.8264e-02 - MI: 0.0480\n",
      "Epoch 681/1000\n",
      "10/10 - 0s - loss: -4.9937e-02 - MI: 0.0500\n",
      "Epoch 682/1000\n",
      "10/10 - 0s - loss: -4.9258e-02 - MI: 0.0497\n",
      "Epoch 683/1000\n",
      "10/10 - 0s - loss: -5.4828e-02 - MI: 0.0548\n",
      "Epoch 684/1000\n",
      "10/10 - 0s - loss: -5.1950e-02 - MI: 0.0521\n",
      "Epoch 685/1000\n",
      "10/10 - 0s - loss: -5.5552e-02 - MI: 0.0557\n",
      "Epoch 686/1000\n",
      "10/10 - 0s - loss: -5.3931e-02 - MI: 0.0541\n",
      "Epoch 687/1000\n",
      "10/10 - 0s - loss: -5.0772e-02 - MI: 0.0498\n",
      "Epoch 688/1000\n",
      "10/10 - 0s - loss: -5.1825e-02 - MI: 0.0517\n",
      "Epoch 689/1000\n",
      "10/10 - 0s - loss: -5.2935e-02 - MI: 0.0528\n",
      "Epoch 690/1000\n",
      "10/10 - 0s - loss: -5.0199e-02 - MI: 0.0503\n",
      "Epoch 691/1000\n",
      "10/10 - 0s - loss: -5.3453e-02 - MI: 0.0528\n",
      "Epoch 692/1000\n",
      "10/10 - 0s - loss: -5.5909e-02 - MI: 0.0560\n",
      "Epoch 693/1000\n",
      "10/10 - 0s - loss: -5.7970e-02 - MI: 0.0580\n",
      "Epoch 694/1000\n",
      "10/10 - 0s - loss: -4.8725e-02 - MI: 0.0486\n",
      "Epoch 695/1000\n",
      "10/10 - 0s - loss: -5.4126e-02 - MI: 0.0541\n",
      "Epoch 696/1000\n",
      "10/10 - 0s - loss: -4.9720e-02 - MI: 0.0498\n",
      "Epoch 697/1000\n",
      "10/10 - 0s - loss: -5.0946e-02 - MI: 0.0509\n",
      "Epoch 698/1000\n",
      "10/10 - 0s - loss: -5.0801e-02 - MI: 0.0512\n",
      "Epoch 699/1000\n",
      "10/10 - 0s - loss: -5.4060e-02 - MI: 0.0545\n",
      "Epoch 700/1000\n",
      "10/10 - 0s - loss: -4.7030e-02 - MI: 0.0473\n",
      "Epoch 701/1000\n",
      "10/10 - 0s - loss: -5.0352e-02 - MI: 0.0501\n",
      "Epoch 702/1000\n",
      "10/10 - 0s - loss: -5.0590e-02 - MI: 0.0509\n",
      "Epoch 703/1000\n",
      "10/10 - 0s - loss: -4.8489e-02 - MI: 0.0483\n",
      "Epoch 704/1000\n",
      "10/10 - 0s - loss: -5.1891e-02 - MI: 0.0517\n",
      "Epoch 705/1000\n",
      "10/10 - 0s - loss: -5.0921e-02 - MI: 0.0503\n",
      "Epoch 706/1000\n",
      "10/10 - 0s - loss: -4.7678e-02 - MI: 0.0479\n",
      "Epoch 707/1000\n",
      "10/10 - 0s - loss: -5.0152e-02 - MI: 0.0494\n",
      "Epoch 708/1000\n",
      "10/10 - 0s - loss: -5.5043e-02 - MI: 0.0547\n",
      "Epoch 709/1000\n",
      "10/10 - 0s - loss: -4.8020e-02 - MI: 0.0477\n",
      "Epoch 710/1000\n",
      "10/10 - 0s - loss: -5.3092e-02 - MI: 0.0534\n",
      "Epoch 711/1000\n",
      "10/10 - 0s - loss: -4.3133e-02 - MI: 0.0426\n",
      "Epoch 712/1000\n",
      "10/10 - 0s - loss: -5.1342e-02 - MI: 0.0511\n",
      "Epoch 713/1000\n",
      "10/10 - 0s - loss: -4.8848e-02 - MI: 0.0483\n",
      "Epoch 714/1000\n",
      "10/10 - 0s - loss: -5.0700e-02 - MI: 0.0507\n",
      "Epoch 715/1000\n",
      "10/10 - 0s - loss: -5.3815e-02 - MI: 0.0531\n",
      "Epoch 716/1000\n",
      "10/10 - 0s - loss: -5.0197e-02 - MI: 0.0507\n",
      "Epoch 717/1000\n",
      "10/10 - 0s - loss: -4.9800e-02 - MI: 0.0501\n",
      "Epoch 718/1000\n",
      "10/10 - 0s - loss: -4.9601e-02 - MI: 0.0493\n",
      "Epoch 719/1000\n",
      "10/10 - 0s - loss: -5.1280e-02 - MI: 0.0508\n",
      "Epoch 720/1000\n",
      "10/10 - 0s - loss: -4.9172e-02 - MI: 0.0494\n",
      "Epoch 721/1000\n",
      "10/10 - 0s - loss: -5.3803e-02 - MI: 0.0535\n",
      "Epoch 722/1000\n",
      "10/10 - 0s - loss: -5.0583e-02 - MI: 0.0502\n",
      "Epoch 723/1000\n",
      "10/10 - 0s - loss: -5.2051e-02 - MI: 0.0521\n",
      "Epoch 724/1000\n",
      "10/10 - 0s - loss: -5.4497e-02 - MI: 0.0539\n",
      "Epoch 725/1000\n",
      "10/10 - 0s - loss: -5.4379e-02 - MI: 0.0541\n",
      "Epoch 726/1000\n",
      "10/10 - 0s - loss: -4.9450e-02 - MI: 0.0487\n",
      "Epoch 727/1000\n",
      "10/10 - 0s - loss: -4.7998e-02 - MI: 0.0480\n",
      "Epoch 728/1000\n",
      "10/10 - 0s - loss: -5.4099e-02 - MI: 0.0542\n",
      "Epoch 729/1000\n",
      "10/10 - 0s - loss: -4.9991e-02 - MI: 0.0501\n",
      "Epoch 730/1000\n",
      "10/10 - 0s - loss: -5.1459e-02 - MI: 0.0518\n",
      "Epoch 731/1000\n",
      "10/10 - 0s - loss: -5.3064e-02 - MI: 0.0533\n",
      "Epoch 732/1000\n",
      "10/10 - 0s - loss: -4.7635e-02 - MI: 0.0479\n",
      "Epoch 733/1000\n",
      "10/10 - 0s - loss: -4.7334e-02 - MI: 0.0468\n",
      "Epoch 734/1000\n",
      "10/10 - 0s - loss: -5.0922e-02 - MI: 0.0512\n",
      "Epoch 735/1000\n",
      "10/10 - 0s - loss: -5.1389e-02 - MI: 0.0517\n",
      "Epoch 736/1000\n",
      "10/10 - 0s - loss: -5.6200e-02 - MI: 0.0566\n",
      "Epoch 737/1000\n",
      "10/10 - 0s - loss: -5.7171e-02 - MI: 0.0570\n",
      "Epoch 738/1000\n",
      "10/10 - 0s - loss: -4.4915e-02 - MI: 0.0448\n",
      "Epoch 739/1000\n",
      "10/10 - 0s - loss: -4.8282e-02 - MI: 0.0478\n",
      "Epoch 740/1000\n",
      "10/10 - 0s - loss: -5.9353e-02 - MI: 0.0598\n",
      "Epoch 741/1000\n",
      "10/10 - 0s - loss: -5.6174e-02 - MI: 0.0561\n",
      "Epoch 742/1000\n",
      "10/10 - 0s - loss: -4.6702e-02 - MI: 0.0465\n",
      "Epoch 743/1000\n",
      "10/10 - 0s - loss: -5.2036e-02 - MI: 0.0515\n",
      "Epoch 744/1000\n",
      "10/10 - 0s - loss: -5.5455e-02 - MI: 0.0554\n",
      "Epoch 745/1000\n",
      "10/10 - 0s - loss: -4.9921e-02 - MI: 0.0504\n",
      "Epoch 746/1000\n",
      "10/10 - 0s - loss: -4.8695e-02 - MI: 0.0492\n",
      "Epoch 747/1000\n",
      "10/10 - 0s - loss: -5.0624e-02 - MI: 0.0508\n",
      "Epoch 748/1000\n",
      "10/10 - 0s - loss: -5.4111e-02 - MI: 0.0539\n",
      "Epoch 749/1000\n",
      "10/10 - 0s - loss: -5.4728e-02 - MI: 0.0552\n",
      "Epoch 750/1000\n",
      "10/10 - 0s - loss: -6.1523e-02 - MI: 0.0614\n",
      "Epoch 751/1000\n",
      "10/10 - 0s - loss: -5.6249e-02 - MI: 0.0565\n",
      "Epoch 752/1000\n",
      "10/10 - 0s - loss: -5.3610e-02 - MI: 0.0531\n",
      "Epoch 753/1000\n",
      "10/10 - 0s - loss: -5.4409e-02 - MI: 0.0540\n",
      "Epoch 754/1000\n",
      "10/10 - 0s - loss: -5.7020e-02 - MI: 0.0578\n",
      "Epoch 755/1000\n",
      "10/10 - 0s - loss: -5.7414e-02 - MI: 0.0572\n",
      "Epoch 756/1000\n",
      "10/10 - 0s - loss: -6.2651e-02 - MI: 0.0618\n",
      "Epoch 757/1000\n",
      "10/10 - 0s - loss: -5.1513e-02 - MI: 0.0516\n",
      "Epoch 758/1000\n",
      "10/10 - 0s - loss: -5.1436e-02 - MI: 0.0506\n",
      "Epoch 759/1000\n",
      "10/10 - 0s - loss: -5.6480e-02 - MI: 0.0567\n",
      "Epoch 760/1000\n",
      "10/10 - 0s - loss: -5.7131e-02 - MI: 0.0567\n",
      "Epoch 761/1000\n",
      "10/10 - 0s - loss: -6.2255e-02 - MI: 0.0624\n",
      "Epoch 762/1000\n",
      "10/10 - 0s - loss: -5.7528e-02 - MI: 0.0579\n",
      "Epoch 763/1000\n",
      "10/10 - 0s - loss: -5.7713e-02 - MI: 0.0569\n",
      "Epoch 764/1000\n",
      "10/10 - 0s - loss: -5.3049e-02 - MI: 0.0534\n",
      "Epoch 765/1000\n",
      "10/10 - 0s - loss: -5.5292e-02 - MI: 0.0560\n",
      "Epoch 766/1000\n",
      "10/10 - 0s - loss: -5.5492e-02 - MI: 0.0550\n",
      "Epoch 767/1000\n",
      "10/10 - 0s - loss: -5.5830e-02 - MI: 0.0561\n",
      "Epoch 768/1000\n",
      "10/10 - 0s - loss: -6.3668e-02 - MI: 0.0634\n",
      "Epoch 769/1000\n",
      "10/10 - 0s - loss: -6.5666e-02 - MI: 0.0663\n",
      "Epoch 770/1000\n",
      "10/10 - 0s - loss: -6.3991e-02 - MI: 0.0639\n",
      "Epoch 771/1000\n",
      "10/10 - 0s - loss: -6.2075e-02 - MI: 0.0619\n",
      "Epoch 772/1000\n",
      "10/10 - 0s - loss: -6.5139e-02 - MI: 0.0655\n",
      "Epoch 773/1000\n",
      "10/10 - 0s - loss: -6.4606e-02 - MI: 0.0644\n",
      "Epoch 774/1000\n",
      "10/10 - 0s - loss: -7.0958e-02 - MI: 0.0717\n",
      "Epoch 775/1000\n",
      "10/10 - 0s - loss: -7.4392e-02 - MI: 0.0743\n",
      "Epoch 776/1000\n",
      "10/10 - 0s - loss: -6.6159e-02 - MI: 0.0658\n",
      "Epoch 777/1000\n",
      "10/10 - 0s - loss: -6.4662e-02 - MI: 0.0653\n",
      "Epoch 778/1000\n",
      "10/10 - 0s - loss: -6.5253e-02 - MI: 0.0656\n",
      "Epoch 779/1000\n",
      "10/10 - 0s - loss: -6.1366e-02 - MI: 0.0609\n",
      "Epoch 780/1000\n",
      "10/10 - 0s - loss: -6.5674e-02 - MI: 0.0652\n",
      "Epoch 781/1000\n",
      "10/10 - 0s - loss: -6.4733e-02 - MI: 0.0652\n",
      "Epoch 782/1000\n",
      "10/10 - 0s - loss: -6.8835e-02 - MI: 0.0693\n",
      "Epoch 783/1000\n",
      "10/10 - 0s - loss: -7.2795e-02 - MI: 0.0731\n",
      "Epoch 784/1000\n",
      "10/10 - 0s - loss: -7.2202e-02 - MI: 0.0724\n",
      "Epoch 785/1000\n",
      "10/10 - 0s - loss: -5.9185e-02 - MI: 0.0577\n",
      "Epoch 786/1000\n",
      "10/10 - 0s - loss: -6.7613e-02 - MI: 0.0672\n",
      "Epoch 787/1000\n",
      "10/10 - 0s - loss: -7.1518e-02 - MI: 0.0718\n",
      "Epoch 788/1000\n",
      "10/10 - 0s - loss: -6.9682e-02 - MI: 0.0697\n",
      "Epoch 789/1000\n",
      "10/10 - 0s - loss: -6.9578e-02 - MI: 0.0693\n",
      "Epoch 790/1000\n",
      "10/10 - 0s - loss: -7.5901e-02 - MI: 0.0758\n",
      "Epoch 791/1000\n",
      "10/10 - 0s - loss: -7.3004e-02 - MI: 0.0729\n",
      "Epoch 792/1000\n",
      "10/10 - 0s - loss: -7.1701e-02 - MI: 0.0712\n",
      "Epoch 793/1000\n",
      "10/10 - 0s - loss: -7.8118e-02 - MI: 0.0788\n",
      "Epoch 794/1000\n",
      "10/10 - 0s - loss: -7.9069e-02 - MI: 0.0793\n",
      "Epoch 795/1000\n",
      "10/10 - 0s - loss: -7.9349e-02 - MI: 0.0794\n",
      "Epoch 796/1000\n",
      "10/10 - 0s - loss: -8.2488e-02 - MI: 0.0824\n",
      "Epoch 797/1000\n",
      "10/10 - 0s - loss: -7.3745e-02 - MI: 0.0737\n",
      "Epoch 798/1000\n",
      "10/10 - 0s - loss: -7.8768e-02 - MI: 0.0791\n",
      "Epoch 799/1000\n",
      "10/10 - 0s - loss: -8.4492e-02 - MI: 0.0843\n",
      "Epoch 800/1000\n",
      "10/10 - 0s - loss: -7.8972e-02 - MI: 0.0788\n",
      "Epoch 801/1000\n",
      "10/10 - 0s - loss: -7.3585e-02 - MI: 0.0738\n",
      "Epoch 802/1000\n",
      "10/10 - 0s - loss: -8.8608e-02 - MI: 0.0885\n",
      "Epoch 803/1000\n",
      "10/10 - 0s - loss: -8.1419e-02 - MI: 0.0818\n",
      "Epoch 804/1000\n",
      "10/10 - 0s - loss: -8.4144e-02 - MI: 0.0847\n",
      "Epoch 805/1000\n",
      "10/10 - 0s - loss: -8.6660e-02 - MI: 0.0864\n",
      "Epoch 806/1000\n",
      "10/10 - 0s - loss: -8.5850e-02 - MI: 0.0854\n",
      "Epoch 807/1000\n",
      "10/10 - 0s - loss: -8.4857e-02 - MI: 0.0848\n",
      "Epoch 808/1000\n",
      "10/10 - 0s - loss: -8.8571e-02 - MI: 0.0880\n",
      "Epoch 809/1000\n",
      "10/10 - 0s - loss: -8.3985e-02 - MI: 0.0837\n",
      "Epoch 810/1000\n",
      "10/10 - 0s - loss: -9.0746e-02 - MI: 0.0911\n",
      "Epoch 811/1000\n",
      "10/10 - 0s - loss: -8.2837e-02 - MI: 0.0833\n",
      "Epoch 812/1000\n",
      "10/10 - 0s - loss: -8.8805e-02 - MI: 0.0890\n",
      "Epoch 813/1000\n",
      "10/10 - 0s - loss: -9.3196e-02 - MI: 0.0925\n",
      "Epoch 814/1000\n",
      "10/10 - 0s - loss: -8.4802e-02 - MI: 0.0846\n",
      "Epoch 815/1000\n",
      "10/10 - 0s - loss: -8.5130e-02 - MI: 0.0847\n",
      "Epoch 816/1000\n",
      "10/10 - 0s - loss: -8.8125e-02 - MI: 0.0884\n",
      "Epoch 817/1000\n",
      "10/10 - 0s - loss: -9.0837e-02 - MI: 0.0908\n",
      "Epoch 818/1000\n",
      "10/10 - 0s - loss: -9.3349e-02 - MI: 0.0938\n",
      "Epoch 819/1000\n",
      "10/10 - 0s - loss: -9.3469e-02 - MI: 0.0936\n",
      "Epoch 820/1000\n",
      "10/10 - 0s - loss: -8.3931e-02 - MI: 0.0837\n",
      "Epoch 821/1000\n",
      "10/10 - 0s - loss: -9.0762e-02 - MI: 0.0912\n",
      "Epoch 822/1000\n",
      "10/10 - 0s - loss: -8.9117e-02 - MI: 0.0892\n",
      "Epoch 823/1000\n",
      "10/10 - 0s - loss: -9.0678e-02 - MI: 0.0907\n",
      "Epoch 824/1000\n",
      "10/10 - 0s - loss: -9.6158e-02 - MI: 0.0967\n",
      "Epoch 825/1000\n",
      "10/10 - 0s - loss: -1.0172e-01 - MI: 0.1023\n",
      "Epoch 826/1000\n",
      "10/10 - 0s - loss: -9.9507e-02 - MI: 0.0996\n",
      "Epoch 827/1000\n",
      "10/10 - 0s - loss: -8.6606e-02 - MI: 0.0864\n",
      "Epoch 828/1000\n",
      "10/10 - 0s - loss: -9.0023e-02 - MI: 0.0901\n",
      "Epoch 829/1000\n",
      "10/10 - 0s - loss: -9.4479e-02 - MI: 0.0943\n",
      "Epoch 830/1000\n",
      "10/10 - 0s - loss: -9.7150e-02 - MI: 0.0966\n",
      "Epoch 831/1000\n",
      "10/10 - 0s - loss: -1.0357e-01 - MI: 0.1038\n",
      "Epoch 832/1000\n",
      "10/10 - 0s - loss: -8.9326e-02 - MI: 0.0898\n",
      "Epoch 833/1000\n",
      "10/10 - 0s - loss: -9.8136e-02 - MI: 0.0988\n",
      "Epoch 834/1000\n",
      "10/10 - 0s - loss: -9.5222e-02 - MI: 0.0936\n",
      "Epoch 835/1000\n",
      "10/10 - 0s - loss: -9.8580e-02 - MI: 0.0986\n",
      "Epoch 836/1000\n",
      "10/10 - 0s - loss: -8.8708e-02 - MI: 0.0881\n",
      "Epoch 837/1000\n",
      "10/10 - 0s - loss: -9.9862e-02 - MI: 0.1007\n",
      "Epoch 838/1000\n",
      "10/10 - 0s - loss: -9.9260e-02 - MI: 0.0993\n",
      "Epoch 839/1000\n",
      "10/10 - 0s - loss: -1.0090e-01 - MI: 0.1010\n",
      "Epoch 840/1000\n",
      "10/10 - 0s - loss: -9.8347e-02 - MI: 0.0980\n",
      "Epoch 841/1000\n",
      "10/10 - 0s - loss: -9.3761e-02 - MI: 0.0932\n",
      "Epoch 842/1000\n",
      "10/10 - 0s - loss: -9.3964e-02 - MI: 0.0938\n",
      "Epoch 843/1000\n",
      "10/10 - 0s - loss: -9.9974e-02 - MI: 0.1001\n",
      "Epoch 844/1000\n",
      "10/10 - 0s - loss: -1.0889e-01 - MI: 0.1084\n",
      "Epoch 845/1000\n",
      "10/10 - 0s - loss: -1.0185e-01 - MI: 0.1025\n",
      "Epoch 846/1000\n",
      "10/10 - 0s - loss: -1.0462e-01 - MI: 0.1056\n",
      "Epoch 847/1000\n",
