{
"cells": [
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Implementation of the CNN model from Cui, Fearn 2018\n",
"\n",
"This is our 2nd exploratory notebook in this series dedicated to the applications of CNN to spectral analysis. \n",
"\n",
"This notebook implements the CNN model proposed by Cui, C. and Fearn, T. 2018, \"*Modern practical convolutional neural networks for multivariate regression: Applications to NIR calibration*\" ( [paper here](https://www.sciencedirect.com/science/article/pii/S0169743918301382?via%3Dihub) ). We will try to replicate the results of section 5.3 of that paper \"*Experiment 3: application on small datasets*\". The data used here (named data set 3 by the authors) was downloaded from the [original source](http://www.models.kvl.dk/wheat_kernels). For details and references about the data check the source website or the \"Data description.txt\" file included. \n",
"\n",
"This is a regression problem. Basically we use the spectra information (our X) to predict the ammount of some chemical compound (our Y). We also implement a visualization of the \"regression coefficients\" associated with the transformation the CNN does on the original data. This can be useful for helping interpreting what kind of spectral features are more important for modeling our data (Y). Some analysis functions (e.g. PLS optimization, etc.) where adapted from this awesome spectroscopy blog, [nirpyresearch](https://nirpyresearch.com/).\n",
"\n",
"\n",
"The long term goal of this series of \"replication experiments\" is to develop and apply a CNN model/methodology on [CEOT@UAlg](https://www.ceot.ualg.pt/research-groups/sensing-and-biology) fruit spectral datasets, for a comparison with other Machine Learning models benchmarked at [Passos, D. et al 2019 (Sensors)](https://www.mdpi.com/1424-8220/19/23/5165)."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"**Practical notes**\n",
"> This notebook was developed for personal academic research purposes. Implementation errors might exist! You can use and adapt its content to your own projects, but if you do so, please be kind and cite the authors.\n",
"\n",
"> Main software setup used: python 3.6, tensorflow 2.x, tf.keras, scikitlearn, scipy, detrend.py from obspy...\n",
"\n",
"> The timestamps of the cells (that can be used as a measure of computation time) where obtained by running this notebook on a Intel i7-4770 CPU + Nvidia GeForce RTS 2080 Ti. Note: the time stamps might only be visible if you are using the ExecutionTime plugin from Jupyter's NBextentions.\n",
"\n",
"**For comments/suggestions please contact me: [Dário Passos](https://www.researchgate.net/profile/Dario_Passos)**"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Import libs and dataset"
]
},
{
"cell_type": "code",
"execution_count": 167,
"metadata": {
"ExecuteTime": {
"end_time": "2020-07-20T12:28:08.084629Z",
"start_time": "2020-07-20T12:28:08.069665Z"
}
},
"outputs": [
{
"data": {
"text/html": [
""
],
"text/plain": [
""
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"## Readjust cell width in Jupyter \n",
"from IPython.core.display import display, HTML\n",
"display(HTML(\"\"))\n",
"\n",
"import os\n",
"import sys\n",
"from sys import stdout\n",
"\n",
"import random\n",
"import numpy as np\n",
"import matplotlib.pyplot as plt\n",
"import seaborn as sns \n",
"\n",
"import scipy.io as sio\n",
"from scipy.signal import savgol_filter\n",
"\n",
"\n",
"from sklearn.preprocessing import StandardScaler \n",
"from sklearn.cross_decomposition import PLSRegression\n",
"from sklearn.model_selection import cross_val_score , KFold\n",
"from sklearn.metrics import mean_squared_error, r2_score \n",
"\n",
"import tensorflow as tf\n",
"from tensorflow import keras \n",
"from tensorflow.keras.models import Sequential\n",
"from tensorflow.keras.layers import Dense, Flatten\n",
"from tensorflow.keras.layers import Conv1D, Reshape\n",
"from tensorflow.keras.callbacks import EarlyStopping, ReduceLROnPlateau\n",
"from tensorflow.keras.utils import plot_model\n",
"\n",
"## Use liveslossplot for training visualization in real time\n",
"from livelossplot import PlotLossesKerasTF\n",
"\n",
"## Package reference versions\n",
"## Numpy 1.17.4 \tScipy: 1.4.1 \tScikitLearn 0.23.1 \tTensorflow: 2.2.0"
]
},
{
"cell_type": "code",
"execution_count": 2,
"metadata": {
"ExecuteTime": {
"end_time": "2020-07-20T05:57:35.553200Z",
"start_time": "2020-07-20T05:57:35.543218Z"
}
},
"outputs": [],
"source": [
"## Define random seeds ir order to maintain reproducible results through multiple testing phases\n",
"def reproducible_comp():\n",
" os.environ['PYTHONHASHSEED'] = '0'\n",
" np.random.seed(42)\n",
" random.seed(42)\n",
" tf.random.set_seed(42)\n",
" \n",
"reproducible_comp()"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Read the data (matlab format) using the scipy.io module and define the train and test datasets"
]
},
{
"cell_type": "code",
"execution_count": 3,
"metadata": {
"ExecuteTime": {
"end_time": "2020-07-20T05:57:39.349931Z",
"start_time": "2020-07-20T05:57:39.336454Z"
}
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"dict_keys(['__header__', '__version__', '__globals__', 'Validation_Y', 'Validation_X', 'Calibration_Y', 'Calibration_X', 'VarLabels_X'])\n"
]
}
],
"source": [
"data = sio.loadmat('NITSingleSeed.MAT')\n",
"print(data.keys())"
]
},
{
"cell_type": "code",
"execution_count": 4,
"metadata": {
"ExecuteTime": {
"end_time": "2020-07-20T05:57:42.101197Z",
"start_time": "2020-07-20T05:57:42.082218Z"
}
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Train set dims X Y = (415, 100)\t(415, 1)\n",
"Test set dims X Y = (108, 100)\t(108, 1)\n"
]
}
],
"source": [
"## Spectra (x) and target variable (Y)\n",
"x_train=data['Calibration_X'].astype(np.float32)\n",
"y_train=data['Calibration_Y'].astype(np.float32)\n",
"x_test=data['Validation_X'].astype(np.float32)\n",
"y_test=data['Validation_Y'].astype(np.float32)\n",
"## The wavelenghts for the XX axis when we plot the spectra\n",
"x_scale=data['VarLabels_X'].astype(np.float32)\n",
"\n",