      "10/10 - 0s - loss: -1.0850e-01 - MI: 0.1080\n",
      "Epoch 848/1000\n",
      "10/10 - 0s - loss: -1.0140e-01 - MI: 0.1011\n",
      "Epoch 849/1000\n",
      "10/10 - 0s - loss: -1.0289e-01 - MI: 0.1033\n",
      "Epoch 850/1000\n",
      "10/10 - 0s - loss: -9.7250e-02 - MI: 0.0979\n",
      "Epoch 851/1000\n",
      "10/10 - 0s - loss: -9.8763e-02 - MI: 0.0990\n",
      "Epoch 852/1000\n",
      "10/10 - 0s - loss: -1.0553e-01 - MI: 0.1064\n",
      "Epoch 853/1000\n",
      "10/10 - 0s - loss: -1.0496e-01 - MI: 0.1049\n",
      "Epoch 854/1000\n",
      "10/10 - 0s - loss: -1.0538e-01 - MI: 0.1057\n",
      "Epoch 855/1000\n",
      "10/10 - 0s - loss: -9.7523e-02 - MI: 0.0978\n",
      "Epoch 856/1000\n",
      "10/10 - 0s - loss: -9.7856e-02 - MI: 0.0981\n",
      "Epoch 857/1000\n",
      "10/10 - 0s - loss: -1.0939e-01 - MI: 0.1093\n",
      "Epoch 858/1000\n",
      "10/10 - 0s - loss: -1.0494e-01 - MI: 0.1054\n",
      "Epoch 859/1000\n",
      "10/10 - 0s - loss: -9.9052e-02 - MI: 0.0992\n",
      "Epoch 860/1000\n",
      "10/10 - 0s - loss: -1.0589e-01 - MI: 0.1063\n",
      "Epoch 861/1000\n",
      "10/10 - 0s - loss: -9.9135e-02 - MI: 0.1003\n",
      "Epoch 862/1000\n",
      "10/10 - 0s - loss: -9.9248e-02 - MI: 0.0993\n",
      "Epoch 863/1000\n",
      "10/10 - 0s - loss: -1.0491e-01 - MI: 0.1053\n",
      "Epoch 864/1000\n",
      "10/10 - 0s - loss: -1.0052e-01 - MI: 0.1006\n",
      "Epoch 865/1000\n",
      "10/10 - 0s - loss: -1.0008e-01 - MI: 0.1000\n",
      "Epoch 866/1000\n",
      "10/10 - 0s - loss: -1.1251e-01 - MI: 0.1123\n",
      "Epoch 867/1000\n",
      "10/10 - 0s - loss: -9.8852e-02 - MI: 0.0977\n",
      "Epoch 868/1000\n",
      "10/10 - 0s - loss: -1.0124e-01 - MI: 0.1013\n",
      "Epoch 869/1000\n",
      "10/10 - 0s - loss: -1.0633e-01 - MI: 0.1064\n",
      "Epoch 870/1000\n",
      "10/10 - 0s - loss: -1.0292e-01 - MI: 0.1033\n",
      "Epoch 871/1000\n",
      "10/10 - 0s - loss: -1.0663e-01 - MI: 0.1068\n",
      "Epoch 872/1000\n",
      "10/10 - 0s - loss: -1.0635e-01 - MI: 0.1068\n",
      "Epoch 873/1000\n",
      "10/10 - 0s - loss: -1.0092e-01 - MI: 0.1006\n",
      "Epoch 874/1000\n",
      "10/10 - 0s - loss: -9.8624e-02 - MI: 0.0992\n",
      "Epoch 875/1000\n",
      "10/10 - 0s - loss: -1.0406e-01 - MI: 0.1030\n",
      "Epoch 876/1000\n",
      "10/10 - 0s - loss: -1.0280e-01 - MI: 0.1023\n",
      "Epoch 877/1000\n",
      "10/10 - 0s - loss: -1.0471e-01 - MI: 0.1040\n",
      "Epoch 878/1000\n",
      "10/10 - 0s - loss: -1.0001e-01 - MI: 0.0992\n",
      "Epoch 879/1000\n",
      "10/10 - 0s - loss: -1.0523e-01 - MI: 0.1048\n",
      "Epoch 880/1000\n",
      "10/10 - 0s - loss: -1.0130e-01 - MI: 0.1019\n",
      "Epoch 881/1000\n",
      "10/10 - 0s - loss: -1.0955e-01 - MI: 0.1097\n",
      "Epoch 882/1000\n",
      "10/10 - 0s - loss: -1.0596e-01 - MI: 0.1059\n",
      "Epoch 883/1000\n",
      "10/10 - 0s - loss: -9.7999e-02 - MI: 0.0986\n",
      "Epoch 884/1000\n",
      "10/10 - 0s - loss: -1.0653e-01 - MI: 0.1063\n",
      "Epoch 885/1000\n",
      "10/10 - 0s - loss: -1.0425e-01 - MI: 0.1039\n",
      "Epoch 886/1000\n",
      "10/10 - 0s - loss: -9.9575e-02 - MI: 0.0993\n",
      "Epoch 887/1000\n",
      "10/10 - 0s - loss: -9.5369e-02 - MI: 0.0949\n",
      "Epoch 888/1000\n",
      "10/10 - 0s - loss: -1.0647e-01 - MI: 0.1069\n",
      "Epoch 889/1000\n",
      "10/10 - 0s - loss: -1.0255e-01 - MI: 0.1025\n",
      "Epoch 890/1000\n",
      "10/10 - 0s - loss: -1.1030e-01 - MI: 0.1104\n",
      "Epoch 891/1000\n",
      "10/10 - 0s - loss: -1.0149e-01 - MI: 0.1011\n",
      "Epoch 892/1000\n",
      "10/10 - 0s - loss: -1.0498e-01 - MI: 0.1051\n",
      "Epoch 893/1000\n",
      "10/10 - 0s - loss: -1.0742e-01 - MI: 0.1072\n",
      "Epoch 894/1000\n",
      "10/10 - 0s - loss: -1.1756e-01 - MI: 0.1176\n",
      "Epoch 895/1000\n",
      "10/10 - 0s - loss: -1.0413e-01 - MI: 0.1043\n",
      "Epoch 896/1000\n",
      "10/10 - 0s - loss: -1.1221e-01 - MI: 0.1115\n",
      "Epoch 897/1000\n",
      "10/10 - 0s - loss: -1.0667e-01 - MI: 0.1054\n",
      "Epoch 898/1000\n",
      "10/10 - 0s - loss: -1.1245e-01 - MI: 0.1123\n",
      "Epoch 899/1000\n",
      "10/10 - 0s - loss: -1.0124e-01 - MI: 0.1002\n",
      "Epoch 900/1000\n",
      "10/10 - 0s - loss: -9.9766e-02 - MI: 0.1008\n",
      "Epoch 901/1000\n",
      "10/10 - 0s - loss: -1.1531e-01 - MI: 0.1147\n",
      "Epoch 902/1000\n",
      "10/10 - 0s - loss: -1.0687e-01 - MI: 0.1069\n",
      "Epoch 903/1000\n",
      "10/10 - 0s - loss: -1.0218e-01 - MI: 0.1021\n",
      "Epoch 904/1000\n",
      "10/10 - 0s - loss: -1.1479e-01 - MI: 0.1146\n",
      "Epoch 905/1000\n",
      "10/10 - 0s - loss: -1.0823e-01 - MI: 0.1075\n",
      "Epoch 906/1000\n",
      "10/10 - 0s - loss: -1.1291e-01 - MI: 0.1130\n",
      "Epoch 907/1000\n",
      "10/10 - 0s - loss: -1.1238e-01 - MI: 0.1119\n",
      "Epoch 908/1000\n",
      "10/10 - 0s - loss: -1.0973e-01 - MI: 0.1092\n",
      "Epoch 909/1000\n",
      "10/10 - 0s - loss: -1.0384e-01 - MI: 0.1030\n",
      "Epoch 910/1000\n",
      "10/10 - 0s - loss: -1.0634e-01 - MI: 0.1064\n",
      "Epoch 911/1000\n",
      "10/10 - 0s - loss: -1.0370e-01 - MI: 0.1037\n",
      "Epoch 912/1000\n",
      "10/10 - 0s - loss: -1.0814e-01 - MI: 0.1083\n",
      "Epoch 913/1000\n",
      "10/10 - 0s - loss: -1.0341e-01 - MI: 0.1041\n",
      "Epoch 914/1000\n",
      "10/10 - 0s - loss: -1.0449e-01 - MI: 0.1036\n",
      "Epoch 915/1000\n",
      "10/10 - 0s - loss: -1.0498e-01 - MI: 0.1054\n",
      "Epoch 916/1000\n",
      "10/10 - 0s - loss: -9.4675e-02 - MI: 0.0953\n",
      "Epoch 917/1000\n",
      "10/10 - 0s - loss: -1.0933e-01 - MI: 0.1095\n",
      "Epoch 918/1000\n",
      "10/10 - 0s - loss: -1.0790e-01 - MI: 0.1074\n",
      "Epoch 919/1000\n",
      "10/10 - 0s - loss: -1.0384e-01 - MI: 0.1038\n",
      "Epoch 920/1000\n",
      "10/10 - 0s - loss: -1.0083e-01 - MI: 0.1010\n",
      "Epoch 921/1000\n",
      "10/10 - 0s - loss: -1.0741e-01 - MI: 0.1078\n",
      "Epoch 922/1000\n",
      "10/10 - 0s - loss: -1.0874e-01 - MI: 0.1082\n",
      "Epoch 923/1000\n",
      "10/10 - 0s - loss: -1.1529e-01 - MI: 0.1146\n",
      "Epoch 924/1000\n",
      "10/10 - 0s - loss: -1.0969e-01 - MI: 0.1099\n",
      "Epoch 925/1000\n",
      "10/10 - 0s - loss: -1.1100e-01 - MI: 0.1109\n",
      "Epoch 926/1000\n",
      "10/10 - 0s - loss: -9.7808e-02 - MI: 0.0969\n",
      "Epoch 927/1000\n",
      "10/10 - 0s - loss: -9.7950e-02 - MI: 0.0976\n",
      "Epoch 928/1000\n",
      "10/10 - 0s - loss: -1.0143e-01 - MI: 0.1010\n",
      "Epoch 929/1000\n",
      "10/10 - 0s - loss: -1.0341e-01 - MI: 0.1026\n",
      "Epoch 930/1000\n",
      "10/10 - 0s - loss: -1.0539e-01 - MI: 0.1053\n",
      "Epoch 931/1000\n",
      "10/10 - 0s - loss: -1.0620e-01 - MI: 0.1067\n",
      "Epoch 932/1000\n",
      "10/10 - 0s - loss: -1.0464e-01 - MI: 0.1035\n",
      "Epoch 933/1000\n",
      "10/10 - 0s - loss: -9.6313e-02 - MI: 0.0957\n",
      "Epoch 934/1000\n",
      "10/10 - 0s - loss: -1.1342e-01 - MI: 0.1134\n",
      "Epoch 935/1000\n",
      "10/10 - 0s - loss: -1.0713e-01 - MI: 0.1066\n",
      "Epoch 936/1000\n",
      "10/10 - 0s - loss: -9.5921e-02 - MI: 0.0957\n",
      "Epoch 937/1000\n",
      "10/10 - 0s - loss: -1.0149e-01 - MI: 0.1014\n",
      "Epoch 938/1000\n",
      "10/10 - 0s - loss: -1.0091e-01 - MI: 0.1009\n",
      "Epoch 939/1000\n",
      "10/10 - 0s - loss: -1.0922e-01 - MI: 0.1086\n",
      "Epoch 940/1000\n",
      "10/10 - 0s - loss: -1.0805e-01 - MI: 0.1079\n",
      "Epoch 941/1000\n",
      "10/10 - 0s - loss: -1.0271e-01 - MI: 0.1023\n",
      "Epoch 942/1000\n",
      "10/10 - 0s - loss: -1.0234e-01 - MI: 0.1025\n",
      "Epoch 943/1000\n",
      "10/10 - 0s - loss: -1.0738e-01 - MI: 0.1074\n",
      "Epoch 944/1000\n",
      "10/10 - 0s - loss: -1.0456e-01 - MI: 0.1049\n",
      "Epoch 945/1000\n",
      "10/10 - 0s - loss: -1.0045e-01 - MI: 0.0997\n",
      "Epoch 946/1000\n",
      "10/10 - 0s - loss: -1.0823e-01 - MI: 0.1080\n",
      "Epoch 947/1000\n",
      "10/10 - 0s - loss: -1.1791e-01 - MI: 0.1182\n",
      "Epoch 948/1000\n",
      "10/10 - 0s - loss: -1.0094e-01 - MI: 0.0999\n",
      "Epoch 949/1000\n",
      "10/10 - 0s - loss: -1.0173e-01 - MI: 0.1020\n",
      "Epoch 950/1000\n",
      "10/10 - 0s - loss: -1.0947e-01 - MI: 0.1096\n",
      "Epoch 951/1000\n",
      "10/10 - 0s - loss: -1.0528e-01 - MI: 0.1065\n",
      "Epoch 952/1000\n",
      "10/10 - 0s - loss: -1.0585e-01 - MI: 0.1060\n",
      "Epoch 953/1000\n",
      "10/10 - 0s - loss: -1.0406e-01 - MI: 0.1050\n",
      "Epoch 954/1000\n",
      "10/10 - 0s - loss: -1.0615e-01 - MI: 0.1065\n",
      "Epoch 955/1000\n",
      "10/10 - 0s - loss: -1.0599e-01 - MI: 0.1068\n",
      "Epoch 956/1000\n",
      "10/10 - 0s - loss: -1.1144e-01 - MI: 0.1114\n",
      "Epoch 957/1000\n",
      "10/10 - 0s - loss: -1.0635e-01 - MI: 0.1058\n",
      "Epoch 958/1000\n",
      "10/10 - 0s - loss: -9.8317e-02 - MI: 0.0979\n",
      "Epoch 959/1000\n",
      "10/10 - 0s - loss: -1.0584e-01 - MI: 0.1056\n",
      "Epoch 960/1000\n",
      "10/10 - 0s - loss: -1.0934e-01 - MI: 0.1096\n",
      "Epoch 961/1000\n",
      "10/10 - 0s - loss: -1.0882e-01 - MI: 0.1090\n",
      "Epoch 962/1000\n",
      "10/10 - 0s - loss: -1.0769e-01 - MI: 0.1076\n",
      "Epoch 963/1000\n",
      "10/10 - 0s - loss: -1.0221e-01 - MI: 0.1022\n",
      "Epoch 964/1000\n",
      "10/10 - 0s - loss: -1.1397e-01 - MI: 0.1149\n",
      "Epoch 965/1000\n",
      "10/10 - 0s - loss: -1.0303e-01 - MI: 0.1028\n",
      "Epoch 966/1000\n",
      "10/10 - 0s - loss: -1.1537e-01 - MI: 0.1151\n",
      "Epoch 967/1000\n",
      "10/10 - 0s - loss: -1.0956e-01 - MI: 0.1099\n",
      "Epoch 968/1000\n",
      "10/10 - 0s - loss: -1.1334e-01 - MI: 0.1122\n",
      "Epoch 969/1000\n",
      "10/10 - 0s - loss: -1.0644e-01 - MI: 0.1067\n",
      "Epoch 970/1000\n",
      "10/10 - 0s - loss: -1.1657e-01 - MI: 0.1167\n",
      "Epoch 971/1000\n",
      "10/10 - 0s - loss: -1.0961e-01 - MI: 0.1099\n",
      "Epoch 972/1000\n",
      "10/10 - 0s - loss: -1.0697e-01 - MI: 0.1063\n",
      "Epoch 973/1000\n",
      "10/10 - 0s - loss: -1.0376e-01 - MI: 0.1031\n",
      "Epoch 974/1000\n",
      "10/10 - 0s - loss: -9.7911e-02 - MI: 0.0975\n",
      "Epoch 975/1000\n",
      "10/10 - 0s - loss: -1.1267e-01 - MI: 0.1122\n",
      "Epoch 976/1000\n",
      "10/10 - 0s - loss: -9.6388e-02 - MI: 0.0971\n",
      "Epoch 977/1000\n",
      "10/10 - 0s - loss: -9.9408e-02 - MI: 0.0995\n",
      "Epoch 978/1000\n",
      "10/10 - 0s - loss: -1.0013e-01 - MI: 0.1000\n",
      "Epoch 979/1000\n",
      "10/10 - 0s - loss: -1.0137e-01 - MI: 0.1011\n",
      "Epoch 980/1000\n",
      "10/10 - 0s - loss: -1.0638e-01 - MI: 0.1076\n",
      "Epoch 981/1000\n",
      "10/10 - 0s - loss: -1.0559e-01 - MI: 0.1055\n",
      "Epoch 982/1000\n",
      "10/10 - 0s - loss: -1.0130e-01 - MI: 0.1006\n",
      "Epoch 983/1000\n",
      "10/10 - 0s - loss: -1.0760e-01 - MI: 0.1079\n",
      "Epoch 984/1000\n",
      "10/10 - 0s - loss: -1.1263e-01 - MI: 0.1132\n",
      "Epoch 985/1000\n",
      "10/10 - 0s - loss: -1.0132e-01 - MI: 0.1016\n",
      "Epoch 986/1000\n",
      "10/10 - 0s - loss: -1.0709e-01 - MI: 0.1072\n",
      "Epoch 987/1000\n",
      "10/10 - 0s - loss: -1.0207e-01 - MI: 0.1019\n",
      "Epoch 988/1000\n",
      "10/10 - 0s - loss: -9.8803e-02 - MI: 0.0983\n",
      "Epoch 989/1000\n",
      "10/10 - 0s - loss: -1.1015e-01 - MI: 0.1104\n",
      "Epoch 990/1000\n",
      "10/10 - 0s - loss: -1.0602e-01 - MI: 0.1050\n",
      "Epoch 991/1000\n",
      "10/10 - 0s - loss: -1.1185e-01 - MI: 0.1115\n",
      "Epoch 992/1000\n",
      "10/10 - 0s - loss: -1.0261e-01 - MI: 0.1018\n",
      "Epoch 993/1000\n",
      "10/10 - 0s - loss: -1.1572e-01 - MI: 0.1156\n",
      "Epoch 994/1000\n",
      "10/10 - 0s - loss: -1.0756e-01 - MI: 0.1077\n",
      "Epoch 995/1000\n",
      "10/10 - 0s - loss: -9.4908e-02 - MI: 0.0946\n",
      "Epoch 996/1000\n",
      "10/10 - 0s - loss: -1.0923e-01 - MI: 0.1091\n",
      "Epoch 997/1000\n",
      "10/10 - 0s - loss: -1.1329e-01 - MI: 0.1136\n",
      "Epoch 998/1000\n",
      "10/10 - 0s - loss: -1.0626e-01 - MI: 0.1058\n",
      "Epoch 999/1000\n",
      "10/10 - 0s - loss: -1.0438e-01 - MI: 0.1047\n",
      "Epoch 1000/1000\n",
      "10/10 - 0s - loss: -9.8916e-02 - MI: 0.0981\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x7ff9bc10ff10>]"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "\n",
    "# GA numerical bias\n",
    "from ifn.Architectures.utils import build_gIFN_DNN\n",
    "from ifn.Architectures.ifn import IFN, mine_loss, MI\n",
    "from ifn.Architectures.dnn import DNN\n",
    "from ifn.Architectures.ifn import IFN, gIFN\n",
    "\n",
    "\n",
    "x_dim = 1\n",
    "y_dim = 1\n",
    "A_layers = [16,16,16]\n",
    "B_layers = [1,]\n",
    "C_layers = [1,]\n",
    "D_layers = [1,]\n",
    "l2_reg = 1e-9\n",
    "d_multiplier = 0.0\n",
    "d_l1_reg = 1e-3\n",
    "\n",
    "\n",
    "X, Y = np.expand_dims(gauss_reco_prior,1), np.expand_dims(gauss_truth_prior,1)\n",
    "\n",
    "# # Build model\n",
    "# ifn = build_gIFN_DNN(1, 1, [16,16,16], l2_reg = 1e-9, acts = 'relu', d_l1_reg= 1e-3, d_multiplier=0)\n",
    "\n",
    "model_A = DNN(x_dim, A_layers, 1, l2_regs = l2_reg, acts = \"relu\")\n",
    "model_B = DNN(x_dim, B_layers, y_dim, l2_regs = l2_reg, acts = \"linear\")\n",