"## check for dimensions\n",
"print('Train set dims X Y = {}\\t{}'.format(x_train.shape, y_train.shape))\n",
"print('Test set dims X Y = {}\\t{}'.format(x_test.shape, y_test.shape))"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"As mentioned in the paper we have 415 training samples and 108 test samples. Lets check the shape of these spectra (of wheat kernels). Note: the Y data seems to have been sorted."
]
},
{
"cell_type": "code",
"execution_count": 5,
"metadata": {
"ExecuteTime": {
"end_time": "2020-07-20T05:57:46.256396Z",
"start_time": "2020-07-20T05:57:45.696470Z"
}
},
"outputs": [
{
"data": {
"image/png": "iVBORw0KGgoAAAANSUhEUgAAA7AAAAEZCAYAAACuMcnNAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8li6FKAAAgAElEQVR4nOzdd3yb133o/88BARAgAAIEwD1EcUjUXtQeluTYjrfTJHZsp7lpm+aV3ts0s82v4za5mc3tdJvkpmnj7CZ27MZ27HjbskTtYW3JokiJ4t4ACC6s8/vjgWhK1hY3v+/X63mJxLMO8EA8z/c553yP0lojhBBCCCGEEEJMdKbxLoAQQgghhBBCCHEtJIAVQgghhBBCCDEpSAArhBBCCCGEEGJSkABWCCGEEEIIIcSkIAGsEEIIIYQQQohJQQJYIYQQQgghhBCTggSw4oYppYqUUmGlVMp4l2U6UErZlVK/VUoFlVK/Hqcy/Fgp9fXxOPdkp5T6uFKqarzLIYQQYvwppb6ilPr5NW57w3Wv1NtiKpIAVlyVUuqsUqo/GayeX/K01ue01k6tdfwGjnnVm3ml1D8opaqVUj1KqZNKqY9dtH6xUmq/Uqov+e/i6y3HaFFKbVFKfWKED/shIBvwaa0/PMLHfo/RDrhG6jNSSm1USjWMRJmEEEJMLcl7mPeNwHGuWCcqpb6hlHr9otdmKaVCSqkFN3t+IcS7JIAV1+reZLB6fmm60sbKcLPfr17gXsAN/A/gMaXUmuTxrcCzwM+BDOAnwLPJ1yc8pZT5BnabAZzSWsdG8JhCCCGEuHlfBXKUUn8Mxn0Q8B/AP2mtj4xryYSYYiSAFTdMKVWslNLnA6dki9o3lFLbgT6gJPnEsjbZinpGKfWoUmoO8H1gdbI1N3Cp42utv6y1Pqm1TmitdwPbgNXJ1RsBM/AvWutBrfW/AgrYfJmy3qWUOp4sR6NS6ovJ1zcqpRqUUn+llOpIPql9dNh+qcmW4HNKqVal1PeVUvZh6+9XSh1MPmGtUUq9Xyn1DWA98J3k+/tOclutlPpfSqlqoDr52mNKqfrk/vuVUusvU/7/A/wt8FDymH+U/Gy3K6X+WSnVBXxFKWVSSv2NUqpOKdWmlPqpUsp90fX6g+Q5u5VSn1JKLVdKHVZKBYaV9UrXKEMp9ULys9ytlCodVs4KpdSrSqkupdQ7SqkHL/N+LvcZXXb/S11DpZQDeBHIU8N6B1zH9c9QSj2vlGpPfh7PK6UKhu23RSn1daXUjuSxf6uU8imlfpG8ZnuVUsXDttdKqT9Lfuc7lFJ/ry7zIOdaPyshhBA3Rin1M6AI+G3yb/hfJF9flfy7HlBKHVJKbRy2zw3dt2itB4E/BP5OKZUPfBLjAfs3rrGsv1ZKtShjmNBWpdS8izbxJ+uMHqXUW0qpGcP2vda615+s5wLJbbddro4SYkLTWssiyxUX4Czwvku8XgxowJz8fQtwDpiHEVy6gRAwO7k+F5iX/PnjQNV1lMEONAPvT/7+OeDFi7Z5HvjCZfZvBtYnf84AliZ/3gjEgH8CUoFbMFp+z5f5X4DnAC/gAn4LfCu5bgUQBG7DeBiUD1QM+yw+cVEZNPBq8lj25GsfBXzJz+sLQAtgu8x7+Arw82G/fzxZ9k8n97djVJ6ngRLACfw38LOLrtf3ARtwOzAAPANkJcvfBtxyuWsE/BjoSr53M/AL4FfJdQ6gHviD5LqlQMf5a36J93PBZ3S1/a9yDRuu8v253L4+4INAWvL6/hp45qIyngZKMb7Px4FTwPuSZfwp8KOLrvGbyWtclNz2Exd/ntf7WckiiyyyyHJjCxfdwyTruk7gLoy6+7bk75nJv803dd8C/CPwevJveuUVtvsKF9bpf5ish1Ix7j0ODlv3Y6AH2JBc/9i11ifJfb+e/PlbGPcAluSyHlDjfY1kkeV6F3nqIq7VM8kndgGl1DNX2O7HWutj2ujmGgMSwHyllF1r3ay1PnaD5/8+cAh4Ofm7EyN4HC6I8cf/UqLAXKVUuta6W2t94KL1/1sbLblvAS8ADyqlFPDHwOe01l1a6x7gm8BHkvv8EfC41vpVbbQSN2qtT17lfXwreax+AK31z7XWnVrrmNb6HzEqptlXOcZwTVrrf0vu3w88itFdqVZrHQb+EviIurB78de01gNa61cwgvVfaq3btNaNGK3cS65yzv/WWu9JXuNfAOfHHt8DnNVa/yhZngPA0xhjd6/F1fa/2jW8kkvum/zsn9Za9yWv7zcwHmIM9yOtdY3WOojR2lujtX4t+f5/zXs/r28nr/E5jJuQh2/gvQohhBgdHwV+p7X+XbLufhXYhxHQws3ft/wNUIbx8Hjfte6ktX5ca92jjZbcrwCLzvegSnpBa701uf6vMVqDC7m++iSKEZTP0FpHtdbbtNb6Ot+fEONOAlhxrR7QWnuSywNX2K7+/A9a617gIeBTQHOy22nF9Z5YKfX3wHzgwWF/aMNA+kWbpmM8obyUD2JUTnXJrjerh63rTpb1vDogD+NpbBqw/3zwDryUfB2gEKi5zrdTP/wXpdQXlFInkl2GAhitfP4bPV6y3HXDfq/DeCKbPey11mE/91/id+dVztky7Oe+YdvPAFYOe9ARwAioc65yvPOutv+VruHVXHJfpVSaUurfldHlOgRsBTzqwsza1/t5Db8m579L1/tehRBCjI4ZwIcv+vu7DsgdifuW5MPkM8A1B75KqRSl1N8pYyhSCKPVGC68Hxh+fxXG6A2Vx/XVJ3+P0avolWQ36f/vOt6aEBOGBLBipF3wJE9r/bLW+jaMJ34nMRIavGe7y1HG2M87gdu11qFhq44BC5OtpOct5DIVhtZ6r9b6foyuss8ATw5bnaGMsZTnFQFNGF1w+jG64ZwP3t1a6/MBSz1G19JLnvJqrytjvOuXgAeBDK21B6MVWV1m32s5TxNGZTb8vcS4MOi60WNfTT3w1rDPyqONhF9/co3Hv+L+V7iGVy3nFfb9AkaL90qtdTpG9yy4vmtwscJhP5//Ll3sej8rIYQQN+ZSdc3PLvr769Ba/x3c/H3LDXoEuB9jeIobY8gPXFgXDdUtSiknxlCVJq6jPkm28H5Ba12CkSTz80qpW0flHQkxiiSAFaNGKZWtlLovGRwOYrSanp9ypxUoUFfIGqyU+kuMP+q3aa07L1q9JXmsP1NGoqU/Tb7+xiWOY00mYXBrraMY41sunvrn/yS3W4/RHefXWusERsX1z0qprOSx8pVSdyT3+SHwB0qpW5WRPCl/2JPaVoxxqFfiwggu2wGzUupveW+r8vX6JfA5pdTMZAX3TeAJfZnMxVdx1Wt0keeBWUqp31dKWZLL8mTyi8sdv+Ra9r/KNWwFfBd1tRpylX1dGA8pAkopL/Dla3yvV/LnykgOVQh8BnjiEttc72clhBDixlxc1/wcuFcpdUey5dOmjISOBTd733ITXMnzdWL0/PrmJba5Sym1Lnn+rwG7tdb1XEd9opS6RylVlnz4f74uvO6pEIUYbxLAitFkwmjhasLo6nIL8D+T697AaC1tUUp1XGb/b2K0YFWrdzPM/hWA1joCPAB8DAhgJD94IPn6pfw+cDbZNedTGGNgzmsBupPl/AXwKf3uWNYvYXS32ZXc9zWSY1S11nswkib8M0bL6Vu82/r5GPAhZWS2/dfLlOlljDGVpzC6mg7w3i7B1+tx4GcYXWHPJI/56Rs81rVcoyHJMaS3Y4wRbsL4XL+NMa73Ui74jK5h/0tew+S1+iVQm+w+dakuu5e7/v+CkfyqA9iF0UX8Zj0L7AcOYoyn/uHFG9zAZyWEEOLGfAv4m2T98MVk0Hc/8FcYD5DrgT/HuGe52fuWG/VTjPuARoxkgbsusc1/YTxk7QKWYXQTvt76pBzjPiYM7AS+p7XeMoLvQ4gxoWTstpjOlJE6/+da64KrbSvE1SilNFCutT493mURQgghhJiKpAVWCCGEEEIIIcSkIAGsEEIIIYQQQohJQboQCyGEEEIIIYSYFKQFVgghhBBCCCHEpGAe7wJcL7/fr4uLi8e7GEIIIaaI/fv3d2itM8e7HJOZ1M1CCCFG0pXq5kkXwBYXF7Nv377xLoYQQogpQilVN95lmOykbhZCCDGSrlQ3SxdiIYQQQgghhBCTggSwQgghhBBCCCEmBQlghRBCCCGEEEJMChLACiGEEEIIIYSYFCSAFUIIIYQQQggxKUgAK4QQQgghhBBiUpAAVgghhBBCCCHEpDAtA9iE1nz5dCNbu3qIJfR4F0cIIYQQQgghJq3wYIyvP3+c023hUT/XtAxga/sH+VlTJw8eqmHhjqP8xTv1VHX3ENcSzAohhBBCCCHE9dhd28l/Vp2hNTQw6ucyj/oZJqCyNBtH187nza4Qz7UFeKq1m582dZJpNXNvpocHsjxUuh2YlBrvogohhBBCCCHEhLatugObxcSyGRmjfq5pGcACpKWYuDvTw92ZHvriCV7vDPFsWzf/1dzJ440d5KdauCfLw/1ZHpa40lASzAohhBBCCCHEe1Sd7mDFTB82S8qon2vaBrDDpaWYuDfLw71ZHsKxOC91BHm2LcDjDR38e307RTYr92V5uC/LwwKnXYJZIYQQQgghhACag/2cbgvzUGXhmJxPAtiLOM0pfCjHy4dyvASiMV7sCPJcW4Dv17fxnXNtFNut3JtpBLPzJZgVQgghhBBCTGPbqjsAWFfuH5PzjVoAq5SyAVuB1OR5ntJaf/ky2y4HdgEPaa2fGq0yXS+PxczDuT4ezvXRFY3xUrsRzH6vvo1/GxbM3ists0IIIYQQQohpqKq6A78zlYoc15icbzRbYAeBzVrrsFLKAlQppV7UWu8avpFSKgX4NvDyKJblpnktZh7J8/FIno/OSIyXOoL8dlgwW2Szck8ymF3skmBWCCGEEEIIMbUlEprtpztYX+4fs/hn1AJYrbUGzk8EZEkul5qn5tPA08Dy0SrLSPNZzTya5+PRvHdbZn/bHuAHDW18r75tKAHUvZkelqanSTZjIYQQQgghxJRzoiVEZ2+E9eWZY3bOUR0Dm2xd3Q+UAd/VWu++aH0+8AFgM1cIYJVSnwQ+CVBUVDRq5b0Rw1tmA1GjZfaF9iA/SiaAyk21cJffzd2ZHlZ6HKRIMCuEEEIIIYSYAqrGePwrjHIAq7WOA4uVUh7gN0qp+Vrro8M2+RfgS1rr+JWanLXWPwB+AFBZWXmpVtwJwWMx85FcHx/J9RGKxXm1I8jz7UF+0dzJDxs78FvM3Jnp5i6/m7UZTqwm03gXWQghhBBCCCFuSNXpDmZlO8lOt43ZOcckC7HWOqCU2gK8HxgewFYCv0oGr37gLqVUTGv9zGiX6WzwLDPSZ4xaX+10cwofzPHywRwvvbE4r3WFeKE9yNOt3fysqRNXiolbfencmelmszcdl3n050wSQgghhBBCiJEwEI2z50wXj66cMabnHc0sxJlANBm82oH3YSRrGqK1njls+x8Dz49F8Nre1869z9yLz+ZjRe4KVuasZGXuSgpcBaNyPoc5hfuzMrg/K4OBeIJt3T282BHk5Y4Qz7QFsCrFugwnd2V6uN2XTlaqZVTKIYQQQgghhBAjYd/ZbgZjCdaPYfdhGN0W2FzgJ8lxsCbgSa3180qpTwForb8/iue+IrvZzlfXfJVdzbvY07KHF8+8CECBs4BVeatYlbuKlTkr8dg8I35uW4qJ2/xubvO7iWvNvmAvv+sI8mJ7kC++U48ClqancYffze3+dGan2SSjsRBCCCGEEGJC2Xa6HUuKYmWJd0zPq4xkwZNHZWWl3rdv34gdT2tNbbCWXc272NW8i70te+mN9qJQzPPNY3XealbnrWZx5mIsKaPXMqq15kTvAC8nW2YP9vQBUGiz8j5fOu/zpbPW48SWIuNmhRBiJCml9mutK8e7HJPZSNfNQgghJr6P/2gP7T2DvPBn60f82Feqm8dkDOxEppSi1FNKqaeUR+c8SiwR42jHUXY272Rn004eP/o4/3HkP0gzp7EiZwVr89eyNm8themFI16OuU47c512PlecQ/NghFc7QrzWGeJXzZ38qLEDu0mxxuNis8/FZm86M9NSR7QMQgghhBBCCHEtmgMDFPnSxvy80z6AvZjZZGZx1mIWZy3mTxb9CT2RHva07GFn006qGqvY0rAFgEJXIWvy1rA2by0rclfgsDhGtBy5qVY+lu/nY/l++uMJdgTCvN4Z4s2uEK9Xh4BGiu1WNmS42Oh1sdbjxG2RyymEEEIIIYQYfU2BflaNcfdhkAD2qlxWF7cW3cqtRbeiteZczzmqGqvY0bSD52qe44l3nsBsMrMka8lQQDvbOxuTGrmuvvZkxuJbfekAnOkb5I2uEG919fB0azc/beokRcFiVxobMlysy3BS6XaQKtP0CCGEuAKl1OPAPUCb1nr+sNc/DfwpEANe0Fr/xTgVUQghxATUMxClZzBGnsc+5ueWAPY6KKWYkT6DGekzeHTOo0TiEd5ue5vtTdvZ0biDxw48xmMHHsNr87Imbw1r8tawOm81fvvIZuaamZbKH6Vl8kcFmUQSCQ6E+nirq4et3T08VtfKP9e1YjcplrsdrPW4WJvhZJErDYtJkkEJIYS4wI+B7wA/Pf+CUmoTcD+wUGs9qJTKGqeyCSGEmKCagwMA5EoAO7lYU6yszDWm4Pn8ss/T3tfOzuadbG/czvbG7Txf+zwAFd4K1uatZW3+2hFPBmU1mVjlcbLK4+RL5BKKxdkZCLOtu4ft3WG+daYZzkBaioll6WmscDtY6XayLD0Nh8w9K4QQ05rWeqtSqviil/8E+Dut9WBym7axLpcQQoiJrTHQD0C+xzbm55YAdgRlpmVyX+l93Fd6Hwmd4ETXCXY07mB703Z+cuwn/PDoD41kULkrhgLaQtfIJoNKN6dwh9/NHX43AB2RGLsCYXYEwuwJ9vJPZ1vRtGIC5jhtLEt3sDQ9jcXpaZSn2UiRKXuEEGK6mwWsV0p9AxgAvqi13nvxRkqpTwKfBCgqKhrbEgohhBhXzYFkC6xbWmCnDJMyMc83j3m+efzxwj8mHAmzp2UPO5p2GMmg6rcAUOQqMsbO5q9lRc4K0iwjm8nLbzVzT5aHe7KMOW1DsTj7gr3sC/VyINjHM23GGFoAu8nEfKedhS4785x25jjtzHbYSJOpe4QQYjoxAxnAKmA58KRSqkRfNO+e1voHwA/AmEZnzEsphBBi3DQF+jEpyHKN/awoEsCOEafVyeaizWwu2vyeZFDP1jzLr9751agngwKjhXazL53NyYRQCa053TfI4Z4+DvX0cbinn1+2dNEXTwCggJn2VGY7bENLeVoqJWkS2AohxBTVAPx3MmDdo5RKAH6gfXyLJYQQYqJoCvaTk27DPA7xgASw4+CyyaAat7Oj6cJkUKvzVhvJoHJXk5mWOeJlMSnFLIeNWQ4bH8ox0mAntObcQITj4X6Ohwc40dvPqd4BXukMEh/2jL3AZqE8zUaJPZWZaamU2lMpSUulwGaVrshCCDF5PQNsBrYopWYBVqBjfIskhBBiImkODIxLAieQAHZCuCAZFEYyqF3Nu9jetJ2dTTt5ofYFAMo8ZazOW82q3FVUZleOeHfj80xKUWxPpdieyl3DYubBRILavkGq+wY53TdATd8g1X0D7A32Ek622AJYlKLIZmVmWirFdiszbKkU2a0U2YxFkkcJIcTEoJT6JbAR8CulGoAvA48DjyuljgIR4H9c3H1YCCHE9NYU7GdBvntczi0B7ASUmZbJvaX3cm/pvSR0glPdp9jRtIMdTTt44uQT/Oz4zzCbzCzOXMyq3FWszF3JfP98zKbRvZypJhNzkmNjh9Na0x6JUds/SG3fIGf63112BsL0DgtuATLMKRTYrBTYrOTbLOSnWsk7/2+qhSyrBbNM+SOEEKNOa/3wZVZ9dEwLIoQQYtJIJDTNwQHePy9nXM4vAewEZ1ImKrwVVHgr+MP5f8hAbIADbQfY1bSLXc27+O7B7/Kdg9/BaXFSmV3JitwVrMxdSZmnbMTHz16OUoqsVAtZqRZWeZwXrNNa0xWNUzcwyLn+CPUDxtIwEOF03wBbu3veE+CagOxUC3mplqFAtyDZeltsN35ONcn4WyGEEEIIIcZaZ2+ESCxBrnvsp9ABCWAnHZvZxpq8NazJWwNA90A3e1r2sLt5N7ubd7OlYQsAXpuXyuxKlucsZ3nOckrcJahxGJeqlMJnNeOzmlma7njPeq01oVicpsEojYNRmgcjNA1EaUz+e6injxfbg0SG9V5TQF6qMf623JFKWZqNCoeNeU47TumeLIQQQgghxKhpDhpzwMoYWHFDMmwZ3FF8B3cU3wFAc7iZ3S272dO8hz0te3il7hXACGiXZS9jWfYyKrMrKc8oH7MW2itRSuG2mHFbzO/pmnxeQmvaIjHO9Q9ydiBCXf8gZ/ojnO4d4OdNvfQn3s2YXGJPZYHLzmJXGpVuBwtcdmmtFUIIIYQQYoQ0JeeAzZcAVoyEXGcuD5Q9wANlD6C1piHcwN6Wvext2cv+1v28WvcqAC6ri8WZi1mavZQlWUuY55uHzTw+3QCuxqQUOakWclItrLhoXUJrGgejnAj3czTcz5GefvYGe3mmLQCAVSkWuOysdDtZk+FkpduBS1pphRBCCCGEuCFNgWQLrHQhFiNNKUWhq5BCVyG/V/57ADSFm9jXuo8DrQd4u+1ttjVuA8BsMjPXO5dFWYtYnLmYhZkLyXGMz8Ds62FSikKblUKbldv972ZCax2Msj/Uy75gH/tCvfxHQzvfq2/DBCx0pbEhw8ktXheVboe00AohhBBCCHGNmoP9pJpNeB3WcTm/BLDTTJ4zj/uc93Ff6X0ABAYCvN32NgfbD3Kw7SBPvvMkPzv+MwCy7FkszFzIgswFLPAvYK5vLg7Le8exTkTZqRbuyvRwV6YHgL54ggOhXrZ3h9keCPPd+jb+9VwbdpOJtRlObvOlc5svnTzb+PxHFCNHJ7ucv9M7QE3/IK2DUVqSS3csRn9c05eI0x/XKMCsFGYTWJWJdHMKHnMKHksKPouZ3FQLeTYjO3aRzUpuqmVcxpILIYQQQkwUTYEB8jz2cbsnkgB2mvPYPGwq2sSmok0ARONRTnad5HDHYQ63G8tr514DQKEocZcwzz+POd45zPXNpcJbMWrz0Y6ktBQT6zJcrMtwAdATi7MjEObNrh7e6AzxWmeILwHznXbe73dzd6abCodNgpVJoHUwyr5QL/uCvbwd6uNk7wCBWHxofYqCLKuFbKsFn8VMms2EPcWEPdnyHtOamNZEEppgLE4wFufcQISOaJRQ7MIM2XaTidK0VErSUpmdZmOu08Ycp50imxWTfFeEEEIIMQ00BfvHrfswSAArLmJJsRgtrpkLeHTOowB0DXRxrOMYRzuPcrTjKDuadvBczXOAEdTOSJ9BhbeC2d7ZzM6YzWzvbDLtmRM6+HOZU7jD7+YOvxutNaf6Bnm1I8irnSH+8WwL/3C2hRJ7Kndlurkvy8MC5/g9ZRIXCsfibE8+fNjSFeJsfwR4d7zzfVkeZjmMzNSlaalkWS2k3OC1C8fiNA5GaRqIUDcQobZvkJq+QQ6F+vhtW4DzubEdKSYWOO0scqWx0GVnSbqDmXarfGeEEEIIMeU0BwZYV+4ft/NLACuuymvzsr5gPesL1g+91t7XzvHO4xzvPM473e9wpOMIL519aWi9J9XDrIxZQ0t5RjmlnlLs5vHJVnYlSilmO2zMdtj40xnZtA1GeakjyAvtQf5ffRvfOddGiT2V+7M83J/tocIx8d7DVNceSV6TtiDbA2GiWhut6h4nH8/zs9ztYP4oZJx2mlOYbU5htuO9Txl743He6R3gRHiAY+F+Dvf08ZOmDgYSRljrs5hZ7k5judvJareDha40zCYJaIUQQggxeUXjCVp7BsiTFlgx2WSmZXJL2i3cUnjL0GuhSIhTXac41f3u8nT10/THjExlCiOpVHlGubF4jH8LXYWYTRPnq5iVauFj+X4+lu+nKxrjd+1Bnm3r5rG6Vv65rpW5DhsfyM7ggewMCmXM7KgJRGM83x7kv1u72RUIkwBm2q38cUEmm30ulo9zAi5HSgpL0x0XzG8cS2hO9Q1wINTHnmCYvcFeXuoIAeBMMbHK42Stx0ggNke6qAshhBBikmkNDaD1+M0BC6MYwCqlbMBWIDV5nqe01l++aJv7ga8BCSAGfFZrXTVaZRKjK92aTmVOJZU5lUOvJXSChp4GqrurOdV9iupANdXd1bxZ/yYJbYwvtJqslHpKKc8op8xTRpmnjPKMcrLTssf9Bt9rMfPRPB8fzfPRHonyXFuAZ1oDfKO2mW/UNrPS7eCD2Rncm+UhwzJxgvDJKpJI8GpniKdbunmtM0REa8rTUvlscTb3ZHomfNBnNinmOu3Mddr5aJ4PMFqPdwTCRgKx7jCvdYagBrKsZm7xutjsTWej1yXfHyGEEEJMeOfngM0bxwBWaa2vvtWNHNi4y3RorcNKKQtQBXxGa71r2DZOoFdrrZVSC4EntdYVVzpuZWWl3rdv36iUWYydgdgAtcFaqruNgPZ8YNve3z60jcvioiyjbKil9vySbk0fx5Ib6voH+U1rN0+3dlPdN4hFKW71ufhgtpfbfOnYUmRqnmulteZIuJ8nmrv4TVs3XdE4mVYzH8jK4EM5GVNu/HHTQIS3unt4q6uHrd09dEXjmIAVbgfv86Vzu99NeVrqlHrPE51Sar/WuvLqW4rLkbpZCCGmh2cPNvKZXx3k1c9toDzbNWrnuVLdPGqP/LURGYeTv1qSi75om/CwXx0Xrx8tsXiCv3nmKJXFXtaU+sb1CcJ0ZTPbmOuby1zf3AteDw4GOR04TXV39dC/L555kSdPPTm0Ta4jl1kZs6jwVjDHO4c5vjnkOnLH9IZ/hj2Vzxbn8JkZ2RwN9/NUazfPtHbzUkeIdLOJezI9fDA7g9Uep2SnvYz2SJSnW7p5oqWLE70DpJoU7/e7eTDHyy0Zrik7XjTPZuXhXB8P5/qIa82hUB+vJjNhf722ma/XNlNqT+X9mW7u9LtZmp4m3yEhhBBCTAjnW2DHswvxqLXAAuThP/gAACAASURBVCilUoD9QBnwXa31ly6xzQeAbwFZwN1a652X2OaTwCcBioqKltXV1d1Uueo6e/nA93bQ1WtkLy32pbG61MfaMj9rSv3jNimvuDStNa19rReMrT3VdYozoTND3ZA9qR7m+eexwL9gaMmwZYxpOeNaU9Ud5qnWLl5oD9IXT5CXauGBZEviXKc8KBlMJHi1I8SvW7t4vTNETMPS9DQeyvFyf5YHzzTvRts0EOGVzhAvtQepCvQQ05BtNXNnpod7Mt2scjunbGA/nqQF9uZJC6wQQkwP//uZozx7sJHDX7ljVM9zpbp5VAPYYQXwAL8BPq21PnqZbTYAf6u1ft+VjjVSlWQioXmntYcdNZ3srOlgd20XPYMxAOblpbOuzM/aMj8rZnqxWVJu+nxi5PXH+jnVfYqTnSc51mlM81MTqBkKaovTi1mctZglWUtYlr2MIlfRmLXS9sbjvNIR4unWbrZ0GYHabIeND2R5uD8rg5lpqWNSjokgoTV7g7081drNc20BgrE42VYzH8rx8lCOl1mXyPArIBiN8XpXDy+0B3ijM0R/QuO1pHCX38N9WR7WeCSYHSkSwN48CWCFEGJ6+MRP9tLQ3c9Ln90wqucZ9wA2WYgvY4x3/YcrbHMGWK617rjcNqNVScbiCQ43Btle3UHV6Q4OnOsmGtdYzSaWF2ewvjyT9eV+5uSkY5KbxgmrL9rH8c7jHGo/xMG2gxxsP0hgMABApj2TymwjydSq3FUUugrHJKDtiMT4bXuAZ1q72R3sBWChy869mR7uzvRQMgWDWa01B3v6ebatm9+2BWgcjGI3mbg7082HcjJYn+G64blZp6PeeJw3O3t4vj3AK50h+uIJvJYU7s70cH+Wh9Uep3yeN0EC2JsnAawQQkwPdz62jVy3jcc/vnxUzzMuAaxSKhOIaq0DSik78Arwba3188O2KQNqkkmclgK/BQr0FQo1VpVkXyTG7jNdQwHtyZYeAPxOK+vK/EMBbVa6tB5NZFprzoTOsL91P3tb9rK/ZT9t/W2AMZZ2Ve4qVuetZnXuajw2z6iXp3EgwnNtAZ5tC3Cwpw+ACoeNO/1ubve7WeSyT9rxjrGEZk+wl5c7grzYEeTcQASLUmz0urg/y8OdfjcOs/RmuFn98QRvdoV4ru3dYDbTauaeTA8PZHlY7nZM2u/QeJEA9uZJACuEENPD4q++wt0LcvnGBxaM6nnGK4BdCPwESAFMGBmGv6qU+hSA1vr7SqkvAR8DokA/8OdXm0ZnvCrJttAA26o72FbdTtXpDjrCxvjZ2dku1pf7WT8rkxXFXuxWuUGfyLTW1IXq2N28m13Nu9jdspueSA8KxQL/Atblr2N9wXrm+uZiUqObSbh+IMJL7UFeaA+wJ9hLAvBbzGz2GVOrrMtw4bdO7DGh7ZEob3X1sKWrhze6QnRF41iVYn2Gi7uz3Nzld0/7ca2jqS+e4PXOEM+2GdMODSQ0+akWPpCdwe9ly7jrayUB7M2TAFYIIaa+vkiMuX/7Mn9+x2z+16ayUT3XhOhCPFImQiWZSGhOtISoqu5gW3UHe852EYklsKaYWD4zg3Vlmawt8zEvz02KdDee0OKJOEc7j7K9cTvbG7dzpOMIGo3P5mN9wXpuKbiF1XmrcVgco1qOzkiMN7tCvN4Z4s2uHgKxOABzHDbWZ7hY43FS6XaMe0DbHomyO9DL7mCYHYEwx8JGJjqvJYVN3nTe73ezyevCKS2tYy4ci/NyR5CnW7t5q7uHuDa+Px/MzuCDORnkpkpyusuRAPbmTYS6WQghxOg61hTk7n+t4t8eXsK9i/JG9VwSwI6y/kicPWe7qKpuZ1v1u92N3XYLq0t8rC0zMhzP9DtkbscJrnugm+1N29lav5Wqxip6oj1YTBZW5KxgY+FGNhZuJMeRM6pliCU0h3r6qOoOUxXoYW+wl4GE8f+01J7KcreDxelpLHTaqXDaSRulOWdDsThHe/o53NPH4XA/B0N91PYPAmAzKZalO7jF62Kj18V85+Tt+jwVdURiPNdmzFO8P9SHAtZlOHkwx8tdmW4cKfKAYTgJYG/eRKybhRBCjKxn3m7ks08c5OXPbmB2zujNAQsSwI65tp4BdtZ0sv10B9tPd9IY6Acg121jTamfdeVGQJvlkvGzE1k0EeVg20G21G9hS/0WzvWcA2COd85QMDvHO2fUH0oMxBMc6uljT7CXvcFe9oV66YoaLbQmoDQtlfI0G8X2VErSUplhs+KzmvFZzHgtZiyX6AWQ0Jq+eIJQLE7zYJSmwSjNgxHq+iOc6huguneQlkh0aPu8VAsLXXaWu52scjtY4LJjNY1uF2sxMmr7Bnm6tYtft3RzbiCCI8XEvZkeHsr1ssotD9VAAtiRMBnqZiGEEDfn/750kh9sreX4V9+P1Ty694ESwI4jrTV1nX1sr+lg++kOdtR0EugzAoPZ2S7WlftZX+5n5UyfjJ+dwM4ng9pSv4W36t/iYPtBEjpBVloWmwo3sbFwIytyVmBNGf1umlprGgajHOnp40hPP8d7+6ntG6SuP0LkEv+fbSaFWRlLilIMJBL0xhOXPLYjxUR5mo1yRyqz0mzMcdpZ5LKTabWM9tsSoyyhNbuDvTzZ0sVzbQF64wmK7VYeyvHyYI6XfNv07WIsAezNm2x1sxBCiOv3iZ/s42xnL699/pZRP5cEsBNIIqE53hxiW3UHVafb2Xu2e2j8bGVxhhHQlmUyL0+m65nIuga62Nqwlbfq32J703b6Y/2kmdNYm7+WTYWb2FCwAXeqe0zLFNeaxoEIDQNROqMxOqIxOiMx+uIJ4loTSy42kwmH2YQzJQWX2USO1UKezUpuqoUMc4q0yE0DvfE4L7QH+VVzFzsCYRRwS4aLj+R6eb/fjW2UuqVPVBLA3rzJXjcLIYS4uo1//yZz89L53qPLRv1cEsBOYAPROHvOdLHtovGzGWkW1pT5WZdcCr1p41xScTmD8UF2N+8e6mrc3t9OikphafZSNhVuYlPhJgpcBeNdTCEuqa5/kCdauniiuYvGwSgecwofyM7gkVwvC1zT4++OBLA3b6rVzUIIIS40EI0z529f4s82l/O522aN+vkkgJ1E2noG2H7ayG5cVd1BW4+RNKfQa2ddmZ9VJT5Wl/pk/OwEldAJjnUc4836N3mz/k1OB04DUJ5RbnQ1LtjIPP+8UZ+iR4jrldCaqu4wv2rp4oX2AIMJzXynnYdzvfxedgYZU3g6JAlgb95Ur5uFEGK6O9oY5J5/q+K7jyzl7oW5o34+CWAnKa01Ne1hqqo72F7Tya6aTnoGYwCUZjpYVeJjZYmPVTO9ZKVLQDsR1Yfqh4LZA20HSOgEfrufWwpuYUPBBlblriLNMj1auUaa1prOgU5a+1rp7O+ko7+Djv4OwpEw4aix9Mf6iSfixHWceCIOCqwmKxaTBWuKFYfFQbo1HZfVhcvqwmf34bf78dv9ZNozsZmn5/+rQDTGb9oC/LKpk8PhflJNijv9bh7J9bEuwznlMk5LAHvzplPdLIQQ09Fv3m7gc08c4tXPbaA8e3QzEIMEsFNGPKE51hRkZ00nO2s72Xumi96IkY222JfGiplelhcbywxfmoxlnGACAwG2NW7jrYa3qGqsojfai8VkoTK7kvUF61mbt5aZ7ply3S7SE+mhNlhLbaCWM8EznA2dpb6nnsZwI/2x/vdsn5qSisPiwGlxYjfbMZvMpJhSSFEpaK2JJCJEE1Ei8Qi90V56Ij0MxgcveW6vzUuuI5c8Zx4FzgKK0ouYkT6DGekzyLRnTotrdbSnj/9q7uLp1m6CsTiFNisP53r5SI6XvCmS+EkC2Js3netmIYSYDr790kn+c5uRgdgyBrkyJICdomLxBMeaQuw508XuM53sPdtNsN/IcJzpSqVyRgbLZmRQWexlXl76mHzZxLWJxqO83fY22xq3sa1hGzXBGgCy0rJYnbua1XmrqcyuJNuRPc4lHTuxRIy6UB3vdL3Dqe5TVAeqOdV9ipbelqFtrCYrRelFFLgKKHAWUOAqINeRO9Rq6rP7SE1Jve5zR+IRQpEQnf2dtPe309HfQVtfG03hJpp7m2kKN9EYbiSaeHdqIZfFRVlGGWWeMsozyqnwVjA7Y/aUbVEfiCd4sSPIfzV3sq07jAnY5E3nkTwvt/vcl5yuabKQAPbmSd0shBBT2yd+spdzXX288rnRz0AMEsBOG4mE5nR7mD1nuth3tot9dd00dBstVKlmEwsL3CwtymBJkYclRRlkS7fjCaMp3MTOpp3saNrB7pbdBAeDABS6CqnMrmRp9lIWZS6iOL14SrT69UR6eKfrHd7pNoLVk10nqQnUDLWEmpWZYncxszJmUZ5RTqm7lFJPKfnOfFJM4zPdVDwRp6WvhbpQHXWhOmoCNVR3V1MdqKYnYiRfUyhmpM9gjm8OC/wLmO+fT4W3ArvZPi5lHi11/YP8qrmLXzZ30RKJ4reYeTDHyyN5XsrSJt/fFQlgb57UzUIIMbVt+L9vsqDAzXcfWTom55MA9iLxWJSnvvG/KV60jJmLl5E5Y+p222wJDrC/rpv9dd0cONfNsaYg0bhxzXPdNhYVeFhc5GFhgZsF+W5cNpnvc7zFE3FOdp9kX8s+9rfuZ3/rfkKREADp1nQWZi5kvn8+c7xzmOubS3Za9oT9/vZEeqgL1XEmeIbqQDWnu09zOnCa5t7moW0yUjOY7Z3N7IzZzPbOZlbGLErcJVhSJsd3UWtNa18rJ7tOcqLrBCc7T3K86/hQy3GKSqE8o5xFmYuGlkJX4YS9ZtcjltC82RXil81dvNIZJKZhpdvBI7k+7sly40iZHHNbSwB78ySAFUKIqas/Emful1/is7fO4jPvKx+Tc0oAe5Hm0+/wX3/9haHfU51OShZXMnPpcmYsWExa+tjO3zmWBqJxjjWFOFgf4FB9gIP1Ac519QGgFJT4HSwq8DA/3838fDdz89Jxpk7d7KOTQUInOBM8w6H2QxxuP8yh9kPUBmtJ6ARgBIBlGWWUuksp85RR4imh0FVIpj1z1FsrB2IDtPW10drXOtTNtjHcSENPA3WhOjoHOoe2NZvMlLhLhrrcng9Yp+pY0o7+Do52HOVw+2GOdBzhSMcReqO9gDG2dmnWUpZkLWFp9lIqvBWYTZP7/1nbYJQnW4xW2Zr+QZwpJn4vO4NHcn0sctkn9DWWAPbmSQArhBBT15GGIPd+p4r/9+hS7lww+hmIQQLY9xjoDXOy6i2q9+yg4eQxErHYBet9hTMoX76K4kXLyC2fjWmStCLcqK7eCIcbAhxuCA79e376HoCZfgdzcl3MyUlnTm46Fbku8j0T+4Z0quuL9nGq+xQnuk5wovMENYEaaoI1QwESgMVkId+ZT7Yj2xgjajPGibqsLtIsaaSZ07Cb7ZiUCZMyoZQioRNE4hEi8QiD8cGhJEc90R5CgyG6B7rpHOg0sv4OdAx1dT5Poch2ZJPvzB9KdlScXkxxejGF6YVYTJOjVXU0xBNxaoI1HGo/xMG2gxxoPUBDuAEAh8XB0qylLM9ZzoqcFVR4K8atq/TN0lqzO9jLL5o7eb4tQH9CM89p4+FcHx+coNPxTOcAVin1OHAP0Ka1nn/Rui8Cfw9kaq07rnQcCWCFEGLqenp/A1/49SFe+/wtlGU5x+ScEsBeQSwSoeHEUWoP7OX0vl30dLRfsN6SamPGwiWULFvOzEXLcHp9I3buiawtNMCxphBHGoMcawpyorlnqKUWwJlqZla2k9k5LsqzXJRnOynPcpGdniqB7TjRWtPS28KZ4Bkawg3G0tNAW1/b0BQzl8u2ey1cFhcZtgx8dh8+mw+f3UdWWhbZadlkpWUNZeu1pkyNzLRjoa2vjf2t+9nXso+9rXs5EzwDgMvqYnn2clbmrmRV3ipmpk/OYQ7BaIxn2gL8ormTwz3GdDx3JafjWTuBpuOZ5gHsBiAM/HR4AKuUKgT+E6gAlkkAK4QQ09e3XjzB41VnxiwDMUgAe10CLc3Uvr2Pmv27aTh+lET8wtZZb34hZZUrmbmkkrxZc6Z86+xw4cEY77SEONnSwzvnl9YeAn3DMrPazJRmOinLMpbzPxdm2DFLFuRxpbWmN9pLOBqmL9pHX6yP/lg/CZ0goRNorVFKkZqSijXFmCvVaXXisrpwmB2TtkVwMuno72BP8x52t+xmd/NuGsONAOQ4cliVu4rVuatZlbcKr807ziW9fpeajucjOV4eyvVSMM7T8UznABZAKVUMPH9RAPsU8DXgWaBSAlghhJi+/vDHe2ns7uflz20Ys3NKAHuDooMD1B87Qu2BPZzet5ve7q4L1lttdoqXVFK6bAXFi5ZO6bGzl6O1prM3QnVrmNNtPZxqDVPTHuZ0W/iCbsiWFMUMn4PSTAclmUZgW5LpoNTvxJ02fbuVCnEl9T317Gzaya7mXexq3kVPpAeFYq5vLmvy1rA2fy2LMhdNqvGzF0/Ho4D1GU4eyvFyV6YH+zg86JIAVhUzLIBVSt0H3Kq1/oxS6iyXCWCVUp8EPglQVFS0rK6ubszKLIQQYuys+/YbLC708J0xykAMEsCOCK01XY31ya7