    "model_C = DNN([x_dim, y_dim], C_layers, [y_dim, y_dim], symmetrize=True, l2_regs = l2_reg, acts = \"relu\")\n",
    "model_D = DNN(x_dim, D_layers, y_dim, l2_regs = l2_reg, acts = \"relu\")\n",
    "\n",
    "ifn = gIFN(model_A, model_B, model_C, model_D, d_multiplier= d_multiplier, d_l1_reg = d_l1_reg, y_dim = y_dim)\n",
    "\n",
    "\n",
    "ifn.pre_train([X, Y], epochs = 10, batch_size = 64, verbose = True)\n",
    "opt = tf.keras.optimizers.Adam(lr = 1e-3)\n",
    "ifn.compile(loss=mine_loss, optimizer=opt, metrics = [MI,])\n",
    "\n",
    "\n",
    "# Fit\n",
    "history = ifn.fit([X, Y],\n",
    "            batch_size= 1024,\n",
    "            epochs = 1000,\n",
    "            shuffle=True, verbose= 2)\n",
    "\n",
    "plt.plot(history.history[\"MI\"][5:])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.8944271909999159\n"
     ]
    }
   ],
   "source": [
    "def analytic_bias(x):\n",
    "    return (sigma**2/(sigma**2+epsilon**2))*x+(epsilon**2/(sigma**2+epsilon**2))*mu\n",
    "\n",
    "\n",
    "def analytic_bayesian_resolution(x):\n",
    "    temp = epsilon ** 2 * sigma **2 / ((epsilon**2 + sigma**2 ))\n",
    "    return np.sqrt(temp)\n",
    "\n",
    "\n",
    "print(analytic_bayesian_resolution(0))\n",
    "\n",
    "def analytic_resolution(z):\n",
    "    return np.sqrt((epsilon**2 *  sigma**4) / ((epsilon**2 + sigma**2 ))**2)\n",
    "\n",
    "def analytic_MSE(x):\n",
    "    return (epsilon**2 * mu + sigma**2 * x) / ((epsilon**2 + sigma**2 ))\n",
    "\n",
    "\n",
    "\n",
    "meanvals_analytic = analytic_bias(np.linspace(-5,5,20))\n",
    "resvals_analytic = analytic_bayesian_resolution(np.linspace(-5,5,20))    \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Estimates\n",
    "xrange = np.linspace(-5,5,20)\n",
    "window = 0.25\n",
    "MSE_estimates = np.array([mypoly(np.random.normal(xx,2,1000)) for xx in np.linspace(-5,5,20)])\n",
    "MSE_means = np.mean(MSE_estimates, axis = 1)\n",
    "MSE_stds = np.std(MSE_estimates, axis = 1)\n",
    "# meanvals = np.array([np.mean(mypoly(np.random.normal(xx,2,10000))) for xx in np.linspace(-5,5,20)])\n",
    "\n",
    "sorted_xs = np.sort(gauss_reco_prior)\n",
    "sorted_indices = np.argsort(gauss_reco_prior)\n",
    "n = 0\n",
    "stds = []\n",
    "for x in xrange:\n",
    "    m = mypoly(x)\n",
    "    temp_array = []\n",
    "    for (index, i) in enumerate(sorted_xs[n:]):\n",
    "        temp_array.append(gauss_truth_prior[sorted_indices[index]])\n",
    "        if i > x + window:\n",
    "            n = index + n\n",
    "            break\n",
    "    std = np.std(temp_array)\n",
    "    stds.append(std)\n",
    "    \n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Plots"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "GA_meanvals =  np.squeeze(ifn.maximum_likelihood(np.expand_dims(np.linspace(-5,5,20), 1)))\n",
    "GA_stds =  np.squeeze(ifn.uncertainty(np.expand_dims(np.linspace(-5,5,20), 1)))\n",
    "\n",
    "\n",
    "fig = plt.figure(figsize=(6, 6))\n",
    "gs = gridspec.GridSpec(1, 1) \n",
    "plt.subplots_adjust(hspace = .001)\n",
    "\n",
    "ax0 = plt.subplot(gs[0])\n",
    "ax0.yaxis.set_ticks_position('both')\n",
    "ax0.xaxis.set_ticks_position('both')\n",
    "ax0.tick_params(direction=\"in\",which=\"both\")\n",
    "plt.xticks(fontsize=20)\n",
    "plt.yticks(fontsize=20)\n",
    "ax0.minorticks_on()\n",
    "\n",
    "\n",
    "# # True z_T sample\n",
    "# plt.scatter(gauss_reco_prior[:2500],gauss_truth_prior[:2500],label=\"True $Z_T$\",alpha=0.05, color = data_color)\n",
    "\n",
    "# MSE\n",
    "plt.errorbar(xrange,MSE_means, yerr = stds,marker='o',ls=\"\",label=\"Fit MSE\", color = MSE_darkcolor, capsize = capsize)\n",
    "plt.fill_between(xrange, meanvals_analytic+resvals_analytic, meanvals_analytic-resvals_analytic, color = MSE_color, alpha = 0.25)\n",
    "plt.plot(xrange,meanvals_analytic,ls=\"--\",label=\"Analytic MSE\")\n",
    "\n",
    "# MLC\n",
    "plt.plot(xrange,np.linspace(-5,5,20),ls=\"--\",label=\"Analytic MLC\", color = MLC_color)\n",
    "plt.fill_between(xrange, np.linspace(-5,5,20) + epsilon, np.linspace(-5,5,20) -epsilon, color = MLC_color, alpha = 0.25)\n",
    "plt.errorbar(xrange + 0.025,GA_meanvals, yerr = GA_stds,marker='o',ls=\"\",label=\"Gaussian Ansatz\", color = MLC_darkcolor, capsize = capsize)\n",
    "\n",
    "\n",
    "plt.xlim([-5,5])\n",
    "plt.ylim([-5,5])\n",
    "plt.xlabel(\"Measured $X_D$\",fontsize=20)\n",
    "plt.ylabel(\"Calibrated $\\hat{Z}_T$\",fontsize=20)\n",
    "leg = plt.legend(frameon=False,fontsize=14,ncol=1, loc = \"lower right\")\n",
    "for lh in leg.legendHandles: \n",
    "    lh.set_alpha(1)\n",
    "_=plt.title(\"Simulation-Based Gaussian Example\",loc=\"right\",fontsize=18, fontstyle=\"italic\")\n",
    "_=plt.text(-4.5,4,r\"$(\\mu,\\sigma,\\epsilon)=(0,1,2)$\",fontsize=20)\n",
    "plt.savefig(\"Gaussian_prior2.pdf\",bbox_inches='tight')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Multiple Fits"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib as mpl\n",
    "\n",
    "def b(x,sigma,epsilon,mu):\n",
    "    return (sigma**2/(sigma**2+epsilon**2))*x+(epsilon**2/(sigma**2+epsilon**2))*mu\n",
    "\n",
    "\n",
    "def analytic_resolution(z,sigma, epsilon, mu):\n",
    "    return np.sqrt((epsilon**2 *  sigma**4) / ((epsilon**2 + sigma**2 ))**2)\n",
    "\n",
    "fig = plt.figure(figsize=(6, 6))\n",
    "gs = gridspec.GridSpec(1, 1) \n",
    "plt.subplots_adjust(hspace = .001)\n",
    "\n",
    "ax0 = plt.subplot(gs[0])\n",
    "ax0.yaxis.set_ticks_position('both')\n",
    "ax0.xaxis.set_ticks_position('both')\n",
    "ax0.tick_params(direction=\"in\",which=\"both\")\n",
    "plt.xticks(fontsize=20)\n",
    "plt.yticks(fontsize=20)\n",
    "ax0.minorticks_on()\n",
    "\n",
    "# colors\n",
    "n_lines = 5\n",
    "c = np.arange(-3, n_lines + 1)\n",
    "\n",
    "norm = mpl.colors.Normalize(vmin=c.min(), vmax=c.max())\n",
    "cmap = mpl.cm.ScalarMappable(norm=norm, cmap=\"Blues\")\n",
    "cmap.set_array([])\n",
    "\n",
    "plt.plot(np.linspace(-5,5,20),np.linspace(-5,5,20),ls=\"--\",label=\"Unbiased\",color=\"red\")\n",
    "markers = [\"v\",\"o\",\"s\",\"d\",\"^\"]\n",
    "myc=-1\n",
    "for (i,epsilon) in enumerate(np.linspace(-2,2,5)):\n",
    "    myc+=1\n",
    "    gauss_truth_prior_B = np.random.normal(epsilon,1,10000)\n",
    "    gauss_reco_prior_B = np.random.normal(gauss_truth_prior_B,2,10000)\n",
    "    mycalibration_B = np.polyfit(gauss_reco_prior_B, gauss_truth_prior_B, 1)\n",
    "    mypoly_B = np.poly1d(mycalibration_B)\n",
    "\n",
    "    meanvals_B = np.array([np.mean(mypoly_B(np.random.normal(xx,2,10000))) for xx in np.linspace(-5,5,20)])\n",
    "    meanvals_analytic_B = b(np.linspace(-5,5,20),1,2,epsilon)\n",
    "    resvals_analytic_B = analytic_resolution(np.linspace(-5,5,20),1,2,epsilon)\n",
    "\n",
    "    plt.errorbar(np.linspace(-5,5,20) + epsilon / 20,meanvals_B, yerr = MSE_stds, marker=markers[myc],ls=\"\",color=cmap.to_rgba(i + 1),label=\"$\\mu=%0.1f$\" % epsilon)\n",
    "    plt.plot(np.linspace(-5,5,20),meanvals_analytic_B,ls=\"--\",color=cmap.to_rgba(i + 1))\n",
    "    plt.fill_between(np.linspace(-5, 5, 20), meanvals_analytic_B + resvals_analytic_B, meanvals_analytic_B - resvals_analytic_B, color = cmap.to_rgba(i + 1), alpha = 0.25)\n",
    "\n",
    "plt.xlim([-5,5])\n",
    "plt.ylim([-5,5])\n",
    "plt.xlabel(\"True $Z_T$\",fontsize=20)\n",
    "plt.ylabel(\"Calibrated $\\hat{Z}_{T}$\",fontsize=20)\n",
    "plt.legend(frameon=False,fontsize=15,ncol=2,loc=\"lower right\",handlelength=1)\n",
    "_=plt.title(\"Simulation-Based Gaussian Example\",loc=\"right\",fontsize=20, fontstyle=\"italic\")\n",
    "_=plt.text(-4,4,r\"$(\\sigma,\\epsilon)=(1,2)$\",fontsize=20)\n",
    "plt.savefig(\"Gaussian_prior4.pdf\",bbox_inches='tight')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZ4AAAGZCAYAAAC5eVe3AAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjUuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/YYfK9AAAACXBIWXMAAAsTAAALEwEAmpwYAADNEElEQVR4nOyddZwkxfn/39U9Prs763Lu3B0ccIK7BgkhhCTEIUaIQeQb/cbzjf1ihBASQoR4iJAEIgR3P+zc92Tdd8db6vdH9ezOzs6s3ezuHfT79ZrX7HZXV1fXdNenq+qp5xFSSlxcXFxcXKYLbaYL4OLi4uLyysIVHhcXFxeXacUVHhcXFxeXacUVHhcXFxeXacUVHhcXFxeXacUVHhcXFxeXaeUVITxCiDuEEFIIMWumy5JBCFHplOnuKT7Px5zzvGUqz+MyHCHELiFE10yX41BFKHqFENtnuiwuhRFCrHbaj58WM9/DUniEEGcKIX4vhGgUQiSFEP1CiB1CiL8LIT4qhMi9rnVAi5SyeSbKW4A1zvdzB5OJEOIy58b4bIEk65zvZw/mPMVECDHLKXPuJy2E2COEuEkIUTPT5ZwsQohyYBHw/ASP8wkh3uG8KDUJIVJCiJhzb/9OCHG5EEKfkkJPP0uBCIfWffnZAvdl9ue+mS7nNLPW+T6odioXTzEzm2qch+7HwHuBKPBvYC9KQFcCFwDrpJTfzzn0bCAxjUUdDxnhWX+Q+WRujEL5fBr4CrDjIM9TTLLF8F/O3xpQC1wKvB91XSdMf9GKwoR/WyHEauAPwBFAF3APsA/1jC4FLgHeAtwLnFfMws4QB4AVqGs9VMjclz8DmgqkeXqaynKoUKx2ahiHlfAA/4sSnTuAd0gp+7J3CiEqGaqoQaSUW6eneBOiWG8So+Yjpdx3kPlPBZkH/FdSyhuzdwghvgE0AscLIWqllO3TXbgiMKHfVghxMuolKgh8Evi+lNLMSVMCXAfMK2I5ZwwpZRw41J7LdYAE/ie3bXkFsxYwgZeKmquU8rD4AALoRN0Y9RM47svOMRdnbfMBadQbdx1wE+oNLA48CZzopKsAvg3scfY9C5yR5xz/dc5RmWffO5x91+Vs3wH0AiJrmwZcA9wO7HLO2QE8Alyac/w1Tr75Pp930pzl/P/tPOVaCPwI2A0kgXbgV8CcPGm/5uTzKtSb971O2buBvwANE/wt/+3kd1yefUHAAFKAN2ff0c7v8QzQhurF7gRuAGrz5BUAPgA8AbQ617kP+Gf2/ZCV/nLgP859lnZ+o88CeoHreDXwIKr33QXcApShei4SWDqOuqhxzmcCrxlH+tw6Gfc946R/s1O2j+XL26mj53O268BbnWs94Pw2Tc598I48v9FvgG1OWbqAF4AbgWBWuocAGyjNOf4twO+A7cAA0OP83u8s8Psazu9bDXzHqYMEStSunMA9WevUy7ZxpteBzc4xa3P2+VA9Vgm8P2t7CfAR1P3f6JSzBdV+nJ7nHJnn9zuoF+q/o565PuBvOO0garTn96hnuM+5v+flyS/TTh2NGjna4/yW7cAPgUCea4wDLxZ4Tj8GPAX0O9fyNPC6cdXfRBqMmfygxoMzDWvVBI670zmmIWvbWmfbY6gG6SHnx73L2d4FHIlqpJ4Avuf86NL54StyztEJ7C5w/h86x52Wta3Meejuz0l7pPMgPQDcDHwD+K3zAErgiqy05wPXO9s3AV/K+qxy0nzc2f/mnPO8BtVYxpwb9luoBleiGpbanPQZofgjqiH4HfD/UA2KBP4zwd+yDdXA+fLsy7wo3JRn35+d3+SPKAG6HjUEIIENgCcrbRg1x2KjGszvO2W+zfl9r815wDJisQP1UF6PajwlqmdWqJwHUAJ+g/P3Q6hGs4+sl4pR6uI3FHg5GGddjvuecdJ/x9mer6HLPBc/z9qmZd0bzzjX+U3nHAeA72WlfRtgOb/vL4GvO2V6AujMSiec+tmWc/5SlAA/AfzcuZafM/TC+amc9Mc72x9ENeBPOL/zL537y8Z5iRxHPV7k5PXbCdT9Jc4x/82prz872z+bk/5iVEN+F+pl95uoF7eUc90n56TPPL93OMf9FXXfb3C2/9spQ9RJk3kpk8CDecrbgXrm96GE7wdOOdrJ88wBq5ztv8jZXpdVhkdR7eNPUW2jJM9LwoiyTOZmn6kPsN+5sOeAd5Pn7TzPMU0ow4Lsbe918rEY2Sjf5+zrAz6Ys++Pzr5XZW2b72z7c4HzP0HOmx1wpnPMd3LSVpEjajk3wF052893tn+9wLl/5+w/ImvbCc5DuQ1YmJP+q076b+Vsb2FIqOuytgec+k2R1eiP8XvMc/JqZ0gov+w8ABucB2zE25dz7JICef7FyfPErG2Zh/b9edL7gLKs/2900n6D4eLlda5ZAiuztr+VoQYhnPP7ZR7iEQ9+nnLUokQjzgRepg7ynnkg937M2vc+55gPZW27PN894ezTcHr5Tl11oUQ3kidtddbfS508/5CTpoQ8oxlAA0pIt+Zsz/T6bUb2vD5MVu9/HPX4BSf9/Qx/icv+HJnnuAed485w/r+ZPM921nWE82x/tXPMTwo8v63A0VnbS1EvgCbq2Twp597e59RJIGt75rmTTr65+2LOvVietf2qPPeDH/WylyKndwPMRYlPK2O8dE34Rp/JD3AyamhFZn32Aj8BVudJn+k+/ytn+4+d7d/Mc8z3nX0/y7Pvf519l2dte52z7dN50ntQjcr2nO0fI09PZJTr1px8ns7Z/mknn9cXOG4L6oHNHs571rkpj8qTfqGT3yM5D4tEdadHDKmh3nhs8vReCpQpU1+FPrcDCyZ4X3zGOfairG2/JKenWeDYE5zy/73A/qvJeotDNY5tKMHN13jf5KT/7jjK/UEn7T8K7P80Ixu/+ZO9Z1A9jV5yGvCs/bc45Tk5a9sXnW1vH+N88510946jbFc4aT8xgd94G9Cesy3TyOcbSj7X2fe1ceZ/xxj3pSSr8c867jjn/nkc1cOTZPUYx3nuzDP2p5ztW5ztF+Q55nln39vy7MsM81VlbXuts21rvmfVee4kcHzWthucbdnC9kln20cKXMvvnf0LC12vlPLwMi6QUj4uhFgGnIqabzgD1XC8D7haCHGNlDLb3rzQJO8aVOX8OM9p5jvfN42yb09OXpDfLPRI1Fjo8znb85ZLCDEbuBbVk1mEerMRWUkO5ORT0CRbCBEGlgGPy0yrI8Qa59z/lFJuzFPejIVRtslu5hx/l1K25DlmEXBASpl2znGyU/5sWqSUNzt/ZwwL3iul/JlzjADqgfegLPDWCSGWSSmTOdf0euDtzjXUoN7ussmun8dRb2x3CiFuQQ2VPimltHKO+TCqjuNCiC/lub6jnO+Mif4VqBeaz0gpB/Kkz9TheAwLjnS+n8jdIYSoRfXAcrkpJ91E7pnFqCHrfxYozzpUI/pi1rbHne9bnN/2duBhKWUq59gm1EvgOUKIh4BbUUOwrXnOk/e+FUJUAB9CDUkdgRqSzl4akfscZfLJ9xwvcr535dmXj3WoHkRp7n03GlLKZ4QQtwFvAk5CDYddnS+tEGI56vrOBBaghoOzOZCVNvP87pZS3pUnu/mo3sVtBfYNSCmzLQYzdfWdzLOaQyatN2vbWtSoUPb9kOlJ1hV4XpY536Mv1ZmIMh+KH6AS+DVDw2PZb/efdbZflrUtM2GWdxIRNfbZVWDfM6hJZ3/Wtsz4dz7DgneRf2w6X0/kHGebheq+/xA19PUl4B9OPl/MyWcn0FOgrCc7x/wga9tHnW3vK3DMEc7+P2Zt+5yz7a150teT88aOGpPPfVO8LWv/3c62daPUsQQuzNrmzaqDJobmpb7kfPpQXX9fTl7vRL0QZMrRBXyX4cNj7Ywsb77PuU7625z/VxUof+aNb8U47t2/O2nfNUa6Uue+2JuzfUL3DEM9jY/mOYcfdW9vybPvNai5K8s5fgBlcpw7Fzgf1dPsYWgI7FHgnJx0mXugMmvb0aghGomasP4x8H/OtfzK2f7LrPQe1IT2jgJ19hPnmBEjIXnSznLSbhwrbYHjMz2DJHmGGZ00b3XqN4Wa47ke9ZL1JaduJVnGEAw9vzfnyWuBs++vefaVOPX+cM72fzrHzC5Qvkwvaa7zv4aaO9qUlWYl43tWTLLayLznm0xFH2of1Nhi5qKzu5eZsf8FWduOcrb9Pk8+Vc6+u/Ps8zg31gs521uBPQXKlTn/+Vnbws4D/HDWNj+qAewk/xBYRlizLfMi5DFQyNr/IeeYq7K2fdfZdkmBY97v7P9A1rZMF3x5nvQXk0cQx/itulBjySPmcJz9/yJH6BgSzBvJmUtCvdlK4JlRzlmH6k1tz+TjbA84/z80gfI/6RxTlmefHzWBGwW0ceSVEakPj5HuTCfdXw7ynvmms+2MPOkz8wwjnousNOWoN/uMmP+zQDod1XDe7NyjSbIaPKeOGnOOedFJd2ae/L7inO+DWduOJuclKeeYZ8hjGVkg7WucvH433vsg69jMcHdmHvRTedI04IgkTsOes/9h59gjs7Zlnt+r86TPzLt9Ns++05x91+dsbwbSo/yucbJeOlBrrCTw66xtFzjbvjzResr9HJaeC/LQ4Hx3O58Ma4BuKWVjzjbIPzS2dpR9K1EP++DwgLNCvQ7V8xiGEGIRqmGG4UMKq1FvE9nbjkcNHf1F5gyBCSHmA5c5/2Yv4joGNaTyfJ6yQv7hjMwi2hFeAYQQAVQDH2d493016g13W55zjLV4Nfcci1A91G0yz3CGEMKDui5yzvdq5/v/ZM76FqfMo5ZBStkm1bDehc6mlZlTOt/V4ym/Q2aYojbPvqudvF6QUtrjyOsZ53usBaHHO99P52yb6D2z3PnemZNeoIZQYJQhQillr5Tyj6ihboOhesxNZ0kpH5dSvg81ke1HDfMhhJiHqqPs52guSkgelFI+mFO2clTPNfdaVjvfI55VIYQXZVyxQUppFLqeLDLDvy+Ommrked6LGg69D3XfdgOfEUJU5SQ9F/WS83Mp5f6cPI5D1WeM4euaJttOZeolu37rUW2k11kPlsu1qCmB7DV1+aYDJvO85OWwEB4hxNuEEOc6D0juvkUoUz5QKi+d7eWoyfLchnm0lbijNaT5GnID9QawRAgxON8ghFiAsrMPoOY/Osc4f6YRPiL7hM5D+jdU97lJDh8vz9zc/XnKmjlPErXWIMOjzvc7sl2vOOPJv0NZG31GOmPDzpj7AuC5TL3mMNFFsJkH/IXcHY7ofBuYjZoQz36oMvWzPCu9EEJ8CPVmCFkPoRDiQueacnmt8/0YgJQygVoYt1II8bp8BRZCnJrjpmaT8/3+nHRno4b/YPyrvH+Nmux/tdOI5Tt/JWo+E4aECiZ3z2REc2VWei+q3jPzcs9n7TvHuQdyuZghiz+EEMuEEEfnKXsNqucTY6hRz/ccZa5lkVOezPFVqJegOajhmxeyjhmtYT6KnJfEMSh4XxZCCPEG1HDeM8BrpVro/A3USMTnc5KPuH+dPI5CjYoI1MtK9vzjGtTvlW8udqLt1Nqsv6/MKcNFKKOpZxlqR7PzyT7HetTv8GYhxOLcEwsh/EKIk/KUaQSHi3HBu1HDDfuEEA+jzKpLUBNZ56K69jehrEoyFJp4zxgW5OspTOgHlVLGhBCPod5YnnYmVeeiHuLbUG9xuefP11i/iHrbOdPJ40lUg38xauhpdZ4yvYi6Ca5zGodeVENzixDCj2pcXsjpIfwXZU57FrDe8TsVRLmpaUCZZd+QlT7z9lSoIV0LtMrx+8DLPOBzsiYmNZTYnIF6K94CvD7nuD+i1ln8TQjxR5Tgn4YS3ydQk7rrYbAh/RuQFkLcixpe86GMUE520v2/rLw/garjvzrpX8oq01rUUE22t4DvoebuPiaEOBLVWK1A9cp2ocR7XA2elLLLacBuB34qhPgwatilF9WbWYi6772oBij7d5jMPfNf4A3An4UQf3au8wzUHNmLqLf27OfiFmCWc59kXmCOQT1ze1DDTKCG374shHgRZYzQg6q/S1HzU1fJIU8A+Z6jDiHE/SjXVk85v0MDqof6AGq4bnNOL3k16jnOV9cT6olnpb/YMaDIxz1SyozQno9ax7QNNRcZddLciPIu8X4hxA1Syt3O9vtRw6JXOmK6GfXCcCHKmm5edlmznt+XZH5DgLWo+b587obWoEY2tuRsAzWn+EMhxFmoXu8xqJeaXah58OzeYWZN1wuZDVLKdiHE11Gm5xuFEHeg7oNS1MvBaai5pBHGMiM42LG66figGo2volZj70W9QcRRjcovyDL3yzoms47jTVnb8i5cy9q/h8KGBY+g5mbCOdvnOT9oD6rBuAv1YGYWl30pJ/0G1BugnrN9IWpCuNf53IdqRC5z8hmxHgF4I6qhjJM1Rp110/wkzzEh1Jt5I+qNqhXVUJ81Sh2OMPumgKn6GL/j/eSfiOx09n2QApOSKA8E253ffjdqvqqSobkBb9b1/S9qAjuzyj6KeqP7n3z5o0xi/+LUheGUZwNqjuKcPOnPdfJLOb/VQ6gGPTMJntfwYJR6mY+abN7s/JZplMn2o6gFn3lNwid6z6Du/y+gnqEESsw+heoddDJy3uVjqAXYmfQJp16+StYkOkocb3Hy63HqsAm1KHddTp6ZBd31OdtrUD3ADtTQ7uMorx8Zgfl5znWM9hxnlkuM8IyRJ232+pbRPq920p/o3E97ye/l4z3kGNQ4249F3eMDqHnOO1AvTJn5y7dnpc08v/kMC+ZT2LDA79T9kznb/5apc9QzvRt17+5BtQXlee6T0er3CpQxS59zvhaUQcg3yDMXnO8jnIxcXFxcXF6GCCH2ol7MDpmwMIfFHI+Li4uLy8RxhvbmMcEQHVONKzwuLi4uL18y8zuu8Li4uLi4TAsZ4XlhJguRizvH4+Li4uIyrbg9HhcXFxeXaeVwWcczYwgh7goGg68Kh8OUl5cTiUQmnEdfX9+kjsumo6ODmpoRDgemtRzFuA63Loqbh1sXQ7h1McRM1kWysRGtt5cNlhWXUuZbyH14rOOZ6c/atWvlwfDe9773oI6XUsqDLUMxylGM63Drorh5uHUxhFsXQ8xYXXz1q1KClFdcIYEOWaBNdYfapoFLLrlkposAHHw5inEdbl0UN49i4NbFEG5dDDGhckgJX/gCfP7z8Pa3w+9+B2pRc15c44JxsG7dOvnss/lcQk1rGZjpMhwquHUxhFsXQ7h1McS018X3vgcf/zi8+91w882g6wgh1ksp1+VL7s7xHCZcfXXe2FKvSNy6GMKtiyHcuhhi2uvijW+E/n7V69HGHkhzezzj4FDo8bi4uLgcUtg2/PrXamhN10fsHq3H487xuLi4uLhMDNuGq6+Gd74Tbr99woe7Q20uLi4uLuPHsuBd71K9nc9/Hl6fG8VkbFzhcXFxcXEZH4YB73gH/PGP8NWvwuc+N6lsXOFxcXFxcRkfW7bAHXfAt74Fn/zkpLNxhcfFxcXFZXRsW1mrHX00bNsGc+YcVHaucYGLi4uLS2ESCbj4YvjJT9T/Byk64AqPi4uLi0shYjF49avhv/8FT/EGyNyhNhcXFxeXkQwMKNF59FG49VZlVFAkXOFxcXFxcRmOYcAFF8BTTym/a296U1Gzd4faXFxcXFyG4/XC614Ht91WdNEBV3jGRV9fH1dffTV33nnnTBfFxcXFZero7ISMe7CPfxwuv3zCWdx5550ZX3EFg/m4vtrGgeurzcXF5WVPezuce6763r0bQqGDys711XYIcuWVV1JbW0ssFpvpogxj/fr1CCH42c9+NtNFOWTrKJtDqb5cXCZNSwuceSbs3KnmdA5SdMakUIQ491O8CKS5PP3001IIIb/73e8WNd9i8drXvlbW19fLgYGBGSvDWHW0f/9++c53vlM2NDRIn88n58+fL6+77jrZ3d096XN2dnbKW265Rb72ta+VixcvloFAQJaVlclTTjlF/uxnP5OWZeU97lCoLxeXSbN/v5RLl0pZUiLlQw8VLVvgWVmgTZ3xRv1w+BRbeM477zwZiURkPB4var7F4qmnnpKA/NrXvjZjZRitjnbu3Clra2slIC+99FL5qU99Sp511lkSkEcccYTs7Oyc1Dl//OMfS0A2NDTIt7zlLfLTn/60fOc73ykjkYgE5OWXXy5t2x5x3KFQXy4uk+YjH5GyrEzKxx8varau8BxCwrNt2zYphChKXPWpZPny5XLevHkF3/KnkrHq6Pzzz5eAvOGGG4Zt/+hHPyoB+b73vW9S573vvvvkHXfcMeKaW1pa5Ny5cyUg//KXv+Q9dibry8XloEinpdy8uejZjiY87hzPNPOLX/wCKSVXXHHFiH0//OEPEULwi1/8YsS+vr4+NE3j7LPPnvS5t27dygc+8AGWLl1KOBymrKyM5cuXc8UVV5BKpYalfdOb3sS+ffu45557Jn2+yTJaHe3atYu7776bBQsW8MEPfnDYvi9/+cuEw2F+85vfTGpe6Oyzz+aSSy5By4mgWF9fzzXXXAPAgw8+mPfYmawvF5cJs20bnH++MiTwemHFimk9vSs808y9996LruuceOKJI/atX78egLVr147Y99xzz2V6X5M674MPPsjq1av5xS9+wTHHHMO1117LVVddxcKFC3nxxRfx+/3D0p9yyikAM9KQjlZHDzzwAADnn3/+CIEoLS3llFNOIR6P8+STTxa1TF6vFwBPAbchM1lfLi4TYvNmOOMMeOEF6OiYkSK4ngumkVgsxgsvvMCKFSsIh8Mj9j/33HMEAgGOPPLIEfsyorRmzZpJnft///d/MQyDp59+elx5HHfccQA8/PDDY6a9/vrr6e3tHXdZjj32WF772tfm3TdWHW3btg2AZcuW5T1+6dKl3H333Wzfvp1zzjln3GUaDdM0+fWvfw3ABRdckDfNROrLxWXGeOklZTKt6/DQQ9Pe08ngCs800tTUhGVZNDQ0jNiXTCbZsmULa9asyftWPVpvaDx0dnYSiURYuXLluNJHIhECgQD79u0bM+3111/P3r17x12WK6+8sqDwjFZHoIYcM+XLR2b7RIRwLD796U+zceNGLrroIl71qlcVPO9468vFZUZ48UU4+2wIBuH++6HAy9t08IoWHiHE24DfOP++V0o5pYsxurq6AKioqBix76WXXsI0zYLCsn79ekpLS1m6dOmkzv29732Pd73rXaxZs4YLL7yQ0tJSzj77bE4//fSCx1RWVtLW1jZm3o2NjZMqUz5Gq6OZ4IYbbuC73/0uy5cv5ze/+c2oacdbXy4uM0J9PRx3HNx0EyxaNKNFecUKjxBiLnAjEAVKpuOcwWAQUL2bXJ577jkgf4+mr6+PnTt3ctpppyGEmPB5pZS0tbUxf/58nnnmGbZs2QLA8uXLRz0ukUgMlnm6GK2OYKhHk+n55JLZXl5eftBlufHGG7nuuutYuXIl9913H5WVlaOmn4n6cnEZk40b4YgjoK4O7rprpksDvEKFR6jW+5dAF3A78D/Tcd7a2lpg6K0+m9GE57HHHkNKOen5nWuvvZYbb7yR97///fzyl79kyZIlI4wJcrFtm97eXhYuXDhm/sWc4xmtjgCOOOIIALZv3553/44dO4DCc0Dj5frrr+ejH/0oRx11FPfdd99guQoxkfpycZk2HnpIBXF7//vh29+e6dIMUcjO+uX8Aa4DbOB04EuABN5TKH2x1vHYti1rampkdXX1iH1r166VgGxvbx+x7+1vf7sE5G9+85sJn7OtrU1qmiZf9apXTei4zZs3S0C+7nWvGzPt/PnzpVOH4/pceeWVBfMarY6kVItHAblgwYIRa2b6+/tlOByWoVBIRqPRCV1vNt/85jclII899ljZ0dExrmMmUl8uLtPCPfdIGQxKuXKllM3N03563HU8QwghVgDfBH4gpZxWEyQhBKeffjqdnZ3s3LlzcLthGGzYsAGARx99dNgxf/rTn/jtb38LwKpVq0bkedVVVyGE4NZbb817zvb2dmzbpr+/H8uyRuxPJBJ5j8uYI5911lljXldjY+OEhL9QWaFwHWVYvHgx559/Po2NjfzoRz8atu+LX/wisViMt7/97SMs4saqpwxf/epX+fSnP83atWu57777qK6uHvP6YWL15eIy5fznPyqI25Il8MADUMBYZ6Z4RQ21CSE8KGOCfcBnZ6IMl19+OX/961/573//y5IlSwDYuHEj6XSauro63vKWt3DZZZdRVVXFiy++yPPPP09tbS1tbW185Stf4ROf+MSw9S22bQOF15ccccQRLFu2jCeeeIKVK1dy3nnnEYlE6OzsZNOmTSxbtizvgtW7774bXde59NJLp6AWRidfHWVz0003cfLJJ3Pttddy3333sWLFCp566ikeeOABli1bxte+9rURx4xVTwC/+tWv+MIXvoCu65x22mnccMMNI9IsWLCAq666asT2mawvF5dhRKMqWuiRR8Ldd0NV1UyXaCQTeVM93D/AVwALOClr25cYY6ht3rx5cu3atYOfm2++eUJdzmxSqZSsra2Vxx9//OC2n/3sZxKQP/rRj+THP/5xWVVVJUOhkDzrrLPkM888I7/zne/IUCgk165dK5uamobld+yxx8rS0tJRnWPu379fvve975ULFiyQXq9XhkIhuWjRIvn6179ePvzwwyPS9/b2ykAgIC+99NJJX+fBkK+Octm3b5+86qqrZH19vfR6vXLevHmjOgkdTz198YtfHHOY8Iwzzhhx3EzXl4vLCJ55Rsqenmk/7c033zzYTgKNslBbXGjHy+0DnACYwP/L2T6m8BTbSejXv/51CcjnnntOSinlBz7wAQnIp556akL59PT0SE3T5Cc+8Ymilu+GG26QgHzkkUeKmu9EyK2jg2Gq6inDoVBfLi7yD3+Q8vvfn+lSDMIr3UkoakhxG7AZ8Ofsm3bhSSQSct68efLVr361lFLKE088Ueq6PmFv1XfccYf0+/2ypaWlaGWLx+OyoaFBXn755UXLczLk1tHBMBX1lOFQqS+XVzi33iqlpkl55plSGsZMl0ZKObrwvFLmeEqAjH1tssBamFuEELegjA4+MpWFCQQC/OY3v+GBBx4gFovx0ksvsXz58gmvAbnkkksKrneZLI2NjVx99dV55zGmk9w6yuc+Z7xMRT1lOFTqy+UVzM9+BldfDeecA//4B4wyj3mo8IoIfS2ECAI/LLB7DbAaeBTVK7pHSnlbdoKpDH29adMmjjrqKN761rcOWq+5uLi4jIubboIPfhAuvBBuvx0CgZku0SCjhb4+9KWxCEgpE8B78u0TQnwJJTy/klPsMicfRx55JK8E8XdxcZkCbBte8xr4059gjEXhhxKvuHU8Li4uLoc9+/er7w99CP72t8NKdMAVHhcXF5fDi698RYUz2LxZ/a8dfs344VfiIiOl/JKUUszEMJuLi4vLuJESPvc5+OIX4fWvV44/D1NeEXM8Li4uLoc1UsInPwnf+Q68973wk58clj2dDIdvyV1cXFxeKfz+90p0PvjBw150wO3xuLi4uBz6XHGFsmB729tgEjG5DjUOb9l0cXFxebliWfD5z0Nzs1oU+va3vyxEB1zhcXFxcTn0ME246ir4v/9TC0NfZrhDbS4uLi6HEoaheje33aaE50MfmukSFR1XeFxcXFwOFdJpeNOb1KLQb38b/ud/ZrpEU4IrPC4uLi6HCtEo7NoFP/gBXHvtTJfmYCk4IeXO8YyDvr4+rr76au68886ZLoqLi8vLkURC9XYqK+Hppw9r0fnTX27n8je+BTTvrEJpXhHeqQ+WqfRO7eLi8gonFlOOPsvL4S9/OWwt1wZiSbY3trFzXzseXef1F52yR5qJRfnSukNtLi4uLjPFwABcfDE89hj86leHpej0DSTY1tjKngNd+Dw6dVVlaJoAFWAzL67wuLi4uMwEfX1wwQXwzDPKM8EVV8x0iSZE70Ccrbvb2Nvchc+rUVdVmhGcMXGFx8XFxWUmeOMbYf16+POf4bLLZro046a7L8aWXS3sb+0m4PdSX11KgajOBXGFx8XFxWUm+L//g/Z2NdR2GNDZE2XzrhZa2nsJBrw01EQmLDgZXOFxcXFxmS7a2uCOO5SH6eOOm+nSjImUks6eKJt2NtPa2U8o6KW+pmzSgpPBFR4XFxeX6aC5Gc45B/btU3M7c+fOdIkKIqWkvXuAjTua6egeoCTkY1ZtpGj5u8Lj4uLiMtXs3w9nnw2trXDXXYes6Egpae3sZ+OOJrp6Y5SE/UUVnAyu8Li4uLhMJY2NSnS6uuDuu+Gkk2a6RCOwbZuWzn42bG+itz9OWUlgSgQngys8Li4uLlPJ448r0+n77oN162a6NMOwLJvmjl427GimP5ogMsWCk8EVHhcXF5epIJ0Gnw/e8ha48EKoqJjpEg1iWTYHWrvZsKOZaCJFeWmQWTVTLzgZXF9tLi4uLsVm0yY44gh44AH1/yEiOqZp0djUyb8e3sCTLzXi8+rMqokQCvimtRxuj8fFxcWlmLz4Ipx7Lni90NAw06UBwDAs9rV0s3FnE8mUSUVZkPLS4IyVxxUeFxcXl2Kxfj2cdx6Ew3D//bB06YwWJ22Y7GvuZsOOJtKGSWUkTEVZaEbLBK7wuLi4uBSHHTvUOp3ycjXEtnDhjBUllTZobOpi085mDMumKhLC5w3PWHlycYVnHGTi8VxyySVccsklM10cFxeXQ5HFi+EDH4BrroF582akCMmUwe79nWzZ3YwtoTISwuvRp7UMjz18P48/8gBIq2DXyo3HMw7ceDwuLi4FefhhmD9ffWaIRDLN7gMdbNnVigSqIiE80yw4uZxx4jG7pZlYnG+f2+NxcXFxmSz33AOXXqqMCe64Y9pPH0+k2bmvnW17WhFCUHkICM54cIXHxcXFZTL8+9/wutcps+mf/3xaTx2Np9ixV0X71BBUV5Sg64fP6hhXeFxcXFwmyt//ruLprFql3OBUVU3LaQdiSbbvaWPnfhVeurqiBF07fAQngys8Li4uLhPBtuGb34Q1a5TDz/LyKT9lJrz07v2d+L2e7PDShyWu8Li4uLiMFylB0+Bf/1ILRMvKpvR0Pf1xtu5uZV9LNz6vRn314S04GVzhcXFxcRkPt94Kf/kL/PWvUz60Vozw0ocyrvC4uLi4jMVPfwrve5/ySmBZU3IKKSVdvTE27Wqmpb2P0EGGlz6UcYXHxcXFZTRuvBE+/GG46CLV2wkEipq9lJKOniibs8JLNxQhvPShjCs8Li4uLoW46SYlOpdeCrfdBn5/0bLOhJfesL2Jzp5o0cNLH8q4wuPi4uJSiJNPhve8RwmQ11uULG3bprWzn007m+nqjVE6ReGlD2Vc4XFxcXHJRkp48EE46yw49li45ZaiZDvd4aUPZQ6/lUcuLi4uU4WU8LnPwdlnF80FjmXZ7G/t5q5HN/PI+h1IaTOrNkJJqHjDdlPFJ779Vz7x7b8WPV+3x+Pi4uICSnQ+8Qn47nfh6qvh1a8+qOyyw0sPxFNUlE1veOlDGVd4XFxcXKSE666DH/4QPvQhuOEGmKRVmWlaHGjr4aXtTcQTaSojIWa/wobULEuC0Ap6K3WFZxy48XhcXF7mPP20Mpv++Mfh29+elOhkwktv2NFEKj3z4aWnG9uG3U1d/OhXd2J1bwZBeaG0rvCMg0gkwk9/+tOZLoaLi8tUccIJ8Mwzyv/aBEUnbZiD0T4z4aUrIzMfXno6kBKiSYPNe9q4/e7neWlLIwBXvPqNvPj0g6lCx7nC4+Li8srENJU3gte/Hi68ENaundDhh3p46akkljTpGkiyvyPK7f95ik3b9qJrGieuXsoZJ66kIhLml6Mc7wqPi4vLKw/DgLe+Ff78ZxVP58ILx31oJrz05l3N2FJSVR6e9vDSM0EybdE9kKKpK8aB1h4qK0oI+b2EAl5OPW45Z5y4krKS8Q0tusLj4uLyyiKVgje9ScXU+e534WMfG9dhiWSaXfs72Lr70AkvPdXYUmJakud3dtIbT9Pa3sNT67eydWcT1737IspLKnj9RSdMOF9XeFxcXF45pNNw+eUqrEHGgm0MYokUu/Z1HFbhpTNrb779icsnfKxpSXpjKVq74/TH00hg5742HnlqM9t2txAM+Djn1FWUl01+HssVHhcXl1cOXi/Mmwc336zW6oxCNJ5ie6MKL62Lwy+89ESwLEl/wqCtJ05bbwLLtgl4dTyahi1tfv3Xh/DqOheceQwnrVlGMOA7qPO5wuPi4vLyJxaDjg5YsED5XRuFgViSbXta2bW/A4+uU1N5eIaXHgspoT+epqMvQXN3AtOy8Hl1SgM6u/Z1sXHbfgzLpKNrgDddchLLFs7C5yuOZLjC4+Li8vKmvx8uvhiammDz5oJhDV5u4aXzIaWySOvsT9LUFSVlWHh0jXDAg6552LKzifse28j+5i4ipSEsyyKZMtixp5WjjphXtHK4wuPi4vLypbcXLrgA1q+H3/8+r+hkwkvvbe7C7/O8bMJLZ5NIWXQPJNjfFSOeMtGFRknAQ0lQedzu6hngN7c/QnNbD5XlJbzuwuNZtrCBb930DwCefWk35566itJxWq2NhSs8Li4uL0+6u+H88+Gll1TI6ksvHb67L8bmnc00tffi93ledsHXbClp7Y7T1B2jL5ZGE4JwwENVqRJf27bp7B6gurKUSGmIYMDHG199EquPXICua9x+19PIrLzufXQDl11wfFHK9ooRHiFEFXAZcDGwCpgNpIENwC+BX0op7ZkroYuLS1H55Cdh40ZlNn3RRUD+8NL11S8fwTFMm95YimjSwDBtNu/vIRzwUFnqH7xGy7J5buMeHnhiE6Zp8cn3vwaPR+d9bz13MJ/+aIJnX9o9+L9l2UXt9bxihAd4A/BjoAV4ANgH1AGvA34GXCiEeIOUUhbOwsXF5bDhu9+Fq66CU08dDC+9aUczrZ19lIT8L5sejmVJ+uJpWnvitPcmsKXEsiVej0ZV2dDQomlaPPvSbh54YhM9fTFm1VVw4ZnHouUxnLj30Q3kNoXF7PW8koRnO/Aa4F/ZPRshxGeBp4HLUSJU/OATLi4u00NTE3zpS/CDH0AkgjzlFNq7+oeFl55dVz7TpRyV8azBsW0YSKRp603Q2hPHtGz8Xp1I2IemCfQ8c1Q7Glu5/a6nmTerite+6jiWL55VUHj3NXViWcMHgCzLZm9T50Fc2RCvGOGRUt5fYHurEOInwNeAM3GFx8Xl8GTfPhXArb0d+/3vp3XOIjbuaKK7L0Zp+PCP9plxyNnZl6SpK0batPB6NEoDHrQ864ts2+bBJzcjhOCME1awfPEs3v/281gwp2bMnt5H3q2GJn/y23sAuOZt5xX1Wl4xwjMGhvNtzmgpXFxcJseePXD22cieHjr++Beei/vpfXaHE166fKZLd1BkHHI2dcVIpEx0TaMk6KE05M2bPpFM09UzQN9AgsYDnaw6Yi6csAIhBAvn1o77vD+5dw/NPUlmVeQ3Pz8YXvHCI4TwAO9w/r1rJsvi4uIyCXbsQJ59NnY0yhPfvpkDooKIE176cMWWkuauOE1dMQYSBroGIb932JxNPl7Y3Mjt/3maZEq9Sx91xFzefvnp01HkCTFlwiOEeBvwRynlod6L+CZwFPBvKeV/8yXo6Ohg3bp1g/9fffXVXD2Guw0XF5epx7Js2g60U6r7eOyrNyJWHMmsg3TnMlOkTZveaIpowsCwbLY19RL2e6gq8496XDSWxLJtIqUhqivLmD+7hm27mwHYtquZgWiiaOtvxuKxe//O4/feof6xUgVPOpU9nl8BdwPtU3iOg0IIcS3wcWAr8PZC6Wpqanj22WenrVwuLi6jY5oWzS9u4YU+k3jSR+Wv/0YoVPwhoanGtCR9sTQt3TE6B5LYEiypLNIqS0cXnL6BOA89uYWnnt/B0Svmc8UlJzGnvpKK8qGYQMVefzMWp5z7Wk4597UAXPv2CxOF0k2l8BzSdopCiA8BPwA2A+dIKbtnuEguLi5jkAkvve8/93PyJ9/H8re8i9Z3XzPTxZoQhRxylod8iAIWadn09EV58InNPP3iLqQtWX3UQs46+Uhg6tffFItX5ByPEOIjwPeBjSjROWR7ZS4uLkPhpTfuaKZkywbO+tyHsUJhes8ffwC3maSQQ86yYH6LtNF46MktPP3CLtYevYizTlpJVUXp4L6pXn9TLKZaeD4uhHgcWC+lPDDF5xoXQohPoeZ1XgDOk1IWxzDdxcWl6KTSBnsOdLJ5VwuGZbNw/w7WfPaDmJEIL/7416RmzZnpIhZkNIecHj2/RVo+2jv7uP/xTRx/7GIWzavjnFOO4syTVlJeNjLM9lSvv5kIQvcVHPucauF5E/A/AEKITuB5YD3wHPCclHLPFJ9/GEKIzwNfccpwvju85uJyaJIJL71ldwuWbVNVHiaQiLH6f96PUVHJSz/+Nan6hpkuZl4yDjkPdMWJpQx0ISgJeAcdco6XlvYe7ntsIxu27MPj1Vk4r5ZF8+pGHTLLrL/5wg/vBOArH75k8hcCzFpxFNecu3BCxwwkDF7c2w/eYEmhNFMtPMcBSWAtsMb5vA74NCCEEL1SysopLgOok12JEh0LeAS4Ns8iqkYp5a3TUR4XF5eRjBZe2iotY9sXvkF0xVGka+tmtqA5pA2LT3znr6QMmzdeehqaJgj5hhxyTgTTsmhq6eb7P/s3fp+HM08+ktOOW05J+NA1nkikLTYd6Of5xj52tsWQEjAS0ULpp1J4JICUsh/lG+2BzA4hRBglQqun8Py5ZGRbBz5SIM1DwK3TURgXF5chcsNLV5WHB6N9VjzxCFoySddZ59F9xjkzXNIhMg45m7vjdPeniKVMdCGGOeScCPuaO5lTX0lPXwzLlsxpqOQ9bzqbUHB067apIm3atPWlaO1NUl8+UvQMy8ajCYQQ3PViG0/s6KGyxMvZK6s5dkGEr/0nnSyU94xYtUkpY6hexyNTeP7cc34J+NJ0nc/FxWVsxgovXfnIA6z81IeJLl1O1xnnwAxHAs3nkDPo06ko9eF1yj0R0ZFSsntfO/c9uoGde9u47FXHEY2q9rqto2/EfM10kTZtWntTmLbklw/t4+MXL8Hn0bBtyc62GC/s7WPj/n7efeZ85teEOPWIKtYuLGduVXBc1z+VwvMqoG8K83dxcTlM6Y8m2N7YNmp46aoH7mHFZz9KbOkRbLzhZzMmOmM55JwMUkq272nhvkc30nigg9JwgFefs4am1u4pi4EzEf70ZBOWrUoykDD5w+MHqAh7eWFvP9Gkid+jsWpeGX6v+k1qxljkmstBCY8Qoh/4O/CuXA8FUsp7DiZvFxeXlx/Z4aV9Xr1geOnqe/7Nis//DwMrV7Hhhp9hlZTmyW3qGOaQsztG2lAOOUsCnmE9ssnnL7nznvWkDZPXnr+O445ZTCJl8LUf/m0wzUytwXl6Vw9bmwYGBdC0JdtbomgCljaUsHp+OctnleD1TL4eDrbHUwK8FQgJId7oBlJzcXHJR09/nC27W9jX3D2u8NKlmzbQv+pYNl7/U6xwQeOoohNPZcyfcxxyTtAiLRdbSjZu3ceTz+3gyjecgd/n5co3nEFFJIxHV8YTd973HLnBwKa719MbM7hjfQtpa3hJDEsS9uu847R5RTnPuIRHCDEfuFVKeVae3RIV2fP3Qog3jxZIzfHftk5K+ZHJFNbFxeXwoqs3ypZdLeMOL63FY9ihMHuu+yQinUb6p35iPZm26ImmONA5MYec48GybV7cvJf7H9tIe1c/NZVl9PTFqK8pp6aybFjafU2dqquVffw0rsG568U27t+kziVgmAh6dcFFxxbPknBM4RFCvBf4LnB7gSQ3AqehInwajOLzDFgMfJjCVmUuLi6HOZMNL91w+x+Z9/ObeOGW35OaNWdKRSfjkLO5O073QBIhxLgcck6EeCLFD2/9L109A9TXRHjLa0/h6OXz8kb8BLUG5yf37qF5y0ZmVQQOKgaOrXsxSuoKWqSlTZtNBwZ4obGX16xroKrEx6LaMB5NcOyCCHe92M6Gff1IwKMJVs4u5bjFFZMuTy6jCo/jWubrwHVSylsKJOsGzgMeBN4ihDCklO8qWgldXFwOC7LDS7d19RMO+sYdXnrWbb9hyXf+j65TzyRdVTMl5bMsSW/GIWd/Ulmk+T1Ulvq5+Xf3Agcf8ExZqbWxaF4doaCfZQvrWXr2alYum4M2TWG206aNXTkPmWORZjlzNc839rHpQD+GJYmEPHRH01SV+FjWUMKyBjWs+cYTZ7P5wACmLSkNenjDibOLWsaxejzC+VijJZJSdgshzkWJz5WO+LyvOEV0cXE5lJFS0t49MCy89ERi4cz5zc9ZdMP/o/PMc9ny9e8jvcULa5BxyNnel6C1Oz7kkDOsHHIWi7Rh0tcfp3cgzi1/uJ/PfvC1lJYEDw2LtMcOcOUZ80imLX718D4CXp01C8tZvSDCgppQXkH0eTTqy/209aV45xnz8B2EIUE+RhUeKeX3hRBx4HohxMlSyveMkrZdCHEW8DDwHkd8PlTU0s4QfX19XH311VxyySVccsnBuaBwcXm5YNs2rZ39BxVeuvbf/2DRDf+P9nMvZNtXv430HNwkPqhpkoFEmo6+JM3dcQzTwueZnEPOsUilDZ54bgcPP7mFaFwN2b3l0pNnzMvA07t62JJjkba5aYCnd/Vw/OIKPnDeQmZVBPCMox58Ho25VcG8Q3Wj8dIzD/PCUw+DZRR8gxhzjkdKebMQ4h7gl+NI2yaEOBvlAeD9jvh8dEKlPgSJRCL89Kc/neliuLgcEti2TXNHHxu2N9EXTRAJTz68dOeZ57LnAx9l/9vfA57JG9lmHHJ2DSQ50BklZapV9eGAB0+BENHFoD+a4D8PvECgtAxNN7Ati9372jlm5YIpO+do3PlcK0aORZoE7nqhjeMXVzCvOjRl57Ztm4FogjlL13DCqefw7JNn9RZKOy75l1LuBs4cZ9om4GxgH8of2rfGc5yLi8uhjWXZ7G/t5q5HN/PI+p2AZFZNhHBoghPyUtJw+x8HLdj2v/OaSYtOIm3R1Bnlme0dPL29jca2fvxenapSP5Gwb1xv