Gu2k+dQKttZH5KPkZZpeUUla5mplLKskqLkGZpneLY7A/Sm17mJr2Xk63GYFtbXuYus4+Yol3v3c+h5WZfgclmQ5m+p1DPxd507BZpNVPCDDG0B7rPMaOph3saNrB4fbDxHUcl8XFqrxVrMtfx9q8tZNq7uC6/kF+3dLNEy1d1A9EcKWYuC/Lw4M5Xla4HWPWbVoCWFVMMoBVSqUBbwK3a62DVwpgh5MWWCGEmJp6B2PM+/LLfP62WfzZrWOTgRgkgB0VA71h6g4fpPbAHmr272GwN3zBepvTRcmyFZQsWc6MhYuxOcZmwPNkEI0nqO/qo7a9lzMdvdR2GEFubXsvHeF3W22VgoIMOzP9Tkr8DkqznJQluyT7ndZJOSZQiJESioTY3byb7Y3b2da4jba+NgAqvBWsz1/PhoINLPAvmBRdvxNaszMQ5omWLp5vD9IXTzDDZuXDOV4+nJPBDHvqqJ5fAlhVzLsB7ALgdeB80oMCoAlYobVuudwxJkrdLIQQYmQdqg9w/3e38/2PLuX988cmAzFIADvqdCJB65kazry9j9N7d9J2tja5RgFGK212STmlyURQWSWlmCbBTeV46BmIcrajj9qOMLXtvdR29HIm+XNfJD60nSfNwqxsFxU5LmbnuKjISWdOros06+TpSinESNFaUx2oZlvDNrY2bOVQ+yHiOo4n1WMEs4UbWJu3FpfVNd5FvareWJzfdQR5sqWLqu4wGljhdvDhnAzuy/TgHoUsxhLAqmIuGgM7bN1ZpAVWCCGmrV/vq+fPnzrM61+4hdLMsWuQkwB2jPWFgpw9dIAzB/ZSe3Afkb6+C9Zb7WnMWLSU4oVLKF64hPTMrHEq6eShtaY5OMDpNmN8bXVbmFOtRiKp8KCRaMukoCTTyby8dBbku1lS5GFenlu6IYtpJzgYZEfTDt5qeIuqxiqCg0HMysyy7GVsLNzILYW3UOgqHO9iXlXjQIT/bu3myZYuqvsGsSrF7f50PpzjZZPXhXWEhmlM5wBWKfVLYCPgB1qBL2utfzhs/VkkgBVCiGkpntD8wY/3svdMF0e+cvuYJmSVAHYcJRJxWmtPc/bgAWoO7KG19rQxZnbY2Nl0fybFi5ZRtGARhfMWTstkUDdKa01joJ8TzT0cawpytDHEsaYgzcEBAMwmRUWui2VFGSyf6WV5sZfsdNs4l1qIsRNLxDjcfpgtDVvYWr+VmmANAGWeMjYWbmRj4UYW+BdgUhN3zL7WmkM9/TzV2sVvWgN0RmN4LSk8VlHEbf6b/3s5nQPYkTLZ6mYhhBBX963fneDft9bytfvn8furi8f03OMSwCqlbMBWIBUwA09prb980TaPAl9K/hoG/kRrfehKxx2JSrI/1s9Dzz/E8uzlrM1fy8rclTgsjps65jWfO9zDuSOHOHtoP7UH9tIXDBgrhgW0voIiihctoXDeIgrnzsdqTxuTsk0lbaEBDtYHhpa3zwXojxpdkAu9dlbN9LG61Fhy3fZxLq0QY6c+VM+Whi28Vf8W+1r3EddxfDYfGws3srloMytzV5KaMrpjTm9GNKHZ0hXiqdZu/mJmDqVpN/9ASgLYmycBrBBCTC3nuw7//qoZfO2B94wwGXXjFcAqwKG1DiulLEAV8Bmt9a5h26wBTmitu5VSdwJf0VqvvNJxR6KSbOlt4Zu7v8nu5t30xfowKzNLspewLn8d6/PXU+YpG5MEQUZm4wbqDh/gzMED1B8/TDyanFP1fECrFDllsyhesJii+YvILa/AbB3fORMno2g8wYnmEHvPdrO7tpPdZ7oI9hufdYnfwdoyP2vL/Kwu8cm0PmLaCA4GqWqs4s36N6lqrKI32ovdbGdd/jo2FW5iQ8EG3KlTv0eIBLA3TwJYIYSYOvae7eKR/9jFiplefvwHK7BMsCnuxqQLcTItfxVGC+vuy2yTARzVWudf6VgjVUk2vfWfuBfdy/GBc1Q1VrGtcRvV3dUA5DhyWJ+/nvX561mZu5I0y9i0gMaiUZreOUHdkbc5e+gAbWdqAY1SivPXKcVsJm/2HIrmL6Zw3kJySstIMUvAdb0SCc3x5hC7ajvZUdPJrtpO+iJxTAoWFnjYUO5n/axMFhd6xuU/rRBjLRKPsLdlL2+ce4M369+kvb8dszJTmVPJrUW3srloM1lpU3O8vgSwN08CWCGEmPy01vx6fwNff/44Pmcqz/zPtePWsDNuAaxSKgXYD5QB39Vaf+kK234RqNBaf+IS60Z0svRQzR721j0MEUhr9+BJXUr23A8zOGMB25u3s61hG7uad9EX68NisrA8ZzkbCjawPn89RelFN3Xu69Ef7qH+2GHOHTnI2cMHCbY2A6BMJnQiAUCKxULerDkUzltAQcU8cspmYUmVMZ7XKxJLcKghwLbqDqqq2zlYHyChwZVqZk2Zjw2zMtlQnkmhV7pzi6kvoRMc7TjK6+de541zb3A2dBaAhZkLubXoVt5X9L4x/Vs42iSAvXkSwAohxORW2x7mr39zlJ21nVTOyOCfHlxMkW/87nsnQgusB/gN8Gmt9dFLrN8EfA9Yp7XuvNKxRqKSjPb3UP/Mzwgm3qIn9ShRr5HwJyWYgjM8A3/WZnzLHuZ4tJWtDVvZ2rB16AZuRvqModbZZTnLxnSsWKijjfpjRzh39BB1h9+mN9ANgCklhUTcGN+pTCayZpZSUDGPvFkV5JZX4PL5x6yMU0WwL8qOmg62Vrez9VQHjYF+AEozHdwyK4tbZmeycqZXMhyLaaEmUMPr517n9XOvc7zzOGAkgbq16FZuLbqVCm/FpJ6XWQLYmycBrBBCTF57znTx0R/uJtVs4i/vnMNHlhdiMo1vvT7uAWyyEF8GerXW/3DR6wsxgts7tdanrnackagkYx39tPzTfkgY7z2RHSTs2UOPdRe9GWfQtgTEwdbuwmNeRFb5B+mfuZCq5h1sa9zG3pa9DMYHsZvtrMxZaYydLVhPnjPvpsp1PbTWBFqaqD9+hPpjR6g/dngooFWmFLRODCWFcmR4yZtVQU7pLLJLysguKcPmGLt5nCY7rTU17b1sPdXOllPt7K7tZDCWwGYxsbrEx6aKLDbNzpLWWTEtNIWbeOPcG7x27jXebnubhE6Q78xnc9Fmbi26lcWZi0mZZPNcSwB78ySAFUKIyesrzx3jyX31bPniRrImyGwd45XEKROIaq0DSik78Arwba3188O2KQLeAD6mtd5xLccdqUoyEYkTbQwzWBcicq6HSF2IRG8UrWIM+Grp9e0j7DrAoNeY+s7UY8IVLMTn30TGkgc5HG9lW8M2tjVuozHcCECpu5R1+etYV7COpVlLsaaMXbIlrTXB1hYaTh6jMbl0NzcZK5UixWx+N0EU/z97bx4cR97e9326577vATADDG6ABEmQIHgtyd0ld18d76HIlhzHkZI4SjmyZEelRFVKnCi2XstOYimvZadKjlWuWLZjVUp+HZUj6T323ffdXe4ud3mBJ3jgvufAzGDuu2e680eD2OUuuS+XIAlwtz9VXWBxenp+MxhM93e+z/N9wNXWTlvvAMGePoK9/QR7+rC63C+0i/K8qDZaXFrc4Nx0indnUiymy4Dqzp4dDvLaniBHerwY9Vrv7POk0ZRZL9TIlBtbW64qUZNaVBstqlKLZkt+4D4GnYjZoMNsELEY9bgsBtwWAx6bAY/VSJvTjM2k36FntPvZqG5wbvUcb628xcX4RSRZwmv2crbrLK9FXuNEx4nn+jn4pGgCdvtoAlZDQ0PjxeWv/uEFWorCn/7qyZ1eyhY7JWBHgX8D6AAR+LaiKL8jCMKvACiK8oeCIPxfwM8D95tamz/uIuJpnCRbksxf/P4FQiMddI34aOtxIIgCzY0ajeUCjeUC9aUCzWSFpjFP2X+Lsu8qZe9dZFNDdWcTdtziKMGBv0yhfz8frF/kfPQ8E+sTSLL0gDt7KnyKTkfnttb8JFQKeWIzU8Rm7hGfmSIxP0uzUQdA1OsRRZFmo7G1v9nuINjTi7+rB3+kB39XN77OLm2Mz49hMV3m3HSSt6eSXFrI0GjJOEx6Xh7y8/qeNs4MB/DZd+9YkheJRlNmMV1mPlViLlliPlViNVMhlquxXqzxqI8zUQCrUY9e99EXNIqiplPXpNb9YoyH4jDpCTpNhNwWurxWujxWIl4rvX4bvX4bFuOL5TY+K0qNEudj53l7+W3ei75HWSpj1Vt5ufNlXo+8zunwaRxGx04v86FoAnb7aAJWQ0ND48VElhVG//6b/NzhML/zs89/XM6j2BUlxE+Lp3GSLCwt8+Y//h7r0iAgYjRB5x4fkf1+uka8OH3qXFC5IlFfKdJYylNfLFCP5qja5yj7b1L2YWKsFgAAIABJREFUXafuUh1OXU6HPRvG53sV59hfYVJMcz56nvPR81vubLezm5Ohk5wOn+ZI25Hnlmz8ceRWi/TqMom5GRLzMyQW5kivLH0UCKU3oDPokRoNlM2eWgBnIIgv3IU33IWvM4I31IknFMbicGqO7Sco15t8MJfm7SlV0CaLdQQBDkc8vL5XLTXe0+7QXrfHoCa1uBMrcGstx51YgbuxArPJIlLro8+ssNtCt89KyG1RN5cZv92Ex2bEZzPishiwmnQYdeIjX3NFUWi0ZKqNFoVqk1y1QbYikSnXWS/USeRrrBdqxHJVVrNVMuXGA/cPuy30B+0MBu0MtzkYbLMz1Ob4Uju3jVaDS/FLvLXyFudWz7FR20Av6jnefpzXIq9xtussAWtgp5e5hSZgt48mYDU0NDReTJbSZc586xy/+/MH+E+O7p6ARk3AfpJ6kT+9fo7jc9+jdjfGanWElcZhSi0fAO6gicj+AJF9PsKDbvSbDosiyTTWitQX89SXCpRji5RcNyn7b1Hx3kY21KElYI47ccnDBHp/htLew1zcuKq6s4kJaq0aBtHAeNu46s6GTtHv7t8xQSM16qSWFkkuLZBcmie5uEB6dWmr3FgQRYxmVdBLtRqy/JGwNdvseDrCuDtCuNs68LR34G4P4Qq2YXG6vvQiTZYV7sQK/OjeOm9NrXM7WgCgw2XmzHCQV4cCvNTvw2XRxiApisJqpsrVlQzXlnPcWM1xL16guWmN+u1GRkIuRjqc7O1wMBC00+e374j7Wao3WdmobDnBC6kSc5uOcE36qES522dlb7uTvR1O9oWcjHa6dk1fyfOkJbeYTE/y9srbvLXyFivFFeCjROPXul6jx9Wzo2vUBOz20QSshoaGxovJd2/F+dv/zzW+82un2R/ePbPfNQH7CWbLNV6+PIUInHBa+AZxvrr2XUyTF1jJ97DaOExU2k9L1qPTC4SHPET2+Yjs8+Jus24JM6WlICXK1BfzVBfT5NKXKTtvUfZPUnesAqArmbCm2vDYDuPZ/3PMeAQ+jH3IB7EPmMvNAdBmbeNU+BSnQqc4ETqB0+jc1vPbLnKrRTYeJbm8SHp5kdTKEhtrKxRSya19BFGHyWpVy5AliUa18sAxDCYzrmAbzkAQZyCIwxdQ/+0P4PAFsHk8iC9Y0Mt2WS/UtkqNz8+mKW/OnT3U5eb0YIATfV7GujxfipLUmtTi1lqeq8tZrq1kub6SJV1SnU2bUcfBLre6dbo51OWmzWna9V+ItGSF1UyFmfUi04ki9xIF7sWLLG2Ut0qbgw4To50uDnW5GYt4GO104TB/eb7AUBTlgUTje5l7APS6enmt6zXORs5ywH8AUXi+/eOagN0+moDV0NDQeDH5vTem+BfvLXDnd34Kk373XINqAvYTKIrCvXKNv0jm+E4qx2yljgAcc1r5GUOeryV+SPDenxFLmFmpj7HcOkmuro6icfjMqpgd8dK5x4PRrH/guM1UlfpintLCLBu59ym7blP23UE2lEEBY8qDvdCFr+0UyuhrTEiLfBD7gIuxixSlIqIgcsB/gFOhU7wUeon9/v3oxd1RitioVkivrrARXSETXSMTXWUjukohmVRTjzcxmEwYLKq4lZtNGrXqA722oDq7do8Pu8+Hw+vH4fv4FsDh82NzexDEL2YQUqMpc30ly/m5NO/Pprm1ps6dNegEDoRdHO31cjjiYazL/cK7di1ZYSFV4uZanhurWW6u5h9wV3t8Vg5HPBzu9jDe7WGozYFuh6PbnyblepN78QK31vLcjua5uZZjPqUGfwkCDLc5GO/2cKTHw5FuL50ey64X60+LWCnGO6vv8M7qO0wkJmgpLXxmH2e6znCm6wzHO45j0Vue+To0Abt9NAGroaGh8WLy1//oMsline//+ss7vZQH0ATsj2G6XOM7m2L2XlmdCXvUaeMb1gZfT52jc+Y/UFhaZqVxiBX5JGvVfUgtPaJOoGPATfc+H937fXg6rA9ceCqKQitTo7aQJTP3AdnKRcruO1Rd8yDKCJIBy3oQR60PT9erxAf7uVi8xYexD7mdvo2CgsPg4Gj7UV4KvcSJjhN0O7t33cVtU5LIryfIxqNk41Fy63Fy6wlyiTjFdOoBcSuIImabHYPZgqgTUWQZqdGgXirRakoPHFen1+PwB3AG2nAFgriC7bjbO3C3qaXKJusXJ1wqX5W4tpzl8lKGy4sZbq3ltno9w24LB7tc7A+7OLC5ua27M9k1X5WYXS8ys17ibjzPnViBqXiRqqSWnttN+i0H8nDEw1jE/aUMuMpXJG6s5bi+kuXqcpbrKzlK9SYA7U4zx3q9HOv1crzXy0DQvuv+5p8F+Xqe96Pv8+7qu5yPnqcklTDrzJwIneBs11le6XwFv+XZzLTWBOz20QSshoaGxovJkX/4I84MB/jWf3xwp5fyAJqA/RzMVVRn9rupPLdLVQDGHFa+7tLxjdwEPXN/RmvuXeKVHpabL7Eiv0Sm7AHA7jXRvc9HZJ/vU+7sfZr5OuX5KOmZH5CTrlB2TyHZ1gHQVRxYEyGcrSFMfSe512XkcuYGF2IXiJXVwKg2axvHO45ztP0ox9uP02HveGavxdOg1WxSTKfIJ9fJJxPkU+vkk+sUNn9W8rkH9hf1eqwOF0arFZ3BALKC1KhTKxWolUoP7Gt1ubcCpbyhTvydEXxd3di9vhf+gv9+gNH1lSzXV3PcWsuxmqlu3R5ymRlqdzDc7mC4zbGViPusha2iKBRqTTXQKFNheaPC4kaZpc1+0PVCfWtfh0nP3pDa/znS4eRQl5v+gH3HB2PvRlqywnSiyNXlDJeXslxa2CBZVF9Ln83I8T4vJ/p8nOjzMfglELRSS+LK+hXeXX2Xc6vniJVjCAgc8B/YcmcH3ANP7XXQBOz20QSshoaGxotHslDj2P/6Fr/9MyP80qnenV7OA2gC9glZrNT5Tkp1Zm8WVfGwz27m614bX69OMbTwFwgzP6CYb7LSGGdZfJ21Yh9SU7flzkb2eene58Mbsj30YqtVapCfvkNy/vsU5GuUPTNquTFgzLVhTXbikAep9x/gTqjJROYGl+OXydazAITtYY60HeFo+1HG28YJ28Mv1MWtVK9RSKcopJJbInfr36l1asXCA/ub7Q4sThcGkwkQkOo1yvksjXJ5ax+T1YY/0k2wp39rzq2vM4JOvztKsZ+UXKXBnViByWieqXiBqUSR+VTpgVRel8VAt89Ku9NMu0vd/DYTDrMep8WAw6zHqBfRiwI6UUQU1HLmelOm3mxRabTIV6WtLVNqkCrVSZfqJAt14vnallN4H4/VQLfPRl/AxlCbg6E2O4NBx5eqFPZpoygKK5kKlxYyXFzc4OL8BrG8Wh3it5s42e/jZL+PUwN+urxfnEqEh6EoCjPZGc6tnuPc6jlub9wG1M++v3vi73IqfGrbj6EJ2O2jCVgNDQ2NF493ppL80r++wrf/5ksc6/Xu9HIeQBOwT4GVap3vpfJ8N5XnSkEVSwNWE1/zOfm6kGB0+Q2Eme/Tit8j3tjDivgay9JRMgU7AHaPich+H92f4c4CtKp1Nu59SGrhDQrCTaquRRSxCbIOcyaCNRnC1uon393LVETiWv42E+sT5Oqqkxm0BhlvG2c8OM6h4CEG3APoXuCwpHqlTD65rpYlJ+LkEjFyiTjZeJRSNrO1n6jX4/QHMNscCKJIo1ohn1zfmnurN5oI9vbTMTBI+8Aw4T0jOLzPphzxeSK1ZJY3yiymK5s/y6xkKqwXaiTyNQq15o8/yGdgNogEHWb8diMBh4kOl4Xw5siasMdCj8+6a8uZv0jcT2m+sJDmw/kNPpzfILXp0Ea8Vk4N+Dk94Odkvw+P7Yv9+0hWkry39h7nVs/xa2O/xrB3eNvH1ATs9tEErIaGhsaLxx+8Pcu33pxh8ps/uetCJTUB+5RJ1CW+n87zvVSOD3MlWgqETQa+GnDxVXOD47G30M98H5bep9hwsCKfZkX3FVazYSRJQBQF2vtdRPZ5iYz48HfaER5RVtmsl0nde5vU4psUxVvU7FEQFISWEUumD8t6B9ZahFwoxEyvxM3mAtfWr5GsqonBDoOD0eAoY4ExRgOjHPAfwG60P8+X65lRr1TIxFbJRNdIry6r2/LiA8LW4Q9g9/jQGfTUiiWy8ehWr60r2EZ4zz469+4nsn8UV7B9p57KM6PSaJKtSBSqm1utidSSacnK1mYyiBh1IiaDDotBh8ti2NrMhkfPT31RkGo1yvkclXyOeqVMs15HatRp1usoioIgioiiiKjTYTCbMZqtGC0WjFYrVqcLs82+68LEFEVhPlXi/GyaD+ZVh7ZYbyIIsD/k4vSgn5cH/Iz3eHZVouBuRROw22c3nJs1NDQ0ND4fv/rHV7kXL3DuN8/u9FI+hSZgnyEZqcmb6TzfT+c5lylSlxW8Bh1f8Tn5aZeRVzOXsc1+H2beoFUpkmjtZ8X0dVaqB0hn1PAai8NA14iXyF4vnXu92FyPDrVp1HOk598mtfAmRWWSui0BgNA0Ycn1YU62Yym0UXW0M9cnc9e5wY3MJPO5eRQUBAT63f0cDBxkv38/B/wH6Hf375qk46dBtVggubhAYmGW9YVZEnOzFDdSAOgMRnzhLixOJ81Gg43o6laZsjPQRmT/QboPHKR7dAyLY2fHGWk8PlKtRnJ5kdTyIrlEdCtErJBKItVr2zq2IIpYHE5sHi8Or28rJdvV1o6nPYSrrR2zbWe/FGq2ZG6u5Tk/m+b8XIrrKzmasoLZIHKiz8crgwFeGfLTH/ji988+CZqA3T677dysoaGhofHjeeX33uFA2MU/+8XDO72UT6EJ2OdEudninUyRN9J5frRRINdsYRYFXvY4+EmvnZ+ozdE+/12Y/j5k5im3PKyav86K8Apr6SDVivq78IXtdO310LXXS8egG8NnzAVtNDbYiH1AevZNCvVb1CwxEBS15LgYwZzuxJzyorT8rHTomAtXuCPGuZ25Q6GhCjezzsywd5gR3wgjvhH2evfS5+7DIO6uUoLtUNxIE5uZIj57j7V7d0kuzqMoMjq9Hn+kB6vLg1SrklxeoFGpgCDQ3jdAz8HD9Bw6Qsfg0Jdubu1uRVEUsvEoa/duE713h8T8LJl4lPvDVvVG02ZadTuuYBtWlwery43V5cJktatjnkwm9EYTCKDICoosI7eaSPU6jWqFRq1KvVymWshTKRSoFHKUsxmKG2mKG2lqpeIDa7I4nPg6I5tbF/5ID4Hu3h0TtqV6k4vzG7w/m+L92TQLabXtIey28MpQgFeHApwa8O26cqGdQhOw22c3n5s1NDQ0ND5NoSYx+s03+c2fGuZvnx3Y6eV8Ck3A7gCSrHApX+KNdJ430wVWauoc1FGHhZ/wOfmKLs/BlR8gzv4AVi6gyAppw2FWLf8RK9W9xONG5JaCqBfo6HfTtddD5x4vgYjjM1Ncm80iucwE6YVz5DITVAwLKDr1sXV1F+Z8L6akH1PaRUG0shRqMhssMatLMlWYo9pUw6oMooF+dz9DniGGPcMMeAYYdA/it/i/EA5OvVImOn2X1TuTrNy+SXJxHgCTzU5bbz8Gs4ViJk1qcQFFkTHb7HQfPEzvoXF6D41jdbl3+Bl8uShm0izfvM7SzWus3p3cSq+2utx0DA6rgV29/QR7enH4Ao/1Hr3/2fck72epViOfTJDd7M3OxqNsrK6wsbZCvfJRoJgr2Eawp5+2vgHaB4Zo7x/EZLV97sfbLquZCu/NpnhvJsUHcxuU6k30osDhbg9nhgOcGQqyt8PxhfjbfhI0Abt9XpRzs4aGhoaGysWFDf7av7jIv/6lo5wZDu70cj6FJmB3GEVRmCrX+OFGgTfTea4WKihAwKjnda+TrzhEXk1fwjH3XZh7C2o5JCzEXD/Hmvgyq5kQG+vqHE2TVU942EPnsOrQuoKfnfQqy01K5Smy61fIrV2gULtNw7i+dbuh3IY534Mx5UK/biIj61gJSCwHKyxai8y3EqTrH/WUOo1O+t399Ln66HX10ufqo8fVQ8gWeqHDoiqFPMuTNzZF0lXKOTXlOdDdi7u9g2ZDIjE/Q7WQ33Jne8eO0HvoCG39A5o7+5SR5RaxmSkWrl1h8doV0qvLANjcHiIHDtG5dz+de/fj6Qhtvf8rhTwbayubY5qSFNJJytkMtVKRWrlEraz2v8qtFrLcAkVBEER0BgM6gx690YTJYsVkt2O22jA7nNjcHmxuD3aPF7vPjyvYht3tfWRPrKIolLMZUitLJBfnSS4tkFyaJ5eIb+3jDXXSMbSH8PAIoaG9eMOdz1U4Si2Za8tZ3p1JcW46xd24WonR7jTz6lCAs3sCnBrwf6ncWU3Abp8X8dysoaGh8WXmX55f5B985y5XfusrBByPbl/cKTQBu8vYaDR5J1PgzY0C5zIFCk0ZvQDHXXZe99h5rbnM8PIPEGbehPVJACqmftacP8+aNMZqwkEppwYR2T0mVdDu8dA57MXu+fFvQEnKUShMkk/eILc+QalxF0n/kUjVV32Yil0YM14McQOVVIOEsUk0WGfFL7Fmq7IsZMi1PprLahANRBwRup3ddDu7iTjVf0ccEQLWAKKwu0JwPgtFUUguLbB04yoL1yeIz0yhKDJWp4uOwWEMJjPZZILk/ByKImNxOOk5NE7v2BF6tN7ZJ6ZRq7J08xrzVy6ycOMqtWIBUaejc+8+eg6O03PwMP5ID4IgUMykic9MEZudJrW0QHp1+VMzhe/3rJodTsw2OyarDYPZjKjTIep0CIKIIrdoShItSaLZqFOvVKiVS9TLJarFAuVclpYkPXBcncGA0x/E0xHCE+rEG+rEG+4kEOl5pLtaK5VILMySmJshPquu+37vtdnu2AwTUwPFgj19iLrn94XIeqHGu9Mpzs0keX8mTXHTnR3v9nBmOMiZ4QB72r/Y7qwmYLfPF+HcrKGhofFl4je+fYPzs2ku/9ZXdnopD0UTsLuYpqxwpVDmrY0CP9ooMFVWA2fCJgNnvU7OWFucTl/EvfBDmH8bqhkUBfLeM6xZvsZaeZDomp5aWR2X4gpYCA97CA+7CQ95PjMQ6uM0GhmKxbsU0rfIb0xSLk9RE9dAkAEQZD3GUgfGYjuGtA39ukhlvUKimiHub5BoE4l5IWatExMLSLS2jm3SmehydG1tnY5O9ae9k7A9jEG3u52earHA0o2rzF+7wtKNq9QrZXR6PaHhERz+AI1KhejUHarFAoIg0jE4TN/ho/SOHSHQ3fuFvvDfLuVclvmrl5ifuMTy5A1akoTZZqf38FH6x4/Rc/AwRouVfHKd1Tu3WLl9k+jU3a1QLr3BiD/SjT/Sg7+rG19nBHdbB3afH71h++8rRVGolUtb/a9bs4qT62TjUTLx6AMC1xloI9DdS7Cnl7a+Qdr7B7G5PQ89bjYeJTp9l+jUXaJTd7ZcWqPFQnjPPrpGDtC1b5Rgb99zc/jvu7PnNt3Zex9zZ88MBzi7J8ipAT920xcn9A00Afs0+KKdmzU0NDS+yFxbyfK3/vgaezsc/KtfOrbTy3komoB9CKuZCp2ezy6/3QmitQbnMkXezhR4L1Ok2JIRgUNOK6+67bwsxxmPvYNp/i1YuwxyE0VnIe3/BlH9q0QL3cTWFBpVVUC6ghbCg25CQx46Blw4vObHfs6yXKdcnqeYu0shdY9ycZqyNIckpj62k4ihGsBYDqLP2dAnBYREg2y2wHo1Q8JSJhmysB40kHDJJEx16uJH4lZAoM0SIOz8SNCGHWH1pz1M0BrcVe5tq9kkNn2X+auXWbh2mWw8BoC/q5tg7wCCKJBaXtzqqbX7/PQdOkLf+FEi+w5iMJt3cvk7jqIoZGJrzE9cYm7iIvHZaVAUnIE2Bo4cp//ICTr37kOq11iZvMnizass37pOIaWOhbK63HSOHCA8tIfQ0F4CPb3o9Dv3BYgstyimU6RXV0ivLJHaTELOxmMoivrlj93ro2NgmI7NNbf19qM3fnpWaymzwdrUHdbu3mb17iSZ6CoAJpuNrpFRug8cInLgIJ6O8HP73Erka7w7k+SdqRTn59KU6k0MOoFjvV7ODgc5uydIn9+26z5HPy+agN0+moDV0NDQ2P1MJQp86wcz/OjeOn67kf/jr41xasC/08t6KJqA/QSpYp2j/8uP8NtNjHe7Ge/2MN7tZX/YuatmJkqywvVCmXPZIu9milwvVJABiyhwzGXnlMPAycocB6M/wjD/NqTuASAb3aT9P0tUPEms0El8VaZeVR1au8dER7+LjgE37X0ufGEbou7zCcRms0S5skhpY5rSxjSV0iJVaZmasIYiNj7aUdZhqPkwVDwYSnZ0GyJCtEIhniFdyZM0tVh3CyTdkPSIJD0iGZv8wGMZ0NGh9xG2dhDx9BIJDNLt7KbLqbq5O52UnIlFWbh2mYWrl1mbuoMiqyXFnSMHsDgclDIZVu9OItWq6AwGukYOqL2zY0fwtId2dO3Pi2ajwdrUHRavXWHh2hVy66rTGOztZ+DICQaOnsDX1U16ZYnF6xMs3pggNjOFIssYLVYi+0eJHDhEZN/B594v+qQ0alWSSwusz8+RmJ8hPjdNfl0deaXT62nrH6JzzwjhvfsID488tPS4nMtuuc7Lkze2RLwzEKR7dIye0TEi+w9htj+fpONGU+bqcpZ3ppOcm04ys662EHT7rJzdLDU+0efDbNg9n6GPiyZgt48mYDU0NDR2J9lyg+/fTvBnN6JcXspgN+n5lVf7+S9P9mDbxRVVmoD9BPmqxHduxbi6nOXqcpbljQoARr3IoU434z0ejvaootZl2T3lrXmpycV8mfPZIu9nS1vlxhZR5KjLygmrwLHiFIejP8S6+A5klwCQzT42/N8goXuJWKmL+BqU86rQ1Jt0tPU4ae9z0t7roq3XicXxaXfocVAUmXp9nXJmnlJqnkp+gWp1lZocoyHGaRlKD+wvSlaMDR+GhhtD1YY+r6OVrrGR3CCVzbOuVEi6BZIuWPcIJNxQNX8kXnSKQEh20m1sp8/Vz96uMfZ0H6HH1bMjc21rpRJLt66pwUPXJ6iViog6HaHhEXzhLppSndjMNNnYGgDutg66Rw/RfWCMrv2jOz5L9Glxvzx26eY1NTX4ziTNRh29wUjkwMHN8uqjmG02lidvsHBtgqUbE5Syah92sLef3kNH6Dk4RsfgHnT63fvh+nko57LEZqeITd8jOnWH9YU55FYLQRAJ9vbRtW+Urn0H6NyzD6PF+sB9FUUhv55g6dZ1lm9dY+X2LRrVilqyPrSH3oOH6R07QrCn75EBU0+b1UyFczMp3plK8uF8mpokYzaIvNTn2+qd7fY9/8TlJ0ETsNtHE7AaGhoau4tmS+a3//wO355YRWop9AVs/KVDYf6Ll7pxW5/sWv95ognYT9Bqyrz5L2/T0e8mNOhGdhm4sZZjYinLleUsd6J5mrKCIMBwm4OjPV6O9no51uOl3bV7SkDTjSYXcyUubG73yjUUQC/AqMPKUZPMeGWWo7FzdCy+CfkVABSji2Lb6yQMp0nU+oknzWzEqiiy+l5wBiy09TjVrdeJv9OO/jNm0T4ujUqOUmKO8sYC1cIK1doK9WaCuphAMqe2xv3cR9eyY5SDmBQ/xqYbQ8VMNdsgsVEgnsuw2kqzYioT80LcCy2dKm6NLYHeupO9pm72tx3k0J4zDPSOP9eU5I+n6C5cvczGmvraezrChIb3YjCZyCfXWbt3B6lW3RIxapDPAcJ7Rl6YMChVsMZYu3eb1Tu3WL07SXlTjHo6QnSPHqbn4GG69h0gn1xXhe2NCdbu3UVuNTFarPSMjtF7+Ci9h8Yf2jO6HeR6nWY8jhSP00xv0CoWkAtFWsUCSkMCWUa5n0psNCKazAhmE6LVhs7tRudxo3e70fkDGIIBhIeU/z4JUq1GfG6a1bu3Wbs7SXx2ilaziSCKtA8M0b3/IJH9B+kY2vupft5Ws0liboalm1dZvHGV9YU5QE1p7jk0Tt/YEbpHx57byJ6a1OLSYoZ3plR3dmnzS8Fev41XhwK73p3VBOz20QSshoaGxu6h2ZL57759k7+4GeMXjkf4hWMR9oWcL0QV2300AfsJkmtZ/uQfXUDXVMWowSTSMaCK2fCQB3u7lVvxvCpolzJcW85Sbqh9m11eC0e7VUF7tMdLf2D39H/lpCYThQqXciUu58vcKFaob4rSsMnAIbPAISnKWHqC0eU3cCZvqHcU9Uj+MVL2syRa+1kvBFiPy5RzqqAURAFvh41At4NgxEEg4sAXtmMwPZ2LUUVRaJUa1JJxyuklKvllqpUV6lKMOgkkcxLJnEbRNbfuIyh6TEIYi7Ebi74DsWZjPV5iMbHOfGWVGV2aea9EzaT+bmx12Ft2M2oZYDzyEuOHv46tPfxU1v845JMJVcxen2D1zi1akoTBZKZr3wHc7SHkVouN1WXis9M0JfV193SEaO8fon1gmPb+QQKRnl3RQ1stFkguLpBYmCU2c4/4zBTVzURdq8tN175RIvvUsl+90cjqnVubI4qubbmsvs7IVtBVaGjvU3FZ5UaD+t271KZnqM/OUp+Zob64QCuVfuj+gtGIYDKpjqUogiCgNBrI9Tp8Inn44+j8fgxtbRg6OzFGIhi7IxgiEUwDA+i93idev1SvEZuZUsuGJ2+SmJ9FUeSt90n36Bjdo2N4Q58uoy7nsizdvMbi9QmWbl2jXi4j6nSE9+yjf/wYfePHnmvJ+lK6zLnpJO9Mp7i4sEG9qbqzJ/p8nBkKcGY4SI9/97izmoDdPpqA1dDQ0NgdSC2Z//ZPbvDdyTj/41f38Ddf7d/pJT0RmoD9BMVikfPnzzNzZ4FKCgwNFxbFBzU1sVdvFGnvcxEechMa9ODrsjOdLnF5MbMlajfKqsjw2Ywc6fGoLm2Pl5GQE8Pn7Cl9VjRkmdulKhP5MlcLFW5ysv8DAAAgAElEQVQUKizXPnI5u0169gslDlQWGUlPsHf1bTpLCwgARgdl30usG0+SbA6SyntIJmRqJfXCXhDA3WbF3+XA32nH32XH3+nA6ny6JQmKrNAqNpCSJSqpVSrZBcqlJWrSCnX9Gg1rgoY1CfeDoRQBsxjGZt2D1TZAOi0zuxrjdmaO22KcNae6fkNTYThtZIwIR9uPceTIz2DfM4LwHMaXSLUay7dvbjmR+aQ6l9cZaKNzZD8OfwC52SQbWyM+N7PlZiIIeDrCBLp78XdF8Ia68IY78bSHHhoKtF3qlTKZ2BqZ6BqZ6Cob0VWSSwsU0x+FeHlCnYQG9xAa3kN4eB8Gs5nYjFoiu3pncst5NtlsdB8Yo+fQYXpGD+PwbT8woFUqU7l8mcrVCarXrlO7fRtlU3gKViumwQFM/QMYOsMYOkIYQiH0AT86pxPR4UA0PTqhW2m1kCsVWrkcrWyWVjZLM51GSiRoJhJI8QTS6iqNaBSaH32xovN6MQ0MYBoawrx3D+a9ezENDDyRa1uvlFm9e5vlW9dZvnWdbDwKqD2wvYfG6Tk4TuTAQYxmywP3k1stYrNTLF67wvzH3f9QJ/3jx+gfP0ZoaO9zG9VTk1pcXNjg3HSKd2dSLKbLgNo7+3F31mrcuVJxTcBuH03AamhoaOw8uUqDv/Onk7xxJ8H//PW9/I2X+3Z6SU/MjghYQRDMwHuACdAD/6+iKL/9iX32AP8KOAz8lqIo3/pxx32aJ0lFUUilUszMzDA9PU10eR19w4lN8WORfUhF9QJPZxBp73MSHvIQHnIT7HaynK8ysZTh8qIqaFcyasmcxaBjLOLmSLeHIz1exiJuHObd00ebkZrcLFS4VawyWapwu1RlqfqRqLWLsEeoMFiPMZC7x2DyCgOlObpqCfSigZLnBCnTcdKtQVKlAOm0bmsmLYDFacQXsuEL2/GF1Z+eDhuGp1CC/EnkWhMpWUFKFCgl5ygVpinX56iZV6g7VpCsya19TUIIp+MQon2YuWieW6szXK/OMm8poghgaijsi+kYVyKc7nqFPce/imVk5Jn3EyqKQi4RY+mm2te4dneSWlntFXYGgnQM7sEb7sJgNFKvVNhYWyG1vEA+lYT7f7uCgM3lxuEP4PD5sXt9mG0OzHY7Zpsdg8mMqNch6vSIOh1yq0VLkmg1JaR6nVqxQLVUVOeeZjMU0ikKqSSNamVrnaJOh7s9RKC7l7befoI9/XhCIQqpJIn5WRLzs8Rnp7aChgwmM+E9I3TtU9NzAz292x4FoygK9Xv3KL33HuXzH1C5cQOaTQSDAfP+/VjGxrAcOoh5ZARDKPRcekGVZhMpkaCxuER9bo76/ByN2Tlqs7Molc3Xz2DAPDSEefQAlgOjWEYPYOz7/L2q+eQ6y7eus3hjguXJm2owmF5P58gB+g4fo3/8KK5g+0Pup7r/81cvs3pnErnVxOxw0jd2hP4jx+kZHftUz+2zZHmjzLubY3ru984adSLHer28OhTg1eEAg0H7c61s0QTs9tEErIaGhsbOsZQu80cfLPLvJ9aoSi3+3jdG+K9O9+70srbFTglYAbApilISBMEAnAd+XVGUix/bJwh0A38JyD5vAftJyuUys7OzTE9PMz8/j1STMcte/OYIupqD8kYLFNDpVUEb2hS0bb1ONqrSljs7sZzhbqyArIAowJ52J0d6PIx3q05tyG358Yt5jhSaLabLNe6Vqtwr15gqV5mr1Ek1PuYsoRBRSvTU4vQWZukuLtJdi9Fdi+MXbTTM46SFfaQbnWQKdjIpmaa0mSgsgMtvwRuyfbR12PG0WdEZnq7IUBQFudCgES1RiSYopm5SrN2lYpmh6pqnZc6pz0ex4bQcwuAbYz6vcG3hDleKd1kzquIxkFMYixo4btrDqZGvEXj5NYydz77kWJZbpJaXWLs7SWxmitjMPUqZDQAEUcTfGSHYN4A31InRYkFRFKqFAsWNNMWNFMV0ilI284D4fBxEnQ6Lw4nV6cIRCOL0B3EGgrjbO3C3tYMCuWSC9MoS6eUlUqvL5D42KsbhC9AxMER4zwjhPfsIdPc+FYdPaTapTExQ/NFbFN9+i2ZMTTA2j4xgO3UK26lTWMYOfaabuhMoskxjeZn6vXvU7t2jevs2tcnbyCX1/SU6nVjGDmEdO4x1/DDmgwcRP4dL22pKRKfubpWl3w8G83VG6D9ynIEjJ2jvH/yUSK5XKizdvMb8xEU1aKxcQqfX07X/IP3jx+k/cgyH9/nF6dekFleWMry76c7OJtXXp91p5pUhP68MBTg94H/mYROagN0+moDV0NDQeD4oisL/dyPK5FqBaK7CWrbK3XgBvSjws4fC/I2Xe9nT/mLkqHwWO15CLAiCFVXA/qqiKJcecvs3gdLzFLCNWAlDmw1B9/Bv+ZvNJktLS0xPTzM9PU2hUEBUDHQ4+3DpQzTzRrLx6iMFbU1WuL6SZWJJTTq+tpKlstlH2+EyM97t2XJp97Q70O+SsuOPk5OazFfqzFXqLFbrLFTrLFbqLFXrFFsPjrtxtip0VuOE6+t01FOE6mncggML7eikIELJQzVjppBpodzXtaKAK7ApbDvUzdNhe+rCVlEUWtk69eU8pbV5ctkrFLlF1T1Dw67OcRUVKy7LEWTPQe7mJS4uXGGiPkNVbKFvKuxbUTiW8fBqx8v0nfwpbMeOIVqfj2tVSCdZn59jfXGO9cV51hfmqBbyW7cbzBZcwTacm8LT4fNjcTjRm0yIOh2iqEMQRQRRAAREUYdOp0PU61SXS9AhtyTq5TKVQo7ixoYqhjfS5NfjFFKpLaGKIOAOtuOP9BDo7qGtb5D2/sGnGrykNJtULl+m8MYPKP7wh7SyWQSTCdupUzhefw37q6+i9/94kVVoFEiWk+TqOfKNPIV6gbJURpIlGq0GkiyhE3QYdAYMogGjzojT6MRpdOIyufCYPARtQUy6pyOOFVmmsbRE9cZNqtevUbl2nca8Oi9YMJmwHDqE9dhRbMePYxkd/Vxlx9l4VHVZJy5tjXOyebz0jx9j4MgJuvYf/FQQlNxqEd2caTw/cZFcQv1yoL1/kP774406I8/VCY3mqrw/o4rZ83NpirUmogCjnW5eGVQF7aEu91P/vPwyC1hBEP4I+AaQVBRl/+b//e/AzwANYB74JUVRcp91HE3AamhoaDwf/uxGlF//kxtYDDrCHgtht4WDXW7+s+MRgs6dz0p5WjyRgBUEoUtRlNVH3PayoijvP8YD64CrwADwzxRF+R8esd83eY4CtpWvE//fLiOYdBi7nZh6XZh6nRg7HQj6T18YKYpCPB5nenqaqakp1tfVnkW/J0jYM4i56aMQb5JeK6mC9hMlx209LhQRphJFJpYyTGyO74nnN8fgGHQc7HIx3u3hcETdPLbdG2+tKAq5ZovlaoOlap1oXWKt1iBarbNWLpFoNMkon3bfLK0agUYWb7OOs6XDJpkw1Uzoi3rEjIytpmCtydgbMgGvFW+7VRW2m66tu82K/imlmMr1Jo3lIqWlJTLJCxSUq1Q8d5Csan+nUQnisJ8gYevlWjzK+dgFVskC0L2ucGRB4LRphIOHfhrnK6+oJaHP8UK/Usizsbay1ZtaSCUpJNfJp9ZpVKvbO7ggYHd7sPv8uILteDrCeDtCuDtC+Dojn+q5fBooskz16lXy3/sexR+8SSuTQbBacZw5g+Onfwr76dMP/cKg3qqzmF9kNjvLfG6ehfwC0VKUWClGSVLdPFmw0DK009K3Ieu8yDonsuhE0dlRBBOKYABBj4IOARmUFihNBKWBKFcwCRIOnYDPoKPTYqHX5mSv089hb4RuR+e20q2b2SzVa9eoXL5C+cpl6vem1DRkqxXr0SPYT57EduoUxv7+x35/VUtFFq9PMH/lIos3ryHVqhjMFnrHjjBw9AR9Y0c+lU6sKAqZ6CpzVy4yP3GJ+Nw0oAaJqWL2JUKDw89tRA+oCYo313K8O5Pm/dkUN1dzyAo4zHpO9ati9isjQYKO7Z+sX1QBKwjC94C/pSjK0jaO8QpQAv7vjwnYnwTeVhSlKQjC7wI86vx9H03AamhoaDx7alKL1751Do/NyJ//N6fRibsjSPZZ8KQCdgH4Q+D3FUVpbv5fG/CPgWFFUY5+jgW4gf8A/JqiKLcfcvs3+QwBKwjCLwO/DBCJRMaXl5cf96EfilxvUZvaoL5YoL6Yp7m+WW6pFzFFHJj6XBh7XZgiDoSHCKZcLrflzC4tLSHLMlarlf6eQQLWbpSSlfWFIqnV4iMFrc4gEs2pfbTXV3JcXc5yN16gtZka3Oe3MRbxMBZxc6jLvWtd2kdRbckk6hKxeoNktUoimyBR2CBVq5KUWiRlPUnRTs7geOj9DS0Ze0PBWlGwVWXsdRlHVaZNb6DLbqLXY2FPu4POLifOgAVxm3/AcqNFY6VIYeEeG+n3KYgTVDx3kQ1VUEQc4ih1xyHuKgY+WLjMzeocsqDgKSqMzymc2PBwov8s3pfPYDtxAtG2MwmriqLQqFapFgtUi3mqhQL1aoVWo0Gz0aApNRAEAUHUqe6sTofZZsNktWO227E4nNg83ucye1VRFGqTkxS++z0Kb7xBc30dwWzGfuYMzq99FfsrryB+LHVZURSWC8vcSN3gdvo2k+lJZjIzNNWPJ/Sing7nPsz2g0jGXjL4WJftlOQHnUej0MItSrjFBhaxhUFQ0KMgCgqSItJQROqKQFUWKckiFVlPTdGjCJ/4LFCa6Jtp7EKRdn2TYZuJY54gZ4N99Do7nugLjVY+T+XKFcofXqD84Yc0lpbU59bRgf30KWynX8Z26iQ6++PNDG42Gqzcucn8lUvMTVykks8h6vREDhxk8OhL9B85/lD3vJTZYP7qJeauXGTl9i3kVhOb28PA0RMMHDtJ18iB5z6fN1dp8MHcBu/PpnhvJkUsX+Of/+JhvnqgY9vHfoEF7F8F/iHwb4DfUxTl0fHZn32cHuA79wXsJ277y8BfURTlFz/rGJqA1dDQ0Hj2/MHbs3zrzRn+5JdPcKLPt9PLeaY8qYD1AP8IOAn8OnAA+A3g94B/rmzVFD72In4bKD9MpO5ECfHHaZUlGot56ot56ksFpJjqpKITMHapgtbU58IYcSJ+IoyoVqsxNzfH9PQ0s7Oz1Go1dDodPT099EYGcOrbKcSaRGeyn+nQ6gwilUaTW2t5rq1kubac4/pKdivt2GLQcSDs4mCXi4Ndbg52uun0WHbNCJ8nQpZp5GNspBZIZdZIFdKkywXStSqppkJa7yJp9JEwBEmafOSMnxaF9qqMpyLTLov0mYzsdVsZ7XByuNuDeRtjfuRak+r8BpnFD8kUzlO0XqXuVAsSjK02DKbjzFvbuZJc4MPUBFUamBswuihzdEHkpPMg4Ze+gv3VVzD29r7Yv6enyP0gpsIbP6Dwve8hra2BwYD99GmcX/86jrNntsT/fcF6KX6JifUJJtYnSFfVkTg2g439vv0MeUeQ9GGWmw5u1ewsNdWeD5EWXcoyvSzQQYx24rQTx0cKMzUEQBB0m6XRP66NQsRg8NI09VDUD5IQOlhuuViSrKxKZpJNE0UcyMJm1YQiY2ilCIp59lpFTnr9fC00TI/j8wstKRaj9MEHlN8/T/nDD9UeWr0e6/g49ldfxX7mDKa+xwtpkOUW8dkZ5q5cYO7yBXLrcRAEwsN7GTj6EoPHTuIKtn3qfvVKmYVrV5i7fIGFGxM063XMNjt9h48ycOwleg4exmB6viVLiqIwnyrR4bJgM21fSL+oAhZAEAQb8PeAnwb+LbB1blYU5fcf8xg9PFrA/gXw7xRF+eOH3PZUv1zW0NDQ0Hg064UaZ791jlcGA/zhfz6+08t55myrB1YQhF8H/gkQA04oirL2mA8aACRFUXKCIFiAN4HfVRTlOw/Z95s8RwGrKAoXLlygp6eH9vZ2xE+Uxcm1JvWlAvWFPPWFHFL0Y4K206GWHPe5MHY7ED928dRqtVhZWWF2dpaZmRnSafVi2+/3Mzg4SE9XH7qag8R88ZGCNjSo9tDqDToURWE1U+X6apbrKzlurOa4Gy/QaKrXJ16bkf1hF6NhF/vDLvaHnYTdL7iovU+rCfkV2FiAjTnYmKO+sUCisEFMklkxhZk1DjFn7mfV2knU4iT3sbRnXUuhowYDgp79Lisnw25OdLmx6p9M1DZzNYrTs6SiPyLb+pCy+zaKroEom7EoY8QsA9ySa7y3+iGpZhZRgeFVhfFZmRPFIMNjr2F7+WVsx48/t97Z3YIiy6rT+uabFH/wpipadTpsL72E86tfxfGV19G5XADk63kuxC9wIaZu8bLalxm0BBlvG+eAJ0KnQeBeVeGtkpOJ5hB1wYxekdgjTDFuiHLIWmefzYjb2o7Z1IHB4MFgcGMwuNHprIiiEUEwIAjq372itJBlCVmu02yWaLaKNKUCkpSl0UjTaKSoN5LUa3GqtSi12hqyXN96fjqdFbOln6TpIJNSF3caLuYbVuItBw1xM0RBaWJpRokYyhxxWvnZUB8vte3FID5+QrkiSVRv3qT07ruUzr1LfXYWAGN3N/bXXsN+9gzWw4cRHsMZVRSF9MoSs5cvMHflAqnlRfV17uln8PhJBo+dxNfZ9an7SY06yzevM3flAvMTl6iVS+iNJnoOHmbw+En6Dh/FbHs8d3g38YILWCPwd4BfAP4dDwrYv/+Yx+jhIQJWEITfAo4AP6f8mIsFzYHV0NDQeLb85r+/yZ/diPHD33iFbt/umaX+rHhSB9YN/C5wHPjvga8Br6MmCb/9GA86ilrWpANE4NuKovyOIAi/AqAoyh8KgtAOTABO1JNuCRhRFKXwqOM+jZNkOp3mD/7gDwDU0t/+fgYGBujv78f+kNI8udakvlygsZCnvpCnES2qqxXBELKrDm2PC1OPE9H60QVpJpNhZmaG2dlZlpaWaLVaGI1G+vr6GBoaoivUQ2m9RXQm96Cg1Yu09ToJDbkJD7pp63NtjaFpNGWmE0VurOWYXMsxGS0ws17cKj12Ww2MdDjZF3IyEnKyt8NJf8C+a2bTPhXqRUjNQPIuJO9B8g4kJinWKtwyHeSGeZzbrkNM28Os2CyUzOpzF2WFzobAAaOJ021OXu/1EbF9fudIkRWqKynSc++wkX+XgvkKTYs6q9VcHyBvGOWuwcSF9CQzpQUAOrIwPiNzZEnkUPth3KdexXbqJKahoefaV/i8kCsVyhcuUHznHUrvvksrlQaDAduJEzh/6iexv/46eo8HWZG5u3GX89HznI+eZzI9iazIOAwOjnUcZczbx5Cpibk2wwf5Cm+3jnKVY9QFMy6hwllbjp/wWTgT6MZr70b4ZJnvM0BRZBqNNJXKAuXKApXyPOXyLKXyNI1Gems/szlMyXKEa61BrtX9TDccpBSf2nML6KUEYXGDo04TP985wOn2fZ9L0ErRKMVz5yi9c47KpUsokoTO5cJ+5gz21197ZN/ww8gl4sxeucDs5Q+Jz0wBaqLx0IlTDJ04jb+r+1P3aTWbrN27zezlC8xfuUApm0HU6ejaN8rgsZfoHz+O3ftilDe9qAJWEISfBn4f+HPgdxRF+Xzx4x8dp4dPCFhBEP468CvA649zXE3AamhoaDwb6s0W782k+eV/O8F//XIf/9PX9u70kp4L2+mB/T+Bf/qxHthDm/+3rCjKf/qM1vuZPK2TZKlUYn5+nrm5Oebn56lszmwMhUIMDAwwMDBAOBxG95BRIHL9/2fvPWMkO9f8vt9JlXNVV3d1qM6TSA455CTmPDMkfTfJuyvpypbWghZrCIZsC9ZCAoz9YliCDRgQIMtYW/DCX3a1WF1L691LXqZlmOHkwDC5c06V8zmnznn94XRXc8gJTfYMw731A16cmZ6a6lOnq7rq//6f5/lbGLPrDu1UEWOuDJYACbQuP+7BMK4BZziUEnTKCnVdZ2pqirGxMcbGxiiVHI3e1dXF6Ogoo6OjdMQ6WZ4osTBWYPFmgcxcGSFAliU6+oN0j0ZIDYfpGg7jDWwOeWqYFleXSlxZKHJlscSVxRI3lssY65OCXYrMSDLA7lSI3akgu7pC7OwKkgi4fjncWnAyUUsLsPQZLH4KCxdg4Tx2vcxN+SGOh17hXGQ/1/wdzERcGJrzuGMGPKa4eLErzLGhBH2+bz5x1qqZ5G6cY23xXQrWSer+CZAEqhGjaT3GTVeCi+YKFzKXMEUTvyGzd7zJ4xOCfbkQPY8+if/wk/gO7P/RlhsL20a/ft0pd/3kJPULFxCmiRwI4H/2GYIvvkjghRdQQiEy9QynFk/xyeInnFw4SV7PIyHxUPwhDnc9yh6fRtKaplg4zaKp8BEvcVx6hQwxgrLFf5bw8lupXp6MBFF/YMMLdCNDpXKdSvkqpfJlyuXL1OubZZWad5Rx7SlOmoN8oceYt+NYkrOJopkL9Kt5nosG+Gn6EfbEtj4YzKpUqZ44QeWDv6H84UfYxaIzufmppwi+8gqBl15EjW5tUnQ5l2H87Clunv6E+etXQAhiPX3sOPwMOw8/Tbyv/2vnJWyb5Ykxxs6eZOzsydZE49TIToYPHGZk/2FiPb0/2Of2j1jAHgf+QAhxZZv3M8CXBOyXhPHzQoi1rdxHW8C2adOmzfYo1k3evrJMpdGkZjQpNZp8Pl/g0mwBvWnTGXLz7n//PCHP1je7f8x8WwHbe6dyYUmS/pEQ4v+6j+e4ZR7Em6Rt2ywtLTE+Ps7Y2BgLCwsIIfB4PLe4s6HQ7TOVhGlhzJVbQ6GMmRJiPQNVTXhbYtY9EEKJOR9WV1dXW+7s3NwcQgi8Xi/Dw8OMjo4yMjKCKrtZniiyOF5gaazAynQJ21p3Wjt9LTHbNRQm2ulbj0lxMC2bybUq15ZKXFsqcXXJEbWr5c3Sx5jfxWgywM6uIKPJAMPJACPJAB0B9w/2g+Y3wrYhNwHz52D2tLMyNyhZcY77jvJx/CU+DfQyFnVTW3dpOw045PJwrDfKkYE4Ae2b9dcJIagtzbE6/jbZ0oeUXZ9iqw0kS0Op72FW7ueqanMq/ylZw3FtB7