9hMhFk/x34de5LY7HwegprKMT15zCTWLlmJbakbi2Zd2MxAtGAOtqAwkTB7b1kUirc69YlbJCJcxXl1wYREt0nKxbJve/hi9/XFm1VVy4uqlLFs0C6Dg8ppx/9qjmUnnSbsvq+fzP0KItJTy8+M93sXF5dDBNC2a2np4aXszsUSKirIgsyc4pDaIlCy6/pvM+f2tiFSK5jdfOeEs0oZFTzTNga4ofbH0kEPOIpg/FyIaS/LwU1t4/LntpNMmq46Yi2XZ6LpGVUUpPc37B9NO9dqbpGGxcb+ySNvhOOQsDXo4el6EN588B1semFKLtAyWZdMfTQCSWXVVVFVEsDWN5r40/a1xhDcwMm6Dw5S5zJFS7skSn88KIYypOpeLi0vxMU2L/a3dbNjRTCKZpqIsRKR0koIDYNss/s7/MfvPv6PpirfT/KZ3jPvQjEPO1u4Enf1JpJCEHYu0yTjknAhbdzXxm78+gmnZHLNiHmefchT1NeWD+/ujCaIdQ7EkD8bjQNq0SUXmkKY37/6BhME37tiBaUkqw17OWlnNsfMjg/MwQogpt0hLpk26++KYpqSsogx/OEyHIehsjyEReHWRMUYo+MMcrPB8DthUaKeUcmeW+HwRmNb4Oy4uLhMnE156w44mUmmTirIg5aUH6bLFtln6zS/R8Lfb2P+2d7Hn2k/CGIJhWZK+RJrW7jgdfQks23HIWVJci7R89PRFiSfSzK6vZG5DNauPWsAZJ6ykpqpsRNp7H91QFI8DGcecUvfSSiXJtMX+7gQvNPahaYLLj59FadDLeUfVsKguzLwCDjmLaZEmJcRTFknDoj+RpqsnimlDXW0l5dWlhAI+fB4djzYx56gHJTxSyq+PI802IcQ5qLAIi2DEb+Ti4nIIkB1e2rAsqiIhKiPFmYz2dXVQ+cj97HvnNTS+/yMFRSfjkLO9N0FLTxzTkvi9GmWh/A45f/Jb5RLyYNffZOjsHuCBJzaxfsNu5s6q4oPveBXhkJ/XX3RiwWOK5XFg0AxaCEyp8ZXbt2HaEr9HY/WCoZ7mWUeOvc5pMhZppmmTsmxShk0sZRJPWUgkO1r6iCdS+Lw6c2bXUV8TGYyRNFmmOhAcAFLKTc46nweA8uk4p4uLy/hIpYeifZqWpDISxOctUtNgmqDrpGvqWP+HOzDLR8Z9VBZpBh3ZDjn14jnkHIuf3LuHA50xfJh8++Y70XWNk9Ys44wTV47r+GJ4HLhvY/swM2iEhiUlJy4p55I1DQflkDMXKcG0bNKmTdKwiKUs4ikTy5ZI1PjYMx0W/WlJqUfi1WDF4gYqyssOWnAyTIvwAEgpXxJCrAJOna5zuri4FEaFl+5g864WbCmpKg/jLVLDAiBMg+Wf+zip2np2f+yzI0QnnjLpchxyxovokHMiZGymLNMkmk5x4uqlnHfa9HiD7o0bvLi3j+cb+2juGRkzTUrYuH+A1x0/+TkaKdX8WNqySaYtYimTWNrCtpXECMDrEfg8OhmNN0wTaUkWzKvisqOrqCgvRdeLd1/ANAoPgJSyGfjTdJ7TxcVlOPFEmt0HOtiyuxUkVJWHivYmm0Gk06z4zHVUP3w/uz76mcHtGYecTV0x+uMGmoBwoDgOOSfC/pYu7nt0I0csagC8SFtZ/grBlItONGny20f3s6c9jgTmVgU5Zl4Zm5sGhq3BmagZtJRqnihtWhimjSUlW5r7HZFRczBeXRDIEpls0oZJIpnC5/PiD3jxeDxUV5Uf5NXmZ1qFx8XFZeaIJVLs2NvOjsY2NE2jOiu8dDHRkklWfurDVD7+MDs++QX2XvZmensTwxxyhvx6UR1yjpfG/e3c+9hGtu9uIRjwsWxRA2Y6jW0qo9upsEZLmzabDwyQtmyOX1xByK8jhODcVTWsXhChulTVw28f3T9uM2hpQ9pSIpNIO3MyaQspJc+0W3QlbMr8goBXRx9jzj+dNkgk0/j8XhbOb6A8UsKGZ1omdP0TxRUeF5eXOSPCS+eJ9lk0pGTFp6+l4olHeO5jX+Sp4y+gY3PrMIecU23+XIh/3P0sjz27jXDIz4VnHctJa5YR8Hu572f3DqYphjVa2rTRNcH2ligv7O1j04EB0qbNnMoAxy+uQBOC952zYEQ+hcygbRsMR2TiaYtY0iJhWIPDhJom8GqCkE9HE6BrNgjU36NUdcoRnGDAz6KFs4iUlUw6oupEcYXHxeVlylB46U68upY3vHQxyTjkXH/axSSWn8zzK8/AH0sX3SHneJFSsn13C7PrKykJB1i5dDYV5WFOPHYpPp9q+oqxBifbGs2SGn9+somQX+eJHT0EfTqrF0Q4dn6EhbWjWwj6PBp1ET/t/SneeOIsuqNpYimTZNpGkiUyukbIr4/P7UweUqk0iZRBOBhgyaLZlJWGp01wMrjC4+LyMqN3IM62PW00NnXh9WjUVZVOWcOSccjZ29KJ/syzbFuxFt/yEwgfo1M1DRZp+bClZMuOA9z32EYOtHTzqjOO4ZxTjmLpwgaWLmwYlvZg1+A8vbObzVnWaFJobG4a4MwV1Vx5+lyOaCgp6LbHtiQpS5IyLBJpi7v3JOhJ2lSW+EimLdKmrUQmMHmRySaZTJNKG4TCAZbOqaWsNDxlvc8f3t+IN1xeW2i/KzwuLi8TMuGl97d04/N6plRwognlkLOpKwq9fbz9+5+k7sBufvWjvxIPVU3JOXP5yb17aO5JMqtiyDDhpS17ue+xjbS091JZXsLrLzqBNasWFsxjsmtwkobF49u7+e9L7bmHY1iSJ3Z084XLl2flKdU8kGkTdyzLUmlrcDmTnvmdBOgalAQOrmleNL+aM+Yo59CJZIp02qSkJMT8+fWUhPMvPD1YpJTKci6p5ptGwxUeF5fDnEx46QNtvQTHGe1zMiTSFt39CZq6EkSTaXRNUGXEef33P0HVvl385+NfI14+PtEp5sLPbDeSz29qxDRtrrjkJI49csGYQ4sTWYMTTZr0xg3mVAbRhOCBTZ1Uhr30xs3B+DegrNHOOaqG3miamGPCnDZtJ5y0wKOBR1em49m/khCjN9YTQkIikSJtmJSVhVk4fxbhcGBK7otk2iSeNLGlxERjVlVBF22DTFp4hBC7J3molFIunux5ZwI3Ho/LoYaUks6eKJt2NdPa0U8o4B13tM+JkOuQcyhEdIBgXzev/fpHKG/ex78++S32rj6pqOceC2nbGOkUjfv72bO/nYVza3nDxScS8HsLhpfOx2gWaUnDYtP+AZ7f28fO1ii1ET8fu0hF9PzMa5fh0wW/e+wAmw4MAGoyf3ZFgPKQhwM9CTy6wKvrlAamZ9hRSolt2dgSgqEgi+oqCYeLbx6eNiwGEgbb2mL0Jy16EiYtfUkMS7IikgJGt1g8mB6Pxkj3Nz4gM4hqAZ1ANZBZJNACpA/inDOCG4/H5VAhE156444m2rsGJhReeryYlqQnmhzmkDPkG2mRtuyxe4i07Oefn/o2+48+rmjnHwvDtHj2xV3sf+lFrLRqTh59eisL59YSCk7MRDufRVrGt9kDmzq4Z2MHpiWpCHs5fXkVK2aXDk76x1ImpiUxAyEEap4n6Nc5/+gavNM8vyWlJJ5IYZoWuu4j4POwZNHEF57eeH8jsaxhsm/dtQuAgFfj1MUVHOhJ4tHgzCOqWTorwp0bO+lPmiysCXPe3HIW1oRZWl/CR377wqjnmbTwSCkXZP8vhCgD7gX2Ap8BHpVSWkIIHTgN+AZKrM6d7DldXF6pSClp6xpg4w4nvHTYP+Fon6Mx0iEnBH1afoecUllwvXjhG2lccwp99XOKVo6xy2nzvVv+SVdPFF8oDI7wbN3VzEA0MeH1N7n+0b73r51cc+5Cwn6d0oCHo+eWsbAmSFnQgy1xfJiZeDwCn64T9ILXa+MPeDDTFq9eXTetoiOlJB5PYto2lRVl1NdWsu2lifmIyyZWYG4madjcu7WLoFdj1dwIJ6+sRwjBly8/itKgJ2/47NEo5hzP11B+2I6SUg72aqSUFvCgExZ7g5Pu2iKe18XlZYtt27R09rNxexM9/ZMLL10470yIaLW407QkvlEccgKUdrTwquu/wP3XfIbuuYumRXSSKYNN2/ezdtUidF3jtOOWU1MV4Y/3biQdj6lrmcT6m1yLNIRGd8zgb880cdziCgI+nVVzy/B6BB5Ny7vaP4OmadRGdKpKCkZ7Liq2rXo4tm1RVRWhrqaSQGDi5+4YSNPYFaetP0Vrf2rUtNe/9VhqI8N7vZEC0V0jQS/No+RVTOG5DPhDtuhkI6VMCiH+AbwJV3hcXEYlE176pe1N9A8kiJRMPrx0NtkOOZu7Y6Qm4JCzrK2Jy77yYfyxKN7USN9i4yWfNVo+Esk0jz27jUef2Uo8kaahtpxZdZWcvO4Itf6mc2LrbzKr/WMpA8uGfz7fhmnlzhZAY0eCs8fhAXomsG1JLJFE2pLq6nJqa8oJ+JXgFBomC/t03n3qXNr6U4MCc+6KasJ+D7s6Yjy0vZsSv051iY+uaOGwaXUT8HT9k3et5ZTvtxe0Ayim8FQBY3n38zrpXFxc8mBZNs3tvUpwYknKS4vTw8nrkDPgoWScDjkjLfu57CsfxpNK8rcv/JCORUccdJkKkUwZPPjEJh5fv51kymDF0tmcc8pRzKobcjI61vqbXJcyrX1JtjYPsLs9TsCjcd4xtRy3qJxHt3cPM4f2aIKTlhY/Wud4yTaDzsa2bWLxFFLa1NZUUFNTgd83/LcrNEwWS1vccH/j4P+RoIeBpEXY7+GI2hCzy3yEfDpVZYFBsZpqiik8u4DXCyG+KKXsy90phKgAXg9M1hrOxeVlS1HDSzuk0hbdIxxyTjxEdFnrAV73pQ+iWSZ/++KNdM1fclDlKoRt22iahiYET7+wi6ULGzjnlCOHCU6GQutvdu5rZ3dbbNClzL6uBFuborT2qWGk2jIfi+rClPh0jp5XxovNcfoHnN6btFlQU8KK2aVTcn2TwbJt4rEkUkB9bSXVVRF8WYJjWDZ7uxKD11eIM5dVUlfmp67MjyYgmjTo6k9SFvKxfE4FlWUB/F4d1YxPPcUUnp8ANwBPCyG+BjwMtAF1wBnA/wL1qDkeFxcXhsJLv7S9iWTKOOjw0mnTpjeaork7Tk80CRy8Q85YZQ1NK1fzzOVX0TNnYdGDr/UNxHnwic3s3tfOde+6AJ/PwyeuuYRgnjkL25KkLcm73nIet9zXSFv3AFVmNxefp6zqbCnZ3hplUW0Iv0cnZaiYM8cvKmdpfZjy8PBewpxZ5WzZ3oqUUBr0cvaR1UW5ponwz90pUlmdlb/uUCLi0yRn1knq6iqpqoqQtKCxJ0Vr/wC1pT6W15dgWJK/Ptc65jnWzIsQTRjEEgbhgIdlsyuoLgsQ9A+XgEjQS19i5HBbpMihKoomPFLKG4UQS4EPA7/Mk0QAP5RS3lSsc7q4HK4YhsXe5i427mwmlTapjISoKJtctE/LkvTG0rT2xujoTQ465KwoOTiHnFX7djFQXUc6VMLd13150vkUwjAt/vqfp3j2pd1IKVlz1EJSaZNgwEcw4FOr/bNcysRSJinDRgJPtaRp70+D7qNXq6G9L8Xujji725WRhEerYWl9mGPmlbF6fmFzc03T8PuVRdrF02yRliFVYN1o2hYcuWI+d27opGnzARLGUOiGtfMiLK8vIeTTefuJs6ku8fH9e/cUPEfSsFhYX0Z1JEA44C1YHz9511oAvvK3TQB84bIjD+LKClNUzwVSyuuEEH8E3gWsBiJAH/AccKuU8vFins/F5XCjWOGlbRv642na+hK09cQxLRu/Vy+aQ87anZu59GsfZf+q47jrY/930PnlYpkG+5v7aWrt5rhjFnP6CSspCYdIGTY93Ym8LmU8ukY4oCOAlrY+NcfjOOa866UO/B6NZfVhltaHBw0X9HHUxXRbpE0En89LwKuxrC48OFRWU+obJpCzxjHpf8rKhml3BDoaRXeZI6V8Anii2Pm6uBzOZMJLb97VjGUzqfDSGYecnf1JmrriGJaNT9coDXjQivimXr9tA6/5+sdIlkZ47O0fHLZvvBZp+Wjr6KOzZwApwTJMAJYtmcOqY1fSFrNpj0VHdSkD0DWQ5vEd3UPzMgCahi7gpKXlHDmnbMLlmi5sKRlIS3pTkpghWVnlwTTHdpNz8dGjB4OzpSSRMgl6tcFeUTaRoPeQEh2YIl9tQogwsAwokVI+MhXncHE5HEgknWifu1onHV46ljTp7E/S3KUmzXUBJUEvHr34IaJnbX6eS77xP8Qqq/nbF35IrKqgg+FxYVo2e5u7efCJTWzbeYCScBCzYdlg4LUdu5o4bu2RlJUUFrL+hMGO1hjbW2N0FzD3tSQ8tat3WoUne26mrLYKGzU/49fhwgU+NKGifu7rt9jZa9GXlmRcuukCGjwp/P7JNcHZDjk1IaiOBPh/V6yivMTP1+/YAhzcMFnLGGt6DpaiCo8QYg7wA+ASlJscmTmHEOJU4KfAB6SUDxbzvC4uhxrxRJpd+9vZuqdtUuGllUNO1bPJOOQMB7yEDtJr8WgI2+KsW/4f0epa/vaFHxKvmNhEu+l4X06ZKiLmgZYennluCwcOtOH1ejjm6KUcs2oJv/v3+sFjpJQ8/8JWzjj12GF5JdIWfq+ycHtp3wAv7uunodzP6csrsW3JYzt6hzkH9WiCk5ZMrxl0obmZlAX/2JXmvPleSn3CiSgKiyM6JR6bAAblQQ+zG1S0z//sLzw3k0vGIacEykv8LGooo6IkgM+r7q1bntxPS3+KhhmI7joRinYXCyEagKdQVmx3ALVAttfAp5xtVwAPFuu8Li6HEgcTXjrjkLO5K0ZPLDXMIed0IDWdOz/1bYxAiET5SBPmYWml8tacMiyiSZNE2sKyVbgy27Lx+3Vsy6Kjo5vj1y7n6CMX4ff7iMWTWL0dg/nYtmTL9n2sW7Mcr8/HnvY421tj7O9O8Jo1dcypDHLs/DKOnldKWZZl1UstiRkxgzZtNVTWmxq58DSbpRU6ujNBNb9MpyFgk0im8Ae8zKpXgpNxZBr26XnX4IR9SkzSpkU0biKRlAR9LJtboZy0+g6t4AJSSlKmPRhLaMQ4aRbFLPkXUcJynpTyASHEF8kSHimlIYR4BDiliOd0cTkkGIglnWifHRMKL53tkLNrIIlNfoecU8mCZx9l7oZneOSqj9Cf4wJHSjBMtRDzH9titEZVCIB42qSxI44QKgyAV9fo7Ojk2ee3UllexhmnHsOcWVVc+eZX4c2ay3pq/dYR62+klNx+z/PESmZh2ZLSgM7q+RHKguq4fLFpBs2gbYku7QmbQY82TPbqRf7BcgkhSJqSlzpNelNqjmY8rKpWZR5PeOkPnb2AB/YNsKWxh3K/zltOmI1p2UQTBl19SQJ+D0tmlVEVCRIOFH94dTLYUpI2bBKGhWHZgEAgKQt6mVMZoiLsxU5G+wsdX0zhuQi4Q0r5wChp9qEchrq4vCyYTHjpjEPO9p4Ebb3xQYecMxEievFTD/Kq6z9P54KleBIJYt6AWu1v2CScgGW2MzGRtmyQ6kVW1wSlAV0t0jzQzrPPbaO1vZtQKMCShUNekb05BhS79rWTi21LYn0DrDyihKX1YeojY4tuxgw6GY1RaXfj1Se2qHW0YbKnWgx6UpLZJRqrqj14NehK2ET8GnNKNCr8gvKAxr/3FHa0n0ylSU4ivLRpS7r6k3g9GnNqSqgrD1ESLGz+PB3YUpLKiIxpI4SSmUjIw/yqEBUlPsJ+DyGfPjzaqrQLWk4UU3jqgB1jpDGAsaMEubgc4vQOxNm6u429zV34vGOHlx7ukDOBadljOuQsJrnBzqQNCx+9mwtu+ioHFh7Bj6/5Oj3dJlJGEUjQBF5NEPDq6IMmzc7wSVZxn3luG888t5WScJDTTzmaFcvm553LklLy9K4ezLplw4KpeDTB6csrWT6rZNyN63YzQEKaaJrGrJoyLls3Mfc9SXP0Xkt30qbcrxHxqfLomuDChRObM/F6PMybW0dpSWjU67KlJJ40MSwbKSU+j8bapbXTdl+MKI8tSZoWUqqfqb0/iSYE5SEv9eUhKkJ+wgGdkM8zLlP1QhRTeLqBuWOkWQaMvcz2EMMNBOeSoac/zpZdLexvHTu8dMYhZ2dfkqYRDjmnb8jEttVHSuXZYHdbjKOf+C8X/+F77Fy4kpve8xVsf5CQpjFWWyKlxDLSdHTEicWTLFsyh5JwgCOWzht1LuvR7d28tG9gxHbTljyxs2fK52d291m0xmx6UzYJc/S04xUZv56/5xT0aixbOreg4EgpiafUvJgmoDYSoiRg4NE1Qn4P5SXTYxhg2ZKkYZE0VK9WItAEVIR9nLCyFo+ucfbKOkI+fUIieOedd3LnnXeCWseZl2IKz2PAa4QQ9VLKEeLieDW4APhtEc85LbiB4Fy6eqNs3tlCU/vY4aXjKZPu/iQHumPEkxN3yHkwZFzKpC2LeMoimjJ55ECa9pgKTWxYEktKjPIKtq1Yy2/e83k8/rFj2GxN+2hvaiS5aysyncICnn1OWaOVR0qGpY0lTXa0xdjRGuOcI6upLPGxpC5MKm2zoz0+OHQHE7dGU3MzytQ3d27mrLk+elM2vUlJT8ombsJ589QwVXfCJmpIqoNqqOylzoMPM33xQh/JZJoHOzSEEFy1rm7U8NLJtEksqVSvqjTA0lnlVJQG8Ho0nmiLH3R5RsOyJYm0RcqwsJzujEfXqAz7mFcVIhL0Eg54CHqVyPxlQxuQf35tLDIv6LfccssIn50Ziik83wYuBR4SQnwECMHgmp7Tge8DNvDdIp7TxWXKyA4v3dbRTzBYOLz0cIecaTQhJuWQcyLkdyljIRGARNfUpH9K6NgoVyuLE134dMH2I49n+8rjYBzDW9t37ufxZ7eRGIhCsBRIAUPWaOFQAMOyB9faNHUra7OaUh8pUy1obCgP0FAeoPeFLto6nJ7PJKzRRpubuatRzbkIoMwnqPQLLAkeAWvrPMN+t4MRHikliUQKw7SIRErQNBMhBCV5wjGkDLXWRkpJWcjHynmVWQ45pw7LlnT0p5xRTYlH06gq9bGwJkRZ0EvY7yHo08cc3pRSrT2ybPXCYtnqk73NsGznXrQxLPW/YUs8wZKCi6qK6avtKSHE+4AfA//M2pWxbDCBd0kpNxXrnNNFW/TgonVf/8heAD5y2vwZz+dQKkux8il2Wa47dd6I8NL1eQTHMG16shxySgn9kSq0CsF8K3pQZdmrl7C9PU4Am9OqlPWbYdqkDJt42hzmUkZK8OgZlzLDV/unKquhRb29XtP0DF/fcz+3HDufnUesHlV0VAhl9SbfkbTw+H3ULl5L+85GMpM0Ukoee3oL55+5GtuGh7d2URLwcNyicpblcciZqqxmxaowbfdvAjkxa7T+tM2evpGr8rNZXeOhPCCI+MSI+Yfc387n1UjnWeXvH0UPssNLV1SUUl9bSavmR+7fP6zODVNZpNm2JBTwsHRWhOpIcIRDzgx37+6he6zxvwIYlnKCmkxbPNUaoytucPSSKj546nyqS/2U+D34vTpej8CWak7JsiFu2AykrbyicceWThq7E1QEPTy4oxMpxdC8nnSsSyQghBOMVqIJgS4EQoAuBL99rplAZcPCQuUutq+2Xzgm0x8ATkTF3ukDngRulFJuK+b5XFyKiZQSw7S494mtdPXmDy+d1yGnTx90yBktwoSwadnYQq2J2by3j117bdbVaM7zPrpLmUJcu/lffHnPffynbgV7Fh9VMJ1hmmza0sjzL+3kjFOOYdGCBlafsprKo47iiUe2QKx7MK1tS3bs3Mcpx68gHArw5pNmUxb0FHyLvvu5FtKGTdWsano7eimvqeGO3cagCXO2S5mepKQ3ZbO8Uqc+rJMy1TzNaCwqH38v4vw1DezpSdG4r4egsPPGwMkwFF5aUukITjDozMMkh4YN+2IpJ4qrzoK6MmrKR3fIORaqZ+F8bEiZFvG06t3aUoWRyLxshP0ePM6LgsejkZKwtzcxKBpKL8YnGq19CfqTBn0JgxsfauSiI2tZO6983OVev6+XbS0DCE0vqC9T4attB/DRYufr4jJVZMJL9/THMU0Lw8MwwZmoQ87t7XG1bmAcIQ9NyyZtqkneWMoikTJIWxK9PqQmfCU07uukt01w2bqGSV3f+x77Pde++Cf+Ur2Czx57Oa/xjJxrSqcNNm7ewwsbdpJIppndUE0opBpXXddpaW4n0dQ4Yg0ODM31FAqDPHgOp4fh9XmpmT0U4TNlQcyQ3L03PcylTLlfDBrAVQUFly728bedBzf6MBHGCi9t2xLTVt6ypZTMqiyhtiJIadCnFtJK1ZOwbbBR4mFLNTxl2xLTkpjOBH9mezxt8VJTH0nDIm1IDNseFA2fV6c04KU05CXs1Qn6NPxeHU0oo4D2/uSgYHz/gd1cdGQt6yYgGDAkGpmfuSdh8qfnWwAGxce0JW39KZKmRdIxs04aNouqQ8yKBLhzYzv2yNtkGMX0XPAO4AUp5UujpFkFrJZS/rpY53VxmSy54aWlDLOlPUFLr+CqI/1EkwYdfYlBh5xeXUzaIaeUKlBZtkuZWNLEdJ5QAei6wKfr+L2Q0gTCeQsdd7cmD0s2P8u1D/+aPyw4hQ/MPoXKAmuM/v6vx+jo7GXenFrWrT6ChvoquqJpntjRw4pjIuzc0QrJWN4La23vHrl9RLLRW6KQB5aU60R8gvKAoNQrhvUUtGlcx2JYklg8gW3bVFWWU1Vdjs/nxZSS3oRBImkQT5lIBLI0rCzMNYGha+ztTWJ1JwfNkWGopyGdngZZPQ2QtPYliSZNqmpKQEq2t8dYVhOmodxHWcBD0JmPGS1kw/p9vWxvHRKM3oTJn59vQaAEw3bEzqNr2FLS2BUnYdgknXhFCcNibkWQf28aKRqGJfnds830JkzOOaKa/oTBd+8fGc/z0qPrmBUJ0J8ce9iwmD2eW4EvAQWFB3gN8BXAFR6XovLYjk72tkf5/uVjO0bMhJd+cdsBookU5aVBZtVG2NpuIVETwk9ubTsoh5y2lESTVo5LGVBGqwKPB3weneA4NCzZuIVODZhEj2fnynV86PIv8nvvAuwDQyPdiWSKTZsbOfboJXg8OieuW4Hf7yNYWsKO1hgPPdFEV9RACGjX2gmWhjHnrBgmIELAWSuqhhkHDLmUUdZlXl1wTE3h4behvMTgav/RKGTCnJmbsSQgVQ9DNf7Ot1TfNsryT7fl4MR52pK09qcwLItkIo1lS8oryyivjmD5PLTFDFI9SdKmjUBQVuKjpqKEshIf61tjVNWoxaF/29jOmtllLKkO5Z0+k45VYcqwME2JJWB/b5J9XXFn9EuAEGxoj7FidhlL61W9ZhZwpkyTEmeeaFtblGhKmUInTJv7tnXmFYzfr2/mLy+0kjJtTlhQzhVrZiGAHz28d0To8NMWV9IzylzTHCf8QmnAw5UnzCHo1Qh4dAJejaBXJ+i4+KkIekbNB6bIO/UoZByHurgASjDg4A0DxoNpWhxo62FDVnjpWTUREmmLps4YA3Edy5IkLAtviTYuh5zZLmWShk0qHMSSki17+9mNxfF1Hry6wOfRmbYYY1JywT9v5aVjT2NPZTUPLDsJa3cL6fI5pIwuHntqIxs378E0LWpqypk3p5Z5c+uIpUxuffgAAPUR5ZBzcV2Y/+w1CZWGSSXSpOJD4Qh8wQD1VUPrwZ9uNdg/MDRh79ehPjz+i86IhkQJQj7ROL1BpylYTuOBPryWyapKodYWScmB3uRg4ysAhHTyA4RASADV05AJg4QpqWsow7ZseuJ9BNIGdTXlVFZF8Ho9pA2LeEo1oBVhHzUVISJh36B38V2dcfZ2xQfXL8UMm8f39aEJWFgVdIxBLJKmmrS3EdSV+qgtC9CZMIilbTa2x/MKxp+eb+Ffm9pVj8SxDFxaE+b9znPylxda6Irl99SdjZRw/Pxygl6NuRXK6k4IwTWnzceva4OiEfBqeHWNjc39eUWjIujhiDplNu/VNY6ZXdgL+EVH1vKH9c2jlmu6hWcZ0DPN53SZIibSy5hJDEOFl96wYyi8dDBQQk80zdYDnYMOOSmrVL7HPBr5QuVIW7mNybiUiadM4lkuZYQQ6CHVo8lM1JaMZiY1BQjb5rI/3cjJj9yJFII9x5yIbdukUyZIQVfPAF3tHSxZOIvquXN5qQO2dHdwwTG1hP0ezl5ZxezKwDCHnMogFSLVETqb0tiWjabrRKojPHjA4NLFPiSCqoBGyKOsysr84HM0J562GN0ejUHRGOwoZImGcIanQJLy+kmYktq6UmzLJmYlCNlpPEJzXLmMTUzz0hNT3SYhBLpHJ1lWQUNYozLgJZ4ykCmLoM/DnJoygkEvaALTloOi09SX5Il9fSNEw7IlT+/vp7EnSXN/CiMrQWXIy+cuWArAPY/uZXt7nqFLB8OSLK0JE/Dqqmfh1akOD80vvevEueiOZ4mAV+Nbd+8sKBiXHVM/YvvSmvwOZDKikX1dXl1w0ZHjD4+xdl45zzcP8JRtFez2HJTwCCF+kbPptUKIBXmS6sA8lJ+2fx3MOQ8WJ3TDV1CLWauAFuDvwJellK8IUTxcBONgSRsm+5qV4KQNi7KSIBYaO5oHlENOKQn5hxxyDmQZCgwTmbRFNKW+B62DNIFPE2rBXVZrl5pBn1rCtnj9H67nhMfv4oHz3sh/L76SdCpNR7MzB9PXBukEelkVjdSyc2+S0oDOsoaSwWHAxfUlJE2b/f3KKWZfllNMTdOoqKt0LNLK0TSNZWWSA71qQacP8OmALemPM2Su7YiGV9cxrJEDHl4NfB5tTNGIaV4GvKrBzAhGnx5GMyBsD3/7dzpP2AhsIZDOd8A26fMERwy7SAR7YpIyj8WcmlL2Rk2e7YhDz/A1kFeuqcewJNs74qTzXAtAwrRZt6Cc9oH0sB5FaZY59ZvXzkIIwfUP7Ka3gGC8ed3sEdszNESGrw8rhmDAkGhscQwMyoMeXrWihiMbSgf99tnOEKWUyvJS/W7CMe1XBagq9ZPsbd9b6DwH2+O5KutvCRzrfPIhUaERZsziTQixGHgc5UX7H8BW4HjgOuACIcQpUsqumSrfeHiliMbBkDZM9hzoZNPOFtJpE83noS9ps721A8uGgHe4Q07bhpRhY/mUBdGmfX3sbLQ4rla93WqawKsL5Trk0ArkiI0SSWyTN//2uxz/zL3851Vv4XcnX8Zj22LE4z3Eo0kwjUFTaGugm8pZc1m5oHLQAWVzX5K+NOyOgmEPXaRfG2paYKRFWnVwfD2NmOZl7rySYQ2+kJIKIzZCNEA1FlaOaPR6Q8gcYZdC0O0NE5MmNoIqI4ZX2kR1P73ekSHFIz2tWP78w38SWLmwmoBXx/IoR50aoAnwCIFHF/QnTSJBHxesqOF3zzXTnxw54VQR9HDKotHDSkScHuXFUygYF6yoYdWsMpKGhSVxrCQl1jhEo6wnQXnQy3tOmqe8j3sEHk1Tf2saPo/6WxdqzVTmowmBrsGLbTHMWF9vofIerPBkFggJYDdwPSoQXC4W0COlLNy3nB5uQonOtVLKH2Y2CiG+hxLErwHXTMWJXcGYemzbZsuuFjbubGYgnsaQgo7+9DCHnDiTyf3OhH80ZZJK20gcE2anaVTelw/u8di9t5Pu1sJm0BnRyJ7TwFnkpxoGOWilJCV0RNPOXIdaPa7mP9R+3Uzj7evh16e+ga/4VtL0twepP/NCoj3dqkHpbSU7s+7mA+yd1YCRhCMiUOcX2Egq/FDihRKP+k55/KxvKmzCXEiIJWAIbbho5KbJEQ1bCCqMOEHbJKl56fSV5Mu64Ak9WQN6ftskYsTRkEjTwkqn0QSUR0qIiczg4XBCXs3xAmFT4tWJVAYpC/mIBD2EA16CPn0wOB3AJUfVFW1YKiMYFUEPFxxZyzFzIqRN2zGzHi4atsxYQmZXvhwUjEjAw1UnzsXrWDAqkRhdNLTBv5VoeDSN8qCXUxaPLqCFGNOYZCwzxwmc6IvAA1LKh4uSYZFxejs7gUZgsZTSztpXihpyE0BtrkDWLjlKvuHbfwZgTsTPZ85eNK5zfuP+3RzoGxlCdiJ5FCufQ6ksxcqnUB4e06CivwtNCLweHdPxU5XrUsajqdX+Vl0t0jtyAaEw0gQ6O0Zsh5GiYdXWgm9kHn29MXxdHYNCkU80JFKNSklB+aI5eAIjnUR2t3ejdfcOmldn5pE8poE/nSQeKiU5EOWOfz4MQuPS972Jp7b2EB+I0d/eDQc2DV+DIzRWnnUmFSV+Kv0Q9AytmcmsMQzOnc2e7gS7dncP86+WQddg6dzSQdGwEUTMJKVWCkNotPrVWqiFdSXsaSvgyUFK/LaJkihJiZnCLy1MBEndiybVdk1KuoORQfPzbDyaoCE+0qTbNE2SyRQSQWl5hNKyErTKMrrjBi09ieHVIeD0BeWsnV8+6E7G79EKNqC3b2qnK27SF0vR3pfEtCQeXbCwJsx7T5w7OCSV6V3Y9pBoZH7re3d105fHRK88oHPRsmq8usCjCTXn6IiGEhDwaNqgYNz4+H5aBg69Z/LPn3gD7Ts35q3AYrrM+XKx8poiznK+784WHQAp5YAQ4jHgfJTHhfvyZaBrsLBybIeKGRZWBmkZSGFlnW2ieRQrn0OpLMXKZ26Zn+a+FDk/JjKVojtuKpcymhKIbJcykiHRkBJkKg0e7zAXMrZtY8eTxOPG4MrxQdHIOh7UYeFQgkC5B5G1TsaybLq6o1QY1qBQIFSjUWh4yk6mkD4v9z65D8OxZjLSxuC8Sijg5aQ6daTHNHj3z79MSXcH17ztWyyaE6HhqKPpMbz0xtWxodIw0d1bB9+SM2hC0tu0l+oT19GDoNsRjhIrRYWZQAL9hlrkOH/+SCeeUkp80gKU6xuvVOLglaoh1aVNVdoJsZDQ8egCM8+ciC5tao2RouRBUmIN72lVeiUdaUYIRqV3eL6GYap1OGhUVJZTUx0hUhaiLOSlIy2xHEu7jGB4dcExcyJccmyD4xFAiUbCsAfXv+SKRpnfQ0/cJBL2EwmrFwUNWFgZIGXYeHSBT9dGFY39AynWH+gnu1p0AasaSifU01hSHaQ9dmg8k/nyyEfRrdqcENjnALOBfP69pZTyq8U+7zjIBO3YXmD/DpTwLKOA8Fg2bGuP8X/37uK6U+fj92jcv7ObZw6MdML6qTMXcs6SKh5r7B2Rx+6uhAoNC9yxuZ0tOdYtJT6dD548D4C/bGhje0dsxA+pITh3SRW/eraJfb3JYfvqSvxctW4WALc+20xbNOU49xuZR2t/iq/eu2vY9sVVIV6/qg6AHz2+j2hWWN7M23ruNW1tjw3mc0xDKRccoXxwfevBPSPq5rg5Ec5ZUsUTe/vItq7P5HP/zi5OXVDBQMrkpif2jzj+6Jogvc0dPLm9FTF3uPt5S0riAwnshIHm8+Kvr0WgxnpNACmJtnWRjsbxBvyUzlbDIjrDhUBKiPXHCM+e5fQwhki0dmCn0njCQfzVWQ2EEMPfgDVBe2ec+ccMjxYSb2pDmiaeshJ8FZFBc2FLSmygL5YeFB3btulp61Ymv2096LOr2T8gOLBjHzf+97uc3Luba+eeya5+i0CwksgiP3VGCqJDjblIJ0bUoW1L+ju68QX8aJqywNMEBHQvdrsSHj2dKigYZiLJkvoS9JzemZ02SLSoYG8VtRVoPjWXUaNLWntzehlATSRAMDi0DshKJEm2q2nW0Jx6hD5kFRgEdEPS0T/Uw6gpC1DqFRBRnh5627voau+nrqGaygUN6JpAahr9GgwYUF/qp783OUwwANA1/vRSO0uqgiyvCWPaNvfs7Bl6QRBqYOvEeWWsnl1KedjLz58ZbjJsA2vmRFg3v5zWgRS/enakSfElK2tYWVfC3p4Eu7sSjKhaIThnSRUbW6Pc7niIzuZNx9YzvyLI5rYod25WvfF8zzYStrTF+Mo9w5/t954wh8qQl6f39/HAzuG9RMuWzvDd8GcyO5+PnJZp97p4Zv/IAKPvP2nuiOc6H0UVHiHEl4FP5+SbfSWZv2dCeCLO90iVGL69PN9OadskezvYtHkvc+YvYH+vileRNC1KfCNNZht71MNeX+qjuV+9uUkpiQQ8lPj1wf2WLUcc7/dog/szx6RMm36nWy6lZEVtCd0JNSmbe7xHGzq/Rxvany+PQUvV7GuVcvB4v2fkRKxHE7QMDL+mSNZ8SNqyB4/PLptj3EQ0bdLSn2RxVZCt7bFB4Qj7NPy6xv6eBA+aNvG0hWXZmLbENG1iSYOBaJJHHmmhr6uXsvJSFtSm8fp9CCGwbEl7Zww9liQZT+OxoCKRyrbRdercxkY19GbKmdgWgmhWoyaAEhP8qZHzG0nDwjQsvGkTkUg57lEgmTDojBqDjatpSyI1pTxwz1OsWHe08hAN+AwT27RIpAXp9viwHptXY9icQV9nH7bTqtiWRV9nHxuefoHbNt7GmdEDXHfMG3ns1Vdy8pxKNCGom1NKui2BagbVb1dz3KnDek1en5c1VTahygi+wPB+l21Z9BvqHgmbaapCQdqzrkmlsYlgkE4Z5HqLMw2DuHO8lkrjcS6mJOCjvjw4rH4rvBAUknQyU8fq72jSUKVPpND04fd2wKOzuKEMTQhsW2Kk0/T3x5C2xOPRqa2KcNqqedSVh9jWnUDXBMKJMyMEzIr48Xk0tnaoMASZ+7e2RA2T1pf5qS3zkzQsqsMjFw1rQjDgPENlfn3Y8xQJeDAsm709CXoTBqV5TOl7EwZ7exJ0xdKUBz2kreHP5MKKID0Jg554Ou/xXbH0YD7Z+3PzmVsRxJNnAq6lP8lAyiSWMvPmX1viY7PzIpy5pvLg0LN9oDeBR9dImzZleeZAf3Lzzeztksxad+6IfdkUc47nrcBvgPuBHwF/RXkzuBs4E3g38GfgZinlQ0U56cTK91PgvcB7pZQ/y7P/a8Bngc9KKb+Rva92yVHyLd/9C19+1ZJhDex46Esa/O9/djqNruTrFy2bcB7Fyudg8hgacpD0xE3+777dg/l8/PT5BL0ahoXjEt3xdGvapG2JZSmfU9bgeheVZyJt86/tXQjnbfucheUICSnDJJoy6YsZ9MXStHf0caCli/hAlLDfQ1lZCM0Z0hK6jlFbN9ig2ZZNtRXPaymVixz8CGLeAP2ewPDenJQEbAOPtAfnMCSCGiOKAHo8QaK6f8zQAul4An/Qj4bEYw8NLQ3ofkyhIW0bHYkuoCdqIsvLaGxU8zMD3QNIIw0djVCzALxertp+Nzdu/gs/fd21PHn8+QXP+3ibxMwz5OHR4OS68Znn5davkTboOtDJcbMCQ3WYZRgx6C0gs92xntI8HiqXzlcvCJbNM8/tY3FYTfFnPEnrmkBDoGkM+h/TEEiGXPFruo5/4Vy1CNS2sXbvYfmccs49fgkrF9Sg62Ovm3q5PJOHaj6ZPP40HXM8wPuBA8AFUkrTeYttlFL+EfijEOJvqDU8fyjiOSdCpkcTKbA/s703384T50cm9UNGAl5K/Tp9SRMdJpXHweaTEY2AR6PEr9OfNBFSYlk2HQMpTNvGtFV0ymzRMG3l4iOfaPh0QdK0MQ2Lxu44UuJ4uVXmx4MNhxB4vBpBMWQNJKUkbdpsaupnR3M/li3RBARNi9kRH4mkSTSeor1rgN7uXhKxOEG/l/LayLBhNQkMSJ2+nsRgr03TNXq0MBgxPNImrvuGTHKdSfDqdBQPkgE9QJ83a/w6z7BHUvOqyW1nglvLSuS3zcGJbw1JjyfE3n29eSfivVqCk+oElpT0GRA1IGokiRoQM+G4GjXBH5OShNPDifZEhyzSUjH1XT2X25afz4o1S1l/1Cmj/u4n1wmM2no6Y2lM0yYVT9CzfTvHrZg/eLmFRGPQus6wqPTrRBrKaGob4JEn9xDSIVHjVfNWmuNnLvOb66AJzfntNSUimfvGspC6TmtnFCElsyNBhJbpl6nymJbEtIbuNwCf10PQqzmyDwfiaUpDPub4JVddeVrBGEmFOBSeyUOxLMXKJ5PHaBRTeFYBf5BSZlsqDp5dSvlfIcR/gU8AdxbxvOMl46hqWYH9S53vEXNAUkouXF6Tu3ncVIV9dEbT+PTxPxwZsnsa5UEvndE0mpB0x9KDgZjSphzR08gVjcxz6dWUl2DLMHnhQP/gBHm2aIjM30LgzxGNDMmUyc62KKYt+dEortOllKRMW1mVJU364gZ9cYPGngTPtwwMjnHbEtY3D7CrA8qsNIm+fhKGSTocQYQjDCDoc3odVUYMv7RIaF56fSNXYEsh6PMEiZhJYrpvmGjoWXYlftsgYgyJRqGey5xU/tHZkG0QyupZ9euBvKIDkAn90pOEzb3qb68GpV6oCkA0YdBvS+oifvpN5d26pKJEWaRlQhHEuqGinqMWV/Ns7SljiwYQ8YIo8XP/g1uI7thAJOghuUTNOTmuwQZFQ2iga3lEwzJJStjX0sei+TWcOntIdCbiBShpWSRt2Nvch0eoMhppezDssgD8Xo1IyEfQp+PzaFiWTdq00YQk5NMoDehs7zUI+Dx88MIjBtfDTJSDeSaLnc+hVJZi5VMV9mGmEgWDUhVTeLxA9uLLBCN7FxuZonUy4+AB5/t8IYSWx5z6FCCOih00DMuSk1J+KeXg23zPQIpgmZ++hDEYxc9ynAYO62lYthrScETDzvgRkRBPm3T2JYkEPIOiIWBQKIQQaNrootETTbOnXa0Z+NHDeyYca0NKybP7+9jeOjA4F9GTMPnjcy3s6IhSFfJRXxogEtBp7k3y6J4e0pbEsNWbrGFLjptTxpbO+IiJVQn0GJDuHyBkmQTCIeI+36BJrS5tvM41A8qqSsq8gmEJjbCdJpxnjiaDX1r4LTUu3q8HsMTItzRdjmGek1UvEXPkJH4uET+srFDrZNKmyYHOJNs6E/QlTEqDHs45qpqWXTuR4dkjLdKkRI92UF8xVzmsHI9oaILHt7WTTJloqDUdsyIBEMqCajykhKCtO45h2HgEeecORsNG9WRsKWntjjGrKshcK44mBBVhHyG/jk/X8enqOpJpi3jaJJmyqSr1E/IJ/B6N2XURjloym3tv30ZvNDVp0QF1Db3RFCWRg4sQW4x8DqWyZPI5sqH0oPwnejRBz4HdBeOvFVN4WoDslXL7gKNz0swi/9qtKUdKuUsIcTfKcu2DwA+zdn8ZCKPmn0YscpVANGUOCYZjXmk4b2OG7YiFbWOYI0VDiY3NQNJk/T5l8SFhUDQyQ1LDRMMjCGqeYaLRNZAajLcxGdF4dl8vW3Nibdz2XAv7ehI0RAKDLtJnlwdYNasM07K55fF9JLJibiQNC6+u5fdRtVdZuaysCbGiNowlJQnTxqdrBL0aXt2LLkCaFrH0yPULzg9FuVeg+fwgLWalRlrOZPBIG13aByUYGSJmghdajGFDPBn2agyaMEsgZUoGTIilIWpC1JBUB2BeOEW2/VvuhH7SUMtTSz2w5cAA21vUrVYZ0pkfiFNfEaY04OXI5fOpjKVIh0toe7hzeGH6O6n0Q0lofI3LMBc+gz2cidTMxLAlgy8YdsZKSqDCLOsamibwe3VW1JeRHaEhlbbojRtIKSkP+1lQW4qQyrhkTl05KxY3UFVeMngdLvk5ZWn1tDjcPViKKTzPA9mhDe8HrhZCvB24HWVg8HrgsSKec6J8AOUy5wYhxDnAFuAE1Bqf7cD/5jtIInlmb6/TYI8uGiKPaCjBMOlLmJMSDFDxNnJF40/Pt9Dcl6Sm1J8VV8OmusTHac46gJsf3UtXzCBpKn9juZi25JFdPcO2nbKoglWzytA1gWVDiV+nIuTFq6nhkPUHCovB21bX49WHYqlcdlQthmnTH0vR3hOnP27Ql5T4NEjn0QZd2hOKvRIxE/T4RrpiKdT7yPYQkD085TVTWHb+8xo2tERtyv3qd1/fBZZUaYO6ClhWGRCUBjSU4fZIM+jq2dW8sKefU5dXUhHyIi1JJKDBQDdbtmxjbyyBd9Fsyo6cD+iE5s7m/r/8F49lYGQJq5SS9U7gtZnGyhIZ9SaiRCbk04mEPAR9Gj6Pjk/XEBrs1ZU/Nk1TBgSGY6lo2ZKSgI9lsyKUl/hJJlOkDYM5DZWsWNRARdlI1zcvNw4XwSgWxRSefwI3CSEWSin3AN8ErkBZtt3qpDGAzxXxnBPC6fWsY8hJ6EWontoPGMVJqG3DjQ8VTzBue24oot+W1gHaB9KqR2GqwExBr8alRyuPsrc+uZ/dXfG8omFYkod2dg/rfQQ8GsvrSgaFpzzkJeTTCXp1Ht+T9/IA+PwFSwl4NTyaIG3YdPar3tUp8yJEHdfwSNB12NKmEc8Trz7sjMuDapCi8TQdfQl6+pNIIODTKQt5SMSThOJR0v4SEEOvvaMJBuQXDa+VoqYiQnvMUivELZuIEUM30ySdxX6DhgdZcxrC8Sk1ODw1uNonP20pjSPrfOrlwmPh9wjKfcqH23BUHvnMoOO2QSxpUhHykujpZNOTG4knUjTUVXLWaccyd85wVyv9bR3DRAfU+pvxBF6bCiSqZzvgBPrSNEHYp1Ph9xDwOiLj0cYy8sO0bLoGDAJejQV1ZVSXBQj6dLr7YvQPxJk/q4rli+ooLz08BOeVJhrFoJieC25lSGCQUu4XQhwHfBxYjHJVc5OUckOxzjkZpJT7gXdO9LiMYHTF0iyoCrG0JowQgm1tUfZ0xQdFI2FYWLbkPc4C0D8/3zJiWMq0Jf/e1M7aeeU8saeXjS0DgFozE/Rq1JUOLWybUxEk7PfwRAHRsCV87oIlBL3KxUdub+GKNbMG/97SOpDXdXpZwEN3f5LehDEYf0RKnDgyGuWh4YG8jptTxiONvcOuS9cEa2aV0B9L09mXoKs/gW074XpDPmwJA9EY+9q7SSbSeH0eyhD0B0qHBMOMoxkpkuDMa+URDTFSNLxBLwNpm8a9fbTua6fML3j1mrqh9RvOm3hGI6SURA0nYFlc0pO0SZqjLys4a6538Pg5pfktdqSU+DyC3p4o6cRw1yPpRJLjl0aoLVXzEkIIKipKOe/sdcxuqB5Wv950khRwzFveQNvOFjY/+RQeDd775nNGLWOxsGwwnWFjCSpoGpLFDSWsDpsEfPrgqvzxdE5tyyaWsjCDaqjR79VZtyRCadCHLW16euP0D9gsnlvNsgV1lJVMbMX9ZHEFY+YoZujr04F+KeULmW1Oz+dDxTrHTGPakru2qDH3b7xmOX6PYEtblId3dg+KRiZ+hpQSIURB1+kZAbhiTQNvErMIeEeKBsC5jgeArQVEoyLooTI00kdYLoZlc9bSKv6+oW24YAhYVhmgYyDliIx3uMmyM4piD7qKkcwq8zO/KkRjZ1xNfNuSBWV+Wlv7aLJtPLqG3+fFoyvz4Y6eAXo6ejAMg4DfR3kkrFbKaxaGR9B4oJ+AbXJUgwcR8DurxZWLGS1HNPJx5wutpJ0eWFldFQD/3G3g1+HihT4GDBUNc3aJclWysctie4+zyFFAxCeoDgoGjMLiU2hCXUpJ50Ca7a0xdrTGKC0LkhhIDEXqNA3oaETWLOCRLV2ku9tZt/oIliyazdLFc0bkV9XRzPtu+CT/fuMH2HP0q0j4Dm6iOJvqpStHOCy1bDBsFREzg64JQn4PVSU6Aa9Oh19HF4KAT6e6bHzxhaQtB33j6ZpGfWWIuObFo6UI+JTros7uASwpWTKvlmUL6igJ5XN0MpJTllaP/6JdDkmKOdT2AHAzah7lZc0HT58/2BC9+shaXrOqruC8RKEwsBXOauCwf3w/wUTibRim6nnFkwY9cZPeeJpESg1FzakIsr87oQRDStbMK2deRRBbSqfXxqDdmJrHUpPRHk2gaxoer+pl1JT66RhIsXtvDyFd0FAfpLSiBI+uOYv+bPr7YrS2diENg4aIH18eb8OtTr1pTmybyZDOM+wHKkTyP3alB63nSn1eKgKC2SUapV5BeUBQ5hODv92e/pEOEkdja3OU5xr76IkZaALmVQdZtKSWlv4UWzc356zBaaGqqox5cxYA+b331rTt55offAKPadA0r5DV/+SYPb+Wc2d7SBiW4wJHmUp6dEHY7yFc6sHvVZ4jPDneKjrHOecmpXLGmkiZaEJQEwlQXxGmPOxD1wVN7RZHzy5llSdOZ3eUpQvqWDq/lnBwfILj8vKhmMLTiTKhfllTEfSwuHpo7YhnDBOhyQZoysSDzwRdWllfyrL6Ura1qvmiSMDDmUurmF0eZG9njLhhMRA3GEiYpC3lnEVgE/DqhHw6VaV+dA26bOiNG+zZ10ttiY8zllQ6frqcFePDVo6LwVXlAImUSWd/kgPtA6RMMbhoNODVmVWpxuMty6and4CWti7SaYNgwE8omD/a4cFg2ZL+tBouG42FEZ1yv6DcLyj1qWupDGhU5ulI+LwaacMeYY2WWQsXS5nsbIuzvCGM36uTNm1CPp1j5pWxuC5EwKuTqqxE9KTYt6+bWHfv4BocEevhkkuPJ1zAGq2uZS/v+8EnEEh+fN13OLBwOfRMTAizMW0wTAtpSxbMiYBhYFlxSgJqzi/g1fF5xJj373hIpk3izotNZWmAxfVllJf4B+f7QIUdTzuudI5YWsfiubWEgmP31F1enhRTeB4ETi5ifocc4421kS0aR80q43LT5vaX2rBstR7orGWVLKgK0x03nHUoqJnbzJulBCHkUOAlXUPXBQ3lQR5+oRmvJvjEG46iP2nS1ZtEd3ol9aV+FlWH8Hv0EaKRYfeAMWiN59XFmMN0KcOiuz9JU1eUvlgaBJQEvBwR1mnOGv+yLIue3gGaW7swDZNg0E+wdHTB+efuFClLNa5x4K871N9+HV69aOgt2LKVwHg1KPNrRA3J3Y3pMdwQKo6pGf8tfv6aBnZ2JHjwfhVCoPdAK287Yz77upL8Y30PTd3KSKLEr7O4LsyquaUcPW8o9nw0muDZ554gsmwZtQ0V7NmzfdCVshDwbAFrtFC0j/df/z/YmsaPr/s27fXjn3fIXu1v2k50VFT47tKgl6RXmTAHfDrLqstGz2wUltWGmG8NrQdMG8oizZZQFvKyfE4pFSV+Ajl+Aw3Toqs3hiYE4WApQb+XVctGDjG6vLIopvB8DnhKCPFV4CtSyrGdZR1GRAIezl5WxcLqEF0xw3GPPj7ROHlRJc839aMh+Mz5S9AzYiIyQZgYiuLniIJhSaIpk/6EQddAio5YmmOrgvxFqJ6GRxPMjgQm/MZ6/qIK9rZHaRwljWHa9ESTNHfF6OpXnq9Dfg9VZfnf1g3DZOOWRizTIhT0E8oTTyYfeQz1Brfv7DXpSSrBGUir+aWFEY01tRphDyyryPRkNP67t/Ai0Yng7+5kywZnDbQQWGj86pEmAMqCHtYuirCsPkxF2OckUb95f3+M9S/uYOv2vSBhZbgUWysZ8jiAmiPbsn0f69YsH9HriZdEuOfCt7JtxTo66+YMlkUzh6cr7FJGpyzoJeT34PeoQF/64EtBnMVVcDCLXzKCY1o20YSJZduE/B4WN5RRVRYglGe4OG2YdPXF8eoaq5bOZtHcal7vm/yCz2yKYRBQLKOCl1tZipXPR06bP2qo6WIKz2dQngk+C7xbCPEi0Epe58fy3UU875Tj92j8v9euUIGZdG1U0chsyx3DD/tUVc+vGm4iKqUknraIp5Q7ma5omq5oGtNWw2WagIBXp8TvwRPSBiM+lgaK8xBnsCybvlialu4Ybb1xpA0Bv05lqT/vfIQaOjGxLJukYeL3evEUcax+R48K11vh15hVojlrZRzHoEJwVHVxI3rYtuTxHd20dw4MbXR8jq1eUMYJiytG1IOUkgcffYEt2/YhhGDlEQtYfcwSykrD/OGu50bc+VLKYb2euY1bQQj2zz+Cx8587fDyAAtFgjl1gn1CvePEktYwlzJ+r45fF2gH6SJlNCzLJpY0MSwbv1dnXk0JNZEA4YA3r0XbkODorF4+lwWzq/B5ix59xeUwp5h3xFVZf9c7n3xIlKfqwwZNwNLaCYThLYCUkmjSJJ4y6XV6Mj0xA9O2By24Aj6dSMibd5hsKuiNpmjrjdPSFcOyJT6vRnnYj1bg/IZh0tXdxx9e7CHtLLiMI/jHHhMwRwyTZUhZkpghB8XjyZbRO8TnzvPlWSOTH7+ev/c0hp9CADr6U2xpjrKzLUYiz4pWW8LmpignLhmKu9M/EKOsNOyYdWusOnIhq49eSkl4yAw43ttLrvJkr8FZsGsj77npf+monc33P/mjQd96GV9vAkHAp1ER9tFQEUBDsGJW6ZSKTHY54ykVSM/j0WioCFFbHqQ06BvmbSCbVNqkuy+O3+dh3cr5zGuoxDtJYxGXlz/FFJ6FRczrkMI0La6++mouueQSLrnkknEdY9uSmDPp2hNL0xdP89y2Ti7b2cWHz1+C7lhxlYe8BRv5qUA6XqltW9IXS/Psjja8ulprM5rYpQ2Tzs5e2tp7kEDazt8CZQSgJ2nTGrPVWpmUTdxUJtGvWexTcWNCGk3Rwm5txis6MCR0Dx1Is3tvJ+V+McJsOJvuaHpQ3He2xdncNMCC6hBBn8bW5tiw8MoeTXDSkgoAOjp7efb5bexubOH1l55BXW0Fp59yTN5zLDn1VHbv7cTct4XqUh+Xvfo0QAnZvK0v8J6bP09vpIobr/wC8bRN0KdTFlSGID6PjlfXBhv5jAXlVIqOlJJEyiSRttA1QU0kSH1liEjQlzVsN5JkyqC7L04o6OP4VfOZW1+Jx+MKziuZO++8kzvvvBMKRwIo6gLSvcXK61DD49H56U9/WnC/5axZiKVMemNpOgfSyu+U88br0ZRvM+FM6NcWmCuZSmJJg66+BPs7Y/QnVDhnTRNUl42+WC+dNujo7KOtoxshIRQOoGsaMLrFVWvMZnO3RYlXUBXQWBxQ8zEZFkZ0nmsvvts+27Zp7ojxp3818saLTxrcPpAw2dEWY3tLlK6owatX1zK/OsTq+WWsWRDB71VlSxg2u9pUkDCkzYKaEiq8Bv/67xM07mvD5/OwbvURRMrGYanneFiwbIgmTSSwYsfzXPPzL9NbXc9tn/4+NdW1zHZcyhTimreddxA1MkrxpCRlWMRSJkIKqssCLJkVpDzsx5snAGA2iWSanv4E4aCfk45ZyOy6CldwXAAGX9BvueWWvkJp3MHXCWJaKjJmNGnSE03TGUvRHzcHXbl4PWoRaVWpb9janukaOssmmTbp6k/S1BmjP5FGE4KSgAevriGcUL6FSKUNOjp6aO/oQQiNcNBP0tawZFasi1FYXK6zpFwftedyMENk+bBtm1TKBN1Ht1Yz6MT1vy910NKrhLIu4ue0IyqpLVO9pFwrrHOOrGZ3+z6krbxhn7asnN/ddjdCCE5Yt4JVKxfi94+0BBxc7W9JpJSsqdboaBb0CBW7aF51CJ9H58K/3k9/wxzu+PwP0CKVeWPDTzUpx/uzbUsiYR8r51RSWerDN46hsVgiRV80SVk4yClrFjOrphx9Kr2OurwsKbrwCCFKgMuA1aiuVh/KgejfpJQF4zMcykgJG/b30TWQYiCp5iYk4NOVyFSX+iYUiGoqSZsWPQNJDnTG6I2qxjYc8FDt9LJuvL9x0DP0QNzgW3epWOphn86Hzl5AMpWmvaOXlvZeugxBUvrUcFmbiWHDcXUe5o1j9fp4YnlMdIhsLA40O45cHYu0+zd1cv6qGnRNcMLicpbWh4mERjfK8GgCj5XESJmI7r1Iew4Xv+okqqsig5Pk2S5lbCkH/b+FfDqVYc+gS5lHt3XTJwRej0aZX0NqGvd+6PN4kwlSJZM3bZ4Mgw45paTE72NpQxmVZUGCecK25yMWV4JTXhbi1DVLmFUTGYwC6+IyUYoqPEKINwA/AcoZbr8pgeuFEO+TUv6lmOecDqIJg4///gU++5oVVBew8ppJTMumN5qipTtGR28SiSTo9+S1SCsUjiCWtvjb03vxWylmhTWCoQAbGg00YRPxCeaUaJQHNKqCh15jY9mSx7Z30z+QHNooNBo7EmxtjnLp2kJ2LkNIKdm7v41nn9+G0a784pnAM+u3csKJq0hbkpRlDkbbzHYp4/NoePTCfsvOb97EGz/ze/7xv98nWVZOqqS4FomFMB2LNNOW+D3KIWdVaYCS4Pgf+4FYkoFYkspImNPXLaW+uswVHJeDppi+2s5DhbW2gV+jFpS2oqzbzgLeAvxBCNErpby3WOedLjyaWqE/Wd574ly+0lI4nMB4aXCGiCxb0h93zJ974ti2xO/VKc8Z4ssmX6yZbHb2GCwt91IS9iKl5Jx53mEuZbIp9jDZRJGDCzMFj2zrZtOBgRFpTFvyxM4eVswuHTWvgWic/9zzFB2dfZSEg4Qb5hBrOQDA1h37OOG4lTRUhAu6lBmN17dv4tvb/0n7siOxPFMvOBmHnGnTwuvRmV0VpiYSoDToG5dDzwz90QTReJrqihLWHbWA2srSQ+6Fy+XwpZg9ni+gZpxPk1I+l7PvV0KIG4GHnXSHnfDMNNf8Yj19iSET5Lf9+CkAgl6ND521IK9lXHt/iua+FG396tM+MPpCy0sX+wdFRgjlZqYQmWGyXz7WDMA7T5lVMG2xyHbIubM1xkWra6kp9XPUnFJ0ARsORIecczLcIm1YPoBhSrp7+wmXhJGaB4/Hy+mnHMOqlQu59W9PDEu/ZdMOll9w/ITLe71nC2dvu5P1VfN57n+/jxGYGjf/0jF/TmY55KyNBCkLeke1SBuRj5T0R5PEEmnqqso4/uhF1FSUuILjUnSKKTyrgdvyiA4AUspnhRB/QgWDO6zwTMPaiUJIKYklzWGik03CsLGkpKVXiUs0aXL6MuWh+cHtXezpTOD3aNSX+Vk7r4ynGwsamkwoANt0kjJsXtrfz47W2JBDzqrg4DKZ6lIfpy2vorHXGBpucyzSls8uxci4lLEkUtrs3tPExo07MU2LD151MaGglxVXno9HF/z/9u48vKkqfeD49yRp0n2jO7SlZasgyC4KCkoRcawrKD9HBR1FXGaUYXR0nFFQcB13Z0TRERx1RBgUUZQRFHFXVFS2ikDZ2gLd17RZzu+Pm5a2pG3SbF3O53nyQJO7nCZp3tx73/O+FVW11BQfbdy3zWZny097yZ4wlAg3yvUP+GID2c8t5su4DH4/9v+Y7eWg01iQs96GDoiLDGZQ7PGCnO5uq7yylhqzheSEKE4b3o9e0WEq4Cg+483AU4fWVK0t+bSXh6sAxwtyHi6qpNrcdtrxExv2NTaaCzXqGd8/Fr1OcNagXpwzWEdksJ6amjryC4va3E4g2e12jpbbKK6qp1e40ZGebiMh0oROBz/klRMfaeSUtF70Swg9IRsN4PIxCbz08QHsdolO2hnbL5pqsxVjkJ4wo55f9x7kqy27KC2vJjkhhsnjTyY2ytQs4G747OcTSm3YpWTDZz9zsRtHPflZp7At+yJuNg6hXu+9U2yuFOR0lZSS0ooa6uqt9E6IZvzIJu2lFcWHvBl4PgXGt7PMeLTTbYoT9Rablv7sKMhpsdmpqrdTUtN24DktM4bESBNJkSYigvWN31Tjwo1UVdWye08BlVU1BBuNAb8240xDGrS0S1Z8vp/kXuEUlNYRGx7EzNN6E6TXMeuMPpiaXGOz06RumU2bJCWA2PAgikqrSJBlZKUMxegoKbN7XyHvb/yO1OReXHjOaE7q39vpN/oDh4saC3s2sNns7D/sWtDu+/3n7D/lVKpj49l0/e1c68kT49C0IGdESOsFOV1ltx8POGkpPae9tNJ5eDPw/Bn4UgjxEHC/lLK64QEhRBhwL3Ay3byCtTMtr8/83z++AiAqJIhnrh5BaZWZXwsq2F1YSXGVhUFJ4aTEBJN7pJo1W4+0u/0zBsQ2+1lKSUVlNfmFxdRUmzEZg4iO1L7Jnp+pLfPJIe16z8Q+gS9N3zQNWko9RyvqGZURxYAkbZKmHRA6QXW9rVlJmRCjnsgwIyEmPUa9ll32aW4xVXmHCY4y8c0PuVhtNs4+/WT6903kxiun0Dc1vs1TSLf97jyWbNhH/s5tpMQEuzV5c+Tb/2b868+x6dr5/HzupZ48Jc0KcoaYDGQmay2inRXkdJXdLikpr6beYiMzNY5BfZOIivBPt09FaarD72IhxL+c3P0TcDswRwjxPXAESARGos3p2QzcQRer1eap1q7PlNda+NvKnyiuslDVJM05LtJE79gQ0mJDmDEqmcRII89+3H5hCLvdEXAKiqitrSPYFNTmDPu9+4soKfRs7kxH2O2Sw6VmfimspleYgcqqJmdfdTqQWm+cIIOOSrMNIWizpEyzbdusWMxm8ioryDt0jJMG9G7sBpuR1n5Li44as+plxr25lNzxU9g25cIObaNpQU6jQSvIGRcVTHgrBTld3q7dTklZDVabjX6p8X5tL60oznhyxDO7jceigbOd3D8ROJMeEniklJRUtZ1JVmm2kRobQmKUiaRII4mRpsa07VCjnsx47RRImFHvdA5OmFGP3W6nvKKa/IJias11hAYbXSvp4kdSSo5U1LO7oIrdjoKcRr1gLzTLRAMtDXprXgWTh8RjNGiTMdsqKdPg510HOLD1e+w27dTkkEF9mHXpRLfHOjc7gyWFv7q2sJSMW/ECY1YvZ+fE89h4411IneunwFoW5EyKCSWxnYKcrrLZ7JSUVXeovbSi+JIngafbFgVtqc5iZ2d+ZbNTZEuuHdVsGSklxyrr2HesGoFgbL9Y7FLyp//82Oa2rz8zzaUx3HJ2XwBe/1rrD3PFqb2x2+2UllWxfVce9XUWQoKNRHeygGOx2QnSa0cu//2mAJ2APrEhZCSEkhoXQn5xLZ/mlmCxHQ8+QXrBb0YkEhHS/kX5qmozVpud6MhQ4mIjMIWFU1tRBsAvewqorKp1KxvNXZHHCjjlvTfZNvkCPr7+DlyJFk0LcuqEICE6hKSYUKJC3c9Ic8Zms1NcVo2UUrWXVjqlDgee7lwUtD1NT529+0M+W/eXkVdUTbXjqn3fuFCSwg0cKqpmfGY0G3NLWttUh0igqLic/MIiLBYbIQE8wmlIDGjIRgMor7HyS2EVvxZWE2zUkX1yPAa94DfDE+ifGEZUmBGjXjuVNiw1iqIqCz8fqND66Uk7g3tHM6bfifNvmqqoqmXz1zv58vtfGDygD7+9aALJCTEYgk3gmKfbkWw0l0ktoaEiIYUVD/+LsqTUNoNOy4KcsZEm+qdEuVSQ01VWq43i8hoEkJWZRGafeNVeWumUVJFQDx0sqaHWYmNMRixx4UEECS1D7NeCCsKDDYzOiPFa4LHZbFgsVuotVvYfrHKr26cvWGx26uusSAlrvz/CyL5R/FJQxZEK7fRicrSJU9KiyEqOQK/XcVJv59u5bFxvdhyqxGqzo8fOjHGtLAiUVVSz6csdfLP1V+x2yfAhfTnr9CGAY7b9Mc/n4LTLbmfSS49RnJrJz+deSllK6x0bTyzIGUNMuAlTBzPSnGnaXnpIvxQyU+MINvmnLI+idIQKPB1ktdmxS8klI5PJL66hpNKMEFqL6GBj86e1reszLu3LaqO0rIL8gmLMdaDXC6Ii/H+EY7WDxWbDZpdIKfh4xzEaqvDUWezkFlQhJUwdlsDw9Ch6Rbj2bdto0JEUbeLw0QqSdGVtzkn57NtcvvphN6OGZnL26UPoFXO8HI635uC0RdhtnPX8wwz5+F22XHSV02WsVjtVjoKcYaYgrSBnRAghXs5Zd9Ze2uSl9tKK4kueZrVJ4C9SyiOtZLk50+VaXzuz80AJx8prkZJWC3I2cHZ9xhVWq43ikgoKCouw2yWhISYMBudFPr1JolVgtlhtWO0S4bjPaNARYjRwpMzMF7tLyS89XpTTapeUVlu4cFQSY/vHtrbpVhkNOkzlhzDGNO9VdLS4go+/2M7Ik/syICOZSeMGM370QGKiTpzo6OkcnPYIm5Xsfy4m69P1fDP9Wr6ecfxt3FiQ09EiOj1BS38OCzZ4lJHmTL3FSnFZNUFBBtVeWul0fN0Ibjba59HDaGnTs11cr8u1vnamvLqe6HCTT8rMNLSXLjhSgpSSsBATen3Dt2Wb19Ogm5aUafh1jAYdESFBhBkNmBypzdV1Vp5Yt4eaVipcW2ySD3482qHA01Lh0TI2frGNn3bsx2DQk9a7FwMykgkPa72J3m2/Ow+AJa9+CHi5gZqUnPPMQgZ+sZEvZ85hyyWzHQU5rdRbbRj0elJij7eI9kW1GXOdhdKKWq299JC+qr200in5uhFcQ1bb4RY/d3thRj3hLmRcuatle2kt4HivBL3k+Gx/rVGZVrFaAlGhRkKMekxBem22vw4Kysx8vrsYnRBMG55IZIiBUZnR9EsMo6Kmnne/P0J9i2y0acMTOz4+u538I6WsWPsF3/28D6PRwMTTBnPm2JPaDDi+cELQEoIj/YdwNGMQn51zOeYKs1aQMyaEhOhQtwtyukO1l1a6G69ltXXnLDeDTpAaE+zyKTJ3NWsvDYSGNrSX7rjWSsqYjDqiQo2EmvTohY5jlXVEBBtIitY+2Iur6vkir5wf8so5WlGHTsCwNO2IWQhBzsjjvW1+PVLjdjZaa8xVlVjq6rBbLZSUV5M9YSgTxgwiNMBpwPr6OqIP7+dgSiYbJ16sFeQMNTIoVsvO82UB2Zbtpfskxapun0q3oE4Mu8JHny11jm6fx4q09tLhocEdarKlBRmt7XJ7JWUaNl9vtXO0og6rXfLSpv3MP68/wUY9n+cW81luCRnxoVw8JplhqZGEBTt/m7iTjdaavQeOsPGzbeTnFTbedyi/mCsvmhDwoGOrqmHaY3fSZ+9OXnl6Jf36pnS4IKc7VHtppbvzdskcV3SL5AJPmBsDThk6nSA8LNRpPx1nzuhtpLhAYLNDpdkKCLdKyjT4zxeHsDqCVHmNleWbD3BDdgZnnhTHGVm9iAlrPyPNnWy0pqSU7M4rZONn29h38CjhYcGYwsKpq9Y6o/t0/k076i12auos6GpquOqffyN15w9sv+t+sob38/m+VXtppafwVcmctnSL5IKOsNvt1Fus7Ni5D51eT2R4SJsFK23y+JEMUrsWI3RaoAnSCdLjQt0qKQNQZbby0sf7OdwkIw3gQHEt3+wpZaybp8pay0ZriwTe3fA9teY6LpgyipP69+aRJWsbH/fZ/JtWaAU5LdjskhCjgQGRes556h5idm0ld8HDlJzXsdprrlLtpZWeRpXM8YPa2joKj5ZQXVPHgUPFFJt0XDymeUaazQ5Wu+N0mdRSmHU6QZhRT0yYgZAgHUEGRwXmncUALpWUsdsle45WY663MzQtkhCjnoIy8wnLWWySD7YecTvwuMIuJdtzD/L5ll+YNf1MQoKNXH3pmURHhmIw6Fn9wTc+n3/Tks2RkWaxNhTkjGgsyJn+4rPEbNvKrvsf49g55/lk/6DaSys9lyqZ40PVNWaOHCmhpKySIIOes9NNvF2kQ0owW21YrSAdH7kGnXZNJjbMQLBRj8mgw6DXdSgtV0rJweJatu4v58f9FVSarSRHmxiaFoleJ7hkTDLvfFfo1Yw0Z+x2Oz/uPMBHn2/jSFE5vWIiKC2vIiQ4lrjY4xM/fT3/5vh4HAU5LTYM+tYLch685gbKxp5OxSkjvbp/ON5eusZsISE2QrWXVnoklVzgA1XVtRQeKaa8vBq93oAx2ITdDlV12qx/gSAkyEBYhIFggw6TQYe+g0HGmXe+K+TzX0rQ6wQn9Q5nRHo0WSnHJ1yO7R/LL4XVXstIc6bWXM8zyz6gqKSSxLgo/u/C0xl2UrrTbD1fzr9xtSCnoayE/g/fx575d2OJi/d60FHtpRXlOBV4vERKSVllLQcPH6O8ogqDIQiTyYjeoCPMFESYSY8pSEeoUZvJnhbX8Y6P9VY7R8rrKCwzE2zU8+N+Lf35itP7kBBl4pT0KFJigjk5VTu15ow3MtJaslpt7D9cRL/0REKCjWT1S6HvxAROzkr1yUTb1tzw22zqHN1cBe0X5AwqLmLYTbMJPnyAgun/R3lcvNfG0rK99IRRKcRGda4K4orib14PPEKIZGAy0Btwlg8rpZT3e3u//iSlxGqHOouNOquN2upaCo+UYKmrIyoimEGpcdpETIPA0CIN1tPP33qrnYIyMzY7PP3B3sbMtNReIdQ6Kgr0jQ+lb3zbga2jGWnO2G02Dhwu4sF/rKG61sxdN19EVEQoF0wZ3eFtdkS9xUa12YpdSqJCjQxObb8gp/HYEYbdNBtTYQHbn3ie8lGnemUsDe2l6y020pJjycpMUu2lFcXBq4FHCLEQuLPFdhtKfTX9f5cKPBEmPReckkhpVX1j4DAZdOis9dSXlGGtqyMzPpSIMO9fmG9Qb7VTXmNh/U9Hsdm1+6x2SWKUiVlnphIX4f6cl45kpDUbU72VsoLD1NdoXc4lkmsvO8uv3S21gpxai+jQYAMDUlwvyGk8UsgpN15NUPExfn76RSpGeB4oVXtpRWmf1wKPEOK3wN+Aj4B/AP8FlgH/AyahpVCvBJ731j79RScEr3y4E71OcP+VY6mprubA4WNU1ZgJCzERFx3tk/3a7JJfCqrYur+c7YcqCTfpqDI3r5NWUlXP3qM1HQo8nqquNVN66HiOSV2dheSEaJ9ftzixIGcEcZEmwtxsES2DgqiPjmXXgoepHDbCozE1bS/dPzWBAX0TVHtpRWmFN494bgQOAedKKa2OD588KeUbwBtCiLeA94D/eHGffhFi1BMcpMNisbEjdx81tfWEhZqaleT3ts9yi9mw7Rg1dTZCjHpG9I3ix/3lzTLRwLdp0C3V1Nbx2be5HCup4LcXTSAmKpxTBvdl6/Y8wLcp0Ha71ApyWo4X5IyPCiEy1P2CnKbCAurj4rDE9uLHl/7j0fnPhvbSdiT9UlV7aUVxhTcDz1DgP1JKa5P7Gs93SCnXCyHWA7cDa1uu3JlZrDYqqszY7XaEEPSKObEkv6uaJgY01EeTUlJQVsfWvDImnhRHWLCBkCA9A5LCGJEezcDkMAx6Ham9QnhnS4HP06Bbqqo28+k3O/niu1+oq7cyZGAfrFYbNeZ6tuUebFzO2xM/pSP92Wyxo9cJEqNDSIzxrCBnSN5eht00i5Lxk9h99/0dDjpN20sPykiiX1q8ai+tKC7yZuAJAoqb/FzLif0YtgFzvbhPv6goK6Wm8CDhEVEedXast9opLNPqo738yQFmT0xjx6FKtu4v50i5VpCzb0IYg3tHMCozmlGZ0c3WH9svhl8KqhrToA06weDeEV5Ng25p974Clq36BKvFxrCT0jl7/BCSE7T9bfjsZ2SL+TfeOOqprbNSU2dFJwRxkcEMjA0lOszkcUHO0D27GXbTbEBy+PIrO7SNlu2l+6XGExKs2ksrSgNf9+NpqQBoOh3/ADCsxTIpgJUuxhQSTlxKmsfbefOrw9gcWWiVtVpvG9Cy0C4ek8zQ1EjCWynI2aAxDdouiQgxeCUNuqWyimoqq82kJveiT3IvRgzJ4MyxWSTENX8fHThchK0h08GhoxM/6+ptWBzbMhn0ZCZFerUgZ9gvuxh682ykwcBP/3yF2gz3aq81tJfW61R7aUVpi6/78bT0A3Byk58/AuYIIa4CVqMlGEwHPvfiPrsEs8XG2u8K+flgRWN6n9Uu0esEU4fFM2mw6/NGmqZBh1UexWgY6LVxFpdW8vGX2/nup30kJURz67XTCAk2Mv085ynGnk78bCjIabNLIkKM3Pf7i4iNMBHsYktwVwmrhcG334w0mvjxueWY0/q6Mcbj7aVPGdSHvr17qfbSiuIhbwaed4F/CiEypJT7gIeAy9Ey25Y5lrEAf/XiPju13YVVfLW7lJ2HKxvn2zRls0s27yx2K/CA52nQTUm7nUMFJby6ejPbcg+h0wnGDu/HpNOGeLxtZ1oW5MxIiqRXRHCrrRe8QRqC2LXoMSyxvTD3TnVpnbp6KyXlqr20oviC1/6SpJTLOB5gkFIeFEKMAeYD/YA84J9Syp+9tc/OpqEgZ0Z8KAa9jrxjNew9Ws3Y/jGYDILPcku0StMO/kgMaMsNk/uyaMf3VFqs/Jx7kAljsjjz1JOIivDuREe7zU5VnRWLTWLU60iNDyc+KoRwN9Of3RW5dQvhO7eT/3+zqBw63KV1GtpLB5tUe2lF8RWffoVzHPnc4st9BJqUkkMlZn7IK2ssyHn1GamcnBrJmVm9OHtIPHpHr53iKotfEwNac6igmI2fbyMlMZaqKq1StV6vZ9K4wV5rQ+CsIGeCI/3ZHxX/o777mpPnzaUuIZHCiy/DHtz279W0vfSpw/rSJzFGtZdWFB/xKPAIIYzAZ0Al2vwdSxvLvQ+EAWe0tlxn5iwNuqLWwnMf5lFcVa8V5EwJZ3jfKAYla+nWphbflP2RGNCW/YeL2PjZz+zak0+wKYjqanPjNSfphWy0lgU546ODSY4JO6Egp69Ff/05Q+bfhDmlDz/9c1mbQafGXE9ZZS3hqr20oviNp0c8VwKjgJy2gomUsl4I8SiwDvgtTU7JdQUS2ZgG/dyGfZwxKI7sofFEBBtIjwvh7CFxbRbkbNCQGHCkvI5rJqb5vIVyU+999D2ffLWT0BATUyeewtBBqTz5r/cbH/dkDo7dLrHZJaVVdcRGBNMvOYqYcOcFOX0t5vNPGHLHLdSkZ/LzP17GEhPrdLlm7aVHqPbSiuJPngaeS4C9Usp17S0opfxACLEbmEEXCzxlNTb0juSA2no7X+4uJntoPEIIZp7ex61tGQ3aRNCGoyZfaWgvnRwfTUR4CFn9ehMeGsy4kQMwGYO05msezMFpWpBTCEFYsJ7Ts5LaLMjpD8FHCqjuN5Cfn3kJa1T0CY9X1dRR4WgvfcaoASSrbp+K4neeBp4RaEcxrtoM+K6lo49Y7ZKmH6dmi71DbaL9QUrJzl8P89Hn2ziQX8zk8SczdeIp9EtPpF/68USGjszBaVmQs39yJL0iQ3j/f1p6cSCDjqGsFGt0DAWXzKTwgkuRhuYpz1p76Tp6Rav20ooSaJ4GnjjgiBvLHwF6ebhPtwkhBqAdnU0FBgCJQCnwFfCklPJjd7bnz/po7tiWe5ANn/1M/pFSYqLCuGTaWEYPzXS6rKtzcGw2Ldi0VZDz0dsv9f4v44b4D9Yy4KF7+ekfy6gaMqxZ0FHtpRWl8/E08NQC7hQuCwfMHu6zI+5Hm1O0A+0IrQQYBFwAXCCEuFVK+XRrK7f8mPIkDTp/5zbtP9kZHVofYG52BksKfwW0I5yGD9Ifd+yn3mLlsvPHMWJIRoevWbRWkLNli+jOIHHtagbe/xfKR46hxlGNQEpJeVUtNbUWkuIiOXVYJnGqvbSidBqeBp6DgDtNTEajldLxtw+Ah6WUPzS9UwgxEfgQeFQIsVJKWeBsZaNBNDYSCmQadFMWq5XCo+U88tw7zJo+kaSEaC4+dwzBpqAOnULyRUFOX0tavYKBD95D6anj2f73f2AzBVNeUUON2UJKQjSnD09W7aUVpRPyNPBsAm4SQoyWUm5pa0EhxCjgdOAZD/fpNsfkVmf3fyKE2ARMQRvbf50tFxGsp1YnApYG3ZTVamPLT3s5lF+CBGrr6qmzaOXvQt2sjiylbMxIK6uu92pBTl+L/voLBj54D8UTJrH9wacoqbNRV1lBn8QYJvRLVu2lFaUT8zTwPIvWh2elEOI8KeVOZwsJIbLQmsDZgH96uE9va0gDb7V4qUAELA26KbvdzhMvruNYSUXjfRaLze0P2bp6G9V12q+rZaQZOH1wUsB+r44oG30qu//4F7ZnX0BdZb1qL60oXYhHnzRSylzgPiAd+EEI8aoQ4lohxDmO2zVCiFfRCoj2BRY61ukUhBDpwGSgBi3jrlX+SoNuqa7ewpaf9iKlRKfTccbYLLL6pzQ+3pAC3R6L1U5ZVR3FFWZAMKh3NOOyEokICcJo0HWZoJO86nWCCgoorjDz3eSLSOmTwLkThnDa8EwVdBSli/C4ZI6U8j4hhBW4F7gC+L8Wiwi0o4q7pZQPero/bxFCmIDXABNwh5SytLVla6vLObhXi5eP3nk/p2dfwPjsizq03xQXC3vWmuv58rtf2PzNLmpq60iIiyQtJY7BA/vwzobvGpdra+JnIApy+oyUpC59loylz2KdsQ/jPfcxMCORiDD/fhFQFKV1L7zwAi+88ELDj3GtLeeVTyAp5QNCiNeAa4HxHO/LU4BWUudlKeV+T/YhhMhDO7Jy1WtSSqfdvoQQeuDfaGNdAfy9rQ2FhEURlzkIgLlz/uXGENxXX29l01c7+HxLLrXmerL6pzB5/FDSUrTXsL3ma4EqyOlLNquN5CcfJWPFyxRfNIPUl54j3MuFTBVF8dycOXOYM2cOAEKIVicFerM69X60ox5f2YN7qdj5zu50BJ1X0SoovAlcKVt+kgeA3W5Hp9MhdIJvtv5Kv7REzp5wMn2Smpd8aW3i596DxyiuMAekIKevWK02SsqqGbr0SQatfg3Ltb+j19IX6NK/lKIoiE7wmes3QoggtNNrM4DXgaullLb21ktMHSDjzltI/s5tpMQEu93wrKmWEzbLK2v45Kud7N5XwG2/Ow+9Xoe5ztJud8slr36IzS6ZedGZAS3I6QtN20sPSYpg0FXT0U2cCE8/TZc9bFOUHkYI8Z2U0ul0my54sr9jHBWy3wQuBF4BrpFS2ttey/usNhtHiyo4mF/Elp/28u2Pe7DbJSNOzqCu3kJoiKnNoGOut1JttmKx2QnS6zg5PTZgBTm9rVl76YwkMpOjCY4Ig88+g4gIFXQUpZvoEYHHkUiwGq1O3EvAnEAEHYDS8mrMdRaeWbYevU7HqGGZnHXaYHrFRLS6TtOCnJGhRganxhAVakQnBAnR3umfE0gntJdOisZ0041QXAyrV0NkZKCHqCiKF/WIwAMsQQs6RcBh4B4ns9k3SSk3+WoAR4vKyTt0rLHxmk4nuOnqc0hNcV66rrWCnCEmrRCnrht8+29oL20MMjAyK5X03r0wCmDWLHj9dVi4EPSqGZuidDc9JfA0FEaLA+5pY7lNrT3QtD6aOwqOlvLR59v5aed+DAZ9Y+M1IQRbftrTLPDYbPbG02imID3p8eHERQU3K8jZHbTaXtpigSt+CytXwgMPwF13BXqoiqL4QI8IPFLKSf7e55Gicj7YtJXtvxzCZDRw2qgBfLN1T+PjDfNvzj79ZHRBQdRb7Bj0uk5dkNNT7baXvuEGLej8/e8wf37gBqooik/1iMDjT1abVtG5rt7C3gNHyZ4wlAljBvHBJz/SMn/Qbpe8/8mPzLp4AgkxIUSFdP2MNGeatpc+fXgmvRNjnFfOvvlmGDMGbrzR/4NUFMVvVODxAiklew8cZePn24iJCmPGb8aRlhLH3b+/GGOQ9hTvP+Rk/o3dTllpBVmp0QEYte81tpcOb6O9dE0NvPUW/Pa3MGqUdlMUpVtTgccDUkp27ytkw2c/k3foGOFhwQwecLxytTHI0FiQ8+rLziYmzMSylR9j0Aseu2N6AEfuWy63l66uhpwc2LQJhg2DoUP9PlZFUfxPBR4PbPpyB+9v2kpURCgXnjOasaf0IyjIgMWq1UizS0l4sJGBKVHERgYTYtTzupfm2wS666czFdVmqhztpSeOGUhSXGTrvXAqKuA3v4EvvoBXXlFBR1F6EBV43NDQfC13Tz6D+qUwfEg6ISFGrb20EFpBzlprY0HOuMguWpDTTQ3tpeNjIxjjSnvpsjI491zYsgXeeANmzPDbWBVFCbzu/6noJT9s38ehghKkhLUbvmNQvxSiwkMZPCid8lorQY6CnHGRwUSEGLtV+rMzHrWX/uQT2LoVVq2Ciy7y9VAVRelkVOBxQUlZJf9Z88Xxn8ur2V9QRnRkKIkxoSR2g4KcrpJSUlZZi9lsITkhmvEjkukVHe7ayna7VuDzwgthzx7oHbhOroqiBE4P+Kj0AikJ19c0+VGSm7uP8SclMah3FNHh3T/oSCkpKa+m4Fg5cdHhTBk/mDNHD3A96BQWwtix8KFWJFUFHUXpntauXdvQGiGqtWW6+celd0RFhWMWx2up2Wx2Nn+bS3lVTRtrdQ92u6S4rIrCokqS4qI594yTGT+yv3vttg8fhkmTYOdOVQJHUbq5nJychmZw5a0towKPCyz1FmjZfM0uee3dbwI0It+z2yXFpdUcKa6kd2IM504YwrhTMoh2twHbgQMwcaIWfD74AM4+2zcDVhSly1DXeFxQV29FtJj8abXZ2bmnIEAj8h2b3U5JWQ1Wm43+aQkM7OtBe+mjR7WgU1qqnWIbN867g1UUpUvqUY3gOqrfwMHy5dff5vZH/wt0zjk0nrLZ7JSU12CXdgakJ9I/LYHwUJNnG7Xb4U9/Ol6VQFGUHkM1glNaZbXaKCmvQQKD+moBJzTE6NlGd+2C4GDo2xcef9wbw1QUpRtRgaeHsjgCjgBO6pdEZp94QoI9DDgA27bB5MmQmalVJejuE5qULm/ms18C8MYtpwV4JD2HSi7oYSxWG4VFlZRV1DCkXwrnTxrGkP69vRN0tm7VstcMBli2rMcHnUmTJjF9uvOafKNHj2b27NkubysvLw8hBO+++26byy1YsIC4uDh3hulTzz77rGuTihWP/Prrr9xwww0MGzYMvV7PpEmT3Fp/x44dTJ48mdDQUFJSUrjnnnuw2Wy+GSzqiKfHaN5eujd9e/fCZAzy3g62bIFzzoHwcPjoI+jf33vbVlx23XXXkZOTE+hhKH62fft21q1bx7hx47BYLG6tW1paSnZ2NoMHD2bNmjXs2bOH+fPnY7fbWbRokU/GqwJPN9e0vfSok9JIS4ltbNXgVXfdBdHRWtDp29f721dc0qdPH/r06RPoYSh+lpOTw4UXXgjA9OnTKSoqcnndJUuWUFtby+rVq4mMjGTKlClUVFSwYMEC7rjjDiIjI70+XnWqrZsy11nIP1pOjbme0UP6cv7EYfRPT/BN0AF4803YvFkFnQ5oOCX3+uuv079/fyIjI5k2bRqHDh06YdmKigquuuoqIiIiSEhIYOHChc0eb3mqrbq6mltuuYVBgwYRGhpKRkYGN998MxUVFc3We+mllxg8eDAhISHExcUxceJEtm/f3vi42WzmjjvuIDU1FZPJxCmnnMK6deuabaOuro5bbrmF6OhoYmNjmTdvntvfvv1pzD0fkjHvPb7eU8LXe0rImPceGfPeY8w9H3ptH4WFhcyaNYvExER0Oh1CiMbbKC9mejptO+Ki999/n6lTpzYLMDNnzqS2tpZPPvnEG8M7gTri6Waatpced0rGie2lvemjj+DZZ+H11yEmRrspHfL111+Tn5/PY489Rm1tLbfeeitz5sw54cP99ttv5/zzz2fVqlVs3ryZhQsXEhcXx8033+x0uzU1NdhsNhYvXkx8fDwHDx5k8eLFzJgxg/Xr1wOwefNm5s6dy3333cdpp51GRUUFX375JeXlxyeeT58+nW+++YaFCxfSr18/3nzzTS644AK2bNnC8OHDAbjzzjt58cUXWbx4MYMHD2bp0qWsXLnSN0+YFxRV1rt1v7vMZjPZ2dlUV1fzyCOPEBcXx8MPP8ynn37KnDlzOOuss5otL6V06bqKweDdj+1du3ZxdouJ3WlpaYSGhrJr1y6fnLpVgaebcLm9tLf8739asc/+/aGyUkuf7uycXXC97DK46SatE+p55534+OzZ2s3HKioqeO+994hxBO/CwkLmzZtHbW0tISEhjcsNGTKE559/HoCpU6dy9OhRHnjgAW688Uan33rj4+N57rnnGn+2Wq1kZGQwYcIEDhw4QFpaGt988w3Dhg3jrrvualzuggsuaPz/xo0bee+999i0aRMTJ04E4JxzzuGXX35h8eLFrFy5kuLiYpYsWcLChQuZP39+4/gGDx7sxWepa1m0aBEHDx5kx44d9HbUJszKyqJ///5MmDCBmTNnNlt++fLlXHPNNe1u19tzL0tLS4mOjj7h/piYGEpLS726rwYq8LihM04cra6to7zSTGSE1l66d0K0R4fdLnn3Xbj0UjjpJNiwATpRFlVXNWbMmMagAzR+YB8+fJj+TRI1Lr744mbrXXLJJbz44oscOnSItLQ0p9v+97//zeOPP87u3buprq5uvP+XX34hLS2N4cOHc8cddzBv3jwuvvhixo0bh9F4PMtxw4YNJCUlMX78eKxWa+P9kydPZtmyZQD8/PPPmM3mxusMoJ3+ufDCC3nkkUc68Ix0fa+99hrXX399Y9AByMzMRAhBWVnZCcvn5OTw7bff+nGEgaMCTxfVrL306DbaS3vbO+/A9Olwyimwfj3Exvp+n96yaVPrj4WGtv14BxgMhlZPndhstmanTFp+42z44Debzc3uT0hIcPpzQUGB08Dz1ltvcfXVV3PjjTfywAMPEBsbS0FBARdffHHjtrOzs3n55Zd5+umneeqppwgPD+eqq67ikUceISwsjKKiIgoLCwkKOjELUu8o+lpYWNjm+HqaXbt2kZeXR3Z2drP7jx07hpSS5OTkE9aJjY0lKqrVgs4+ExMT0+y0aoPS0tJmX4a8SQWeLsat9tK+kJEBU6fCq69CAP5IupL4+Hjy8vKcPlZQUNChD+WjR486/dnZBxnAypUrOfXUU/nnP//ZeJ+zC8azZs1i1qxZHDt2jNWrVzNv3jwiIiJ46KGHiI2NpXfv3rz99tutjispKalxPLFNvoy0HG9P0ZAY0vI1Xr9+PUFBQUyZMuWEdQJ1qi0rK4tdu3Y1u+/gwYPU1NSQlZXl1X01UIGni3C7vbS3/fADDB8OQ4fC2rX+228XdsYZZ/Df//6Xw4cPNzvd8vXXX3PkyBHOOOMMt7f51ltvceONNzb+vHr1apKTk1tNoa6trcVkal5z77XXXmt1+/Hx8dxwww2sXr2aHTt2ANoptccee4zw8PBWP4iGDh1KcHAwa9asaVzGbrezZs0at34/f4qLMDpNJIiL8HwydcMRbG5uLiNHjgS0o9dFixZx2WWXOT2yCdSptmnTpvHoo49SWVlJRITW/mXFihWEhIQ0XtPzNhV4XFBdVcmji//G6Wecxfgz/VfW36P20t70r3/BddfByy/DrFn+3XcXdvXVV/P4449z5pln8te//pX09HR27tzJwoULOf3005k6darb29y+fTs33HADl156KZs3b+all17iqaeeavU065QpU7j55ptZvHgxp556KuvWrWPjxo3Nlrn33nspKSlh0qRJxMXF8cMPP/DJJ5/w0EMPNW5j6tSpTJkyhT//+c8MGTKEiooKtm7ditls5sEHH6RXr17MmTOHe++9F4PBwJAhQ1i6dClVVVXuP3F+8u192lGHL0rmDB8+nMzMTP785z+j1+vR6XQ8/PDDmM1mnnnmGafr9OrVi169enVofzU1NY0ZkIcPH6aiooJVq1YBcN555xEaqrUzeeWVV7j22mvZs2cP6enpAMydO5enn36aSy65hD//+c/s3buXBQsW8Mc//rFDc3jWrl3LWu3LaaunRFTgcUFYeAS3332/3/bnUXtpb3v+eZg7V6tKcNllgRlDFxUeHs7mzZv5y1/+wp133klJSQmJiYlcfvnlLF68uEPX5B555BHeffddLr30UoKDg/nb3/7GLbfc0uryN9xwA3v37uWpp57CbDYzZcoUXn/9dcY1aVExZswYnnjiCd544w0qKytJT09nwYIF3HrrrQAIIVi9ejUPPPAATz75JAcOHCA2Npbhw4fz+9//vtnYLBYL9913HzqdjiuvvJI//vGPjVluPYnBYOCdd95h7ty5XH311URERHD++efzwAMP+OS6ydGjR5kxY0az+xp+3rdvH30d8+vsdjs2m63Z6bqYmBg2btzILbfcQk5ODtHR0cybN48FCxZ0aCw5OTnk5OSwdOnSVhvBqbYILmhoi+BrUkpKK2qoq7fSJzGGk/olu9fp09ueeQb+8Af4zW9g1aqukTKtKG5SRUJ9Q7VF6OTsdklpRTX1FjvpKb3Iykx0v9Ont/3yC9x2G1x8MbzxBhi9UERUURQFFXgCyma3U1pWg8VmJzM1jkF9E4kMD2l/RX8YOBA2boTx48FJGq2idBfqSMf/VOAJgIb20ja7nX6p8Z61l/YmKeH++2HECMjJcT7TX1EUxUMq8PhRQ3tpm93OwL5eai/tLVLCX/4CDz2kJROo0vqKoviICjx+YLXaKC6rAeHF9tLeJCXMnw9PPKEFnX/8I9AjUhSlG1OBx4ca2kvrhGBwfy+2l/Ymu13LXPvHP7R/n3yyx3cOVRTFt1Tg8YF6i5WS8lr0Oji5fwoZfeIINnXSC/RCaEc8f/oTPPKICjqKovicCjxedLy9tN437aW9yWaDwkLo3VvrqQMq6CiK4hcq8HiB1l66BpPRx+2lvcVqhauv1jqGbtumtaxWFEXxk0786dj5messlFbUEGwKYszJ6aQmxRIU5KNun95iscAVV2iVCB58UAUd4MlP9wNw2xnpAR6JEgjn/PVdAP636PwAj6Tn8EMDl+6n1lxP/tFyzPVWTh2WwW/OHEpmanznDzp1dVovnVWr4PHH4c47Az0ixct27NjB5MmTCQ0NJSUlhXvuuceldsqerqsE1q+//soNN9zAsGHD0Ov1THJjDt6yZcsQQpxwW7Jkic/Gq4543NC0vfT4Ef1ISYj2bXtpb1u4UGvk9uyzcPPNgR6N4mWlpaVkZ2czePBg1qxZw549e5g/fz52u51Fixb5bF0l8LZv3866desYN24cFoulQ9v46KOPmrVZz8zM9NbwTqACjwvsdkn+0XIiI0KYMLI/KfFR/un26W133gmjR8MllwR6JIoPLFmyhNraWlavXk1kZCRTpkyhoqKCBQsWcMcdd7RZ4t6TdZXAy8nJaWw7Pn36dIqKitzexpgxYwgP908V/C746el/deZq1v7nH1hK9tAnMaZrBZ2qKrj9dqipgchIFXQCoLCwkFmzZpGYmIhOp2t2OmPUqFFe28/777/P1KlTmwWJmTNnUltb67TrqLfW7eqqzRa++/UYOw6Uen3b/nrtO9Nn0tq1a5kzZw600Y+n84y2E0tJSuC1fy9r/EbRZVRUwLnnahUJPv880KPpdB78aC83v7WT3UU17C6q4ea3dnLzWzt58KO9XtuH2WwmOzubzZs388gjj7B27drGzqNz5szh9ttvb7a8lBKr1druzZldu3ad0CE0LS2N0NDQE1obe3PdrqzabGH7gVJq6mxcvGg91eaOnaZyxp+vvTf069cPg8HAoEGDeP755zu8nZycHF544QWAVvvxqFNt3VVpqRZ0vv9ea2vgpMd7T5cRG0JBZR02+/H79Drtfm9ZtGgRBw8eZMeOHY3tr7Oysujfvz8TJkxg5syZzZZfvnw511xzTbvbddZHq7S0tLHlclMxMTGUlrb9bd6TdbuyG57dTL1VewMcLatl7rOf8u8/eafLsD9fe08kJydz//33M3bsWGw2G2+88QZz586lpqaGefPmeXVfDVTg6Y6Ki7VAs22blsHW1Y7U/GRaVhxf7S/HxvE/ZB2CaVnxXtvHa6+9xvXXX9/4wQPaRVshBGVlZScsn5OTw7fffuu1/SutW74hl/e3HKThc9xssbFuywGWb8hlVvYgj7ffVV77qVOnNmvDPm3aNMxmM4sWLeLWW2/1yWk8FXi6o+JiKCmBNWtg2rRAj6bTigoOYlx6FJ/uKwO0o51x6VFEBXvnz2LXrl3k5eWRnZ3d7P5jx44hpSQ5OfmEdWJjY4mKavXUeJtiYmIoLz/x7EZpaWm77ZY9Wber+tur31JT1/zUVU2dlb+9+q3Hgcffr723TZ8+nTfffJO8vDyfZLepazzdSXm5Vndt4EDIzVVBxwXTsuJoKBTk7aOdQ4cOAZCQkNDs/vXr1xMUFMQUJ6c/ly9fTlBQULs3Z7Kysk64HnPw4EFqampOuH7jzXW7qvuvHEOoqfmXjFCTgUVXjfV42/5+7b1NOMpnCR+V0VJHPN3F4cNw9tlw+eVw331g6iR9fjq5qOAgIkx6KupsXj3aARqvmeTm5jJy5EiAxlMYl112mdNvt56cbpk2bRqPPvoolZWVREREALBixQpCQkKYOHGiz9btqmZlD+LDrYdY/cU+pITgID3njU7j6skDPd62v197b1u1ahVxcXGkp/uomoeUUt3auY0aNUp2anl5UmZmShkRIeVnnwV6NF3Oo5v2yT++s0uW1Vq8ul2LxSIzMzNlamqqXLFihVy5cqUcPXq0TEtLkyUlJV7dl5RSlpSUyKSkJJmdnS0//PBD+fzzz8uwsDB59913N1tu+fLlUq/Xy7y8PLfX7W6qautl1IyXZPBFS+XA6/8jq2rrvbJdf7/21dXVcuXKlXLlypVy3LhxcvDgwY0/V1dXNy7n7LW/5JJL5EMPPSTXrVsn165dK6+88koJyKefftqjMQFbZCufqQH/UO8Kt04dePbskTI9XcqoKCm/+irQo+mSnticJ5/YnNf+gh2wbds2OWHCBGkymWRcXJycPXu2zM/P98m+pJRy+/bt8qyzzpLBwcEyKSlJ/vWvf5VWq7XZMi+//LIE5L59+9xetzs6ff5qGXv5v+T2/d4NCP587fft2ycBp7emr7Oz1/6uu+6SAwcOlCEhITI4OFiOHDlSvvLKKx6Pqa3AI7THlbaMHj1abtmyJdDDOFFdHZx0knZt58MPwXFIr7hHFQnt2VSRUN8QQnwnpRzt7LEem1wghHhRCCEdt/6BHk+HmEzw2GPw8ccq6CiK0mX0yCMeIUQO8A5QBYQDA6SUv7a2fKc74vn5Z9izBy66KNAjURRFcaqtI54el9UmhIgHlgIrgCSga6Xs/PCDNjk0IkJLl1bZa4qidDE98VTbC45/u15fgG++0VKmw8JgwwYVdBRF6ZJ61BGPEGI2cBFwkZSy2FeTo3ziiy+0I5xevbRrOr7Kr1cURfGxHnPEI4RIB54CXpVSrgn0eNz2/vuQmAibN6ugoyhKl9YjkguEEDrgI2AAcLKUstRx/ya0azxtJhekp6fL+PjjpVTmzJnT0G/C9+rrwWjUSuGUlUE3rZulKErX98ILLzS0ROC7777bL6Xs62y5LhN4hBB5gDtf9V+TUl7pWHc+8HfgN1LKdU22uQkXAk/AstrWr4e5c7V/B3pexkNRFMVfuktW2x7A7Mby+QBCiIHAYuDlpkGn01u7FqZPh8GDITY20KNRFEXxmi4TeKSUkzu46mDABFwjhGity9JuR6LBxVLKtzu4H+9ZvVor9jlihHa0o06vKYrSjXSZwOOBPOClVh77DdpcnpVAhWPZwNqwAS67DMaO1RIKOkl/ju5s3n+3A/DEpUMCPBIlEM657kkA/vfibQEdR0/S7bPapJRbpZTXObsBuY7F/uK4b2sAh6o57TS49VbtSEcFHcVNO3bsYPLkyYSGhpKSksI999yDzWZrd71ly5YhhDjhtmTJEj+MWvHUypUrueCCC+jduzfh4eGMGjWK//znPy6v39H3TUf1hCOeruHtt7XJoZGRWv01RXFTaWkp2dnZDB48mDVr1rBnzx7mz5+P3W5n0aJFLm3jo48+IiQkpPFnX3SfVLzv8ccfJyMjgyeeeIK4uDjWrVvHFVdcQVFREb///e/bXNcb7xt3qcDTGTz3HNx0E9x5Jzz4YKBHo3RRS5Ysoba2ltWrVxMZGcmUKVOoqKhgwYIF3HHHHURGRra7jTFjxhAeHu6H0SretHbtWuLi4hp/Pvvss8nPz+fxxx9vN/B4433jrm5/qq0tUspJUkrRViq1zz31lBZ0cnJgwYKADaOn2pB7jB2Flfx4uIKZL3/HhtxjXt9HYWEhs2bNIjExEZ1O1+xU1qhRo7y2n/fff5+pU6c2+6CYOXMmtbW1fPLJJ17bT3dTb7HyY+4hCosqvL5tf732TYNOgxEjRpCfn9/uuoF43/TowBNwjzwCt90Gl1wCq1ap2mt+tiH3GI9t3IvFps1lO1pZz2Mb93o1+JjNZrKzs9m8eTOPPPIIa9eu5YwzzgC0ici33357s+WllFit1nZvzuzatYusrKxm96WlpREaGsquXbtcGm+/fv0wGAwMGjSI559/vgO/cddzIL+EiiozDy5936vb9edr78yXX37JQBfm/3njfeMudaotUMrK4OmnYeZMeOUVCAoK9Ih6nBe/OECd1d7svjqrnRe/OED2oPhW1nLPokWLOHjwIDt27KB3794AZGVl0b9/fyZMmMDMmTObLb98+XKuuaa1rP/jnE38Li0tJTo6+oT7Y2JiKC0tbXN7ycnJ3H///YwdOxabzcYbb7zB3LlzqampYd68ee2Op6sqOFbOkeJKAP695ivuun4aSXHeObXkz9e+pY0bN/L222/zr3/9q91lPXnfdJQKPP7W8KaJjoavvoKkJDColyEQjlXWu3V/R7z22mtcf/31jR88oF2wF0JQVlZ2wvI5OTl8++23Xtu/q6ZOncrUqVMbf542bRpms5lFixZx6623otN1z5MjDy59H4n2N2mz23lw6fs8ddflXtl2oF77vLw8rrjiCi688EJmz57t8fZ8QX3i+ZOUWgJBba12badPn0CPqEeLjzBy1EmQiY8wemX7u3btIi8vj+zs7Gb3Hzt2DCklycnJJ6wTGxtLVAfT6GNiYigvLz/h/tLSUmI6MAl5+vTpvPnmm+Tl5XXL7LaCY+X8+52vG78L1ltsXjvq8fdr36CkpIRp06aRnp7Oa6+95tI63n7fuKJ7fo3pjKSEefO06zpW6/EjHyVgrjs9DZOh+Z+AyaDjutPTvLL9Q4cOAZCQkNDs/vXr1xMUFMSUKVNOWGf58uUEBQW1e3MmKyvrhHPyBw8epKam5oRz+K5oaBvSpdqHuOHBpe9jtzc/1dpw1OMpf7/2ADU1NZx//vnU19fz7rvvEhoa6tJYvf2+cYU64vEHux1uuUVLm771VnjiCeimf8xdScN1nEc37MFikyREGLnu9DSvXd9pOG+em5vLyJEjARpPX1122WVOv916crpl2rRpPProo1RWVhIREQHAihUrCAkJYeJE9xvtrlq1iri4ONK7aRuOr3/Ko97SfJJkvcXGVz/u83jb/n7trVYrM2bMYPfu3XzxxRcnBLy2ePt94xIppbq1cxs1apT0yE03SQlS3nGHlHa7Z9tSvO62Vdvkbau2eX27FotFZmZmytTUVLlixQq5cuVKOXr0aJmWliZLSkq8vr+SkhKZlJQks7Oz5Ycffiiff/55GRYWJu++++5myy1fvlzq9XqZl5fXeN8ll1wiH3roIblu3Tq5du1aeeWVV0pAPv30014fZ2cz5XdPyCm/e8Kr2/T3a3/99ddLQD711FPyyy+/bHYzm82Nyzl77V1937gL2CJb+UwN+Id6V7h5HHjWrJHy3ntV0OmkfBV4pJRy27ZtcsKECdJkMsm4uDg5e/ZsmZ+f75N9SSnl9u3b5VlnnSWDg4NlUlKS/Otf/yqtVmuzZV5++WUJyH379jXed9ddd8mBAwfKkJAQGRwcLEeOHClfeeUVn42zM/FF4JHSv699enq6BJzemr7Ozl57KV1737irrcDTZfrxBNKAAQPkWWedRU5ODjk5Oa6tZLHAl1/CmWf6dnCKx1SR0J5NFQn1rrVr17J27VqWLl36q5RygLNlVHKBC6KionjhhRdcDzr19dr8nLPPhtzc9pdXFEXpJnJychq6kJ6YKuegkgu8ra4OZszQGrk98QQMGhToESntUEc6PZs60vE/FXi8qbZWK3/zwQfwz3/CjTcGekSKoiidjgo83vTmm1ofnRdfhN/9LtCjURRF6ZRU4PGmq6+GoUPBkbevKIqinEglF3iqvBwuvBC2bdMmhaqgoyiK0iYVeDxRWgpTpsC6dbBnT6BHoyiK0iWoU20dVVQE55wD27fD6tVaIzdFURSlXSrwdERRkTZHZ/duWLMGzj030CNSFEXpMtSpto4ID4fMTHj3XRV0FEVR3KSOeNxx+DCEhkJMDLz9dqBHoyiK0iWpIx5X7d+v1V273DvdCRVFUXoqFXhcUVenBZ2SEli8OCBDcNQ+UlDPRVPquThOPRfHdfbnQgUeV+TmQnU1fPQRjBnj9upr1671eAjeeCN5Og5v/B7qufDuNtRzcZx6Lo7rDM8F0GofbxV4XCElfPwxjBjRodW98Ubyhu7yR+UN6rk4Tj0Xx6nn4jgvjCO6tQdUP552CCE+AE4DdnuwmSjaKBHuojigyMNteDoOb/we6rnw7jbUc3Gcei6O6wzPxQAppdOjHhV4FEVRFL9Sp9oURVEUv1KBR1EURfErFXgURVEUv1KBpxsQQrwohJCOW/9Aj8dfhBADhBB/FkJ8JIQ4KISoF0IcEUKsEUKcFejx+YoQoo8Q4l9CiHwhRJ0QIk8I8aQQIibQY/MXIUQvIcR1Qoi3hBC/CiFqhRDlQojPhBC/E0L06M82IcSVTT4Trgv0eFpSyQVdnBAiB3gHqALC0TJJfg3sqPxDCPEGcDmwA/gMKAEGARcAeuBWKeXTgRuh9wkh+gFfAAnAGmAXMBY4C8gFxkspiwM3Qv8QQswFngMKgI+BA0AicAlaNtZ/gRmyB37ACSFSgZ/R/gbCgeullC8GdlTNqcDThQkh4tHeYJuAJGAiPSvwzAZ+lFL+0OL+icCHgAT6SikLAjA8nxBCrAfOAf4gpXymyf2PA/OA56WUcwM1Pn8RQpwNhAHvSSntTe5PAr4BUoHpUsr/BmiIASGEEGjv/QxgNfAnOmHg6dGHo91Aw/TkmwM6igCRUi5rGXQc93+CFoyNwOn+HpevOI52zgHygH+0ePheoBq4SggR5ueh+Z2U8iMp5dqmQcdxfyGwxPHjJL8PLPD+AJwNXIP2fuiUVODpohzf9i8CbugJp1Y6wOL41xrQUXhXw3Wr/zn5wK0EPgdCgXH+Hlgn0x1f+3YJIU4CHgKeklJuDvR42qICTxckhEgHngJelVKuCfR4OhvH8zMZqAE69R+gmwY5/v2llccbqmsM9MNYOiUhhAG42vHjB4Eciz85fu9/o13r+kuAh9Mu1Y+ni3Fk6yxHSyb4Q4CH0+kIIUzAa4AJuENKWRrgIXlTQ/mR1sqYNNwf7fuhdFoPAScD66SU6wM9GD+6BxgBTJBS1gZ6MO1RRzwB4Eh/lW7cXm2y+jy0JILru8OHqofPRctt6dG+9Y0HVgB/99fvoQSeEOIPwHy0TL+rAjwcvxFCnIp2lPOYlPLLQI/HFeqIJzD2AGY3ls8HEEIMBBYDL0sp1/liYAHQoeeiJUfQeRWYAbwJXNkNU2kbjmhaKzffcH+Z74fSuQghbkE7/bwDmCylLAnwkPzCcYrtFbTTr38L8HBcpgJPAEgpJ3dw1cFop5CuEUJc08oyu7WMSi6WUr7dwf34jQfPRSMhRBDa6bUZwOvA1VJKm6fb7YRyHf+2dg1ngOPf1q4BdUtCiNuAJ4BtaEHnaGBH5FfhHH8/mB1/+y0tFUIsRUs6uM1fA2uLCjxdSx7wUiuP/QZtLs9KoMKxbLcnhDCiHeFciPbN75qWGV/dyMeOf88RQuhazF+JQDvFWAN8FYjBBYIQ4s9o13W2AlOklJ62Auhq6mj9M2Ek2nWfz9C+tHSa03BqAmk3IYTYRM+bQGpCmyR3Htof35xuHHQANYG0KSHE34D7gO+Ac3rK6TVXCSEWoM3v6nQTSNURj9KVLUELOkXAYeAeJ6caNkkpN/l5XL50E1rJnKeFEJOBncCpaHN8fgHuDuDY/EYIMQst6NiAT4E/OHnt86SUy/w8NMUFKvAoXVmG4984tHTS1mzy/VD8Q0q5RwgxGu1D91y0wFuAdmF9YXfIdHRRw2uvB25rZZlPgGX+GIziHnWqTVEURfErNY9HURRF8SsVeBRFURS/UoFHURRF8SsVeBRFURS/UoFHURRF8SsVeBRFURS/UoFHURRF8SsVeBRFURS/UoFHUVzgZs8g6WhN3qkJIe524/epcjQhVBSPqZI5iuKahU7uuw2tB85TnNgDZ6tvh+MVO3D+ezWIAW5B+4K6prsXYFX8R5XMUZQOEkLkAelAhpQyL7Cj8S4hRC/gQ7Sy+m8Dl0sp6wM6KKXbUIfOiuIjQoi+jtNUy4QQA4UQK4QQR4UQdiHEJMdNOsrXO1s/zxHcnD12qhBilRCiUAhRL4Q4KIR4XgiR4oVxJ6D1/hkBrAJmqKCjeJM61aYovtcP+BqtbcFrQAhas77IjmxMCHEt8AJaE7B3gINo3UevA3KEEOOklAc6uO0k4CPgJOA/wFXdtJurEkAq8CiK700AHpRS/qXpnUKISe5uSAgxEK0PUR4wUUp5uMljk4H/oV1zurgD2+6NFnQG0v27uSoBpE61KYrvHaHti/juuBEIAm5tGnQApJQb0Y6AchytsF0mhEhD618zEK2bqwo6is+oIx5F8b0fpZR1XtrWaY5/Jwohxjh5PAGtOdpAtJbQ7RJCZKAd6fRFO5q6STrJOhJCnAOsb2dzs6SUr7iyX6XnUoFHUXyv0Ivb6uX49/Z2lgt3ZWNCiP5oQScVeEZK+Yc2Fv8USG7y8zfAm8Dfm9xX4sp+lZ5NBR5F8b3W5iw0nMpq7e8wmhPnB5U7/o2SUlZ4MighxCC0oJMCPCal/FNby0spa4Fax7pRQB/gcymlNwOr0gOoazyKEjiljn9TWz7gOBKJcrLOV45/z/Bkx0KIIWjXdFLQEh/aDDpOjAQELp7OU5SmVOBRlMDZhZZWfaFj7gwAQogQ4OlW1nkWsABPODLcmhFCGIUQbQYlIcQwtHk6icB9LbPtXDQKKOpo2rbSs6lTbYoSIFJKixDiKeBvwA9CiLfQ/ianAPmOW8t1djnm8fwL2C6E+ABtflAQkIZ2JHQMyHK2T0fK9MdALLAPaHUCq8MeKeW/ndw/EvjBld9TUVpSgUdRAuteoAa4HpiDlojwBrAArZbaCaSUrwohfgTmA2cB5wDVaIFqFbCijf1NQAs6ABmO/bflGaC1wPNWO+sqilOqVpuiKG4RQoSjnSK8XEq5MtDjUboedY1HURR3jUBLLPg+0ANRuiYVeBRFcddItLTuvYEeiNI1qVNtiqIoil+pIx5FURTFr1TgURRFUfxKBR5FURTFr1TgURRFUfxKBR5FURTFr1TgURRFUfxKBR5FURTFr1TgURRFUfxKBR5FURTFr/4fJ4wgET4GDyYAAAAASUVORK5CYII=",
      "text/plain": [
       "<Figure size 432x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig = plt.figure(figsize=(6, 6))\n",
    "gs = gridspec.GridSpec(1, 1) \n",
    "plt.subplots_adjust(hspace = .001)\n",
    "\n",
    "ax0 = plt.subplot(gs[0])\n",
    "ax0.yaxis.set_ticks_position('both')\n",
    "ax0.xaxis.set_ticks_position('both')\n",
    "ax0.tick_params(direction=\"in\",which=\"both\")\n",
    "plt.xticks(fontsize=20)\n",
    "plt.yticks(fontsize=20)\n",
    "ax0.minorticks_on()\n",
    "\n",
    "# colors\n",
    "n_lines = 5\n",
    "c = np.arange(-3, n_lines + 1)\n",
    "\n",
    "norm = mpl.colors.Normalize(vmin=c.min(), vmax=c.max())\n",
    "cmap = mpl.cm.ScalarMappable(norm=norm, cmap=\"Blues\")\n",
    "cmap.set_array([])\n",
    "\n",
    "markers = [\"v\",\"o\",\"s\",\"d\",\"^\"]\n",
    "myc=-1\n",
    "for (i,sigma) in enumerate(np.linspace(0.01,2,5)):\n",
    "    myc+=1\n",
    "    gauss_truth_prior_B = np.random.normal(0,sigma,10000)\n",
    "    gauss_reco_prior_B = np.random.normal(gauss_truth_prior_B,2,10000)\n",
    "    mycalibration_B = np.polyfit(gauss_reco_prior_B, gauss_truth_prior_B, 1)\n",
    "    mypoly_B = np.poly1d(mycalibration_B)\n",
    "\n",
    "    meanvals_B = np.array([np.mean(mypoly_B(np.random.normal(xx,2,10000))) for xx in np.linspace(-5,5,20)])\n",
    "    meanvals_analytic_B = b(np.linspace(-5,5,20),sigma,2,0)\n",
    "    resvals_analytic_B = analytic_resolution(np.linspace(-5,5,20),sigma,2,0)\n",
    "\n",
    "    plt.errorbar(np.linspace(-5,5,20) + sigma / 20,meanvals_B, yerr = MSE_stds, marker=markers[myc],ls=\"\",color=cmap.to_rgba(i + 1),label=\"$\\sigma=%0.1f$\" % sigma)\n",
    "    plt.plot(np.linspace(-5,5,20),meanvals_analytic_B,ls=\"--\",color=cmap.to_rgba(i + 1))\n",
    "    plt.fill_between(np.linspace(-5, 5, 20), meanvals_analytic_B + resvals_analytic_B, meanvals_analytic_B - resvals_analytic_B, color = cmap.to_rgba(i + 1), alpha = 0.25)\n",
    "\n",
    "plt.plot(np.linspace(-5,5,20),np.linspace(-5,5,20),ls=\"--\",label=\"Unbiased\",color=\"red\")\n",
    "\n",
    "plt.xlim([-5,5])\n",
    "plt.ylim([-5,5])\n",
    "plt.xlabel(\"True $Z_T$\",fontsize=20)\n",
    "plt.ylabel(\"Calibrated $\\hat{Z}_{T}$\",fontsize=20)\n",
    "plt.legend(frameon=False,fontsize=15,ncol=2,loc=\"lower right\",handlelength=1)\n",
    "_=plt.title(\"Simulation-Based Gaussian Example\",loc=\"right\",fontsize=20, fontstyle=\"italic\")\n",
    "_=plt.text(-4.5,4,r\"$(\\mu,\\epsilon)=(0,2)$\",fontsize=20)\n",
    "plt.savefig(\"Gaussian_prior3.pdf\",bbox_inches='tight')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "interpreter": {
   "hash": "0c189c3790ad0f879b11359947e0f24508572947ea4470e1374962daa7aecefe"
  },
  "kernelspec": {
   "display_name": "Python 3.8.3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.3"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 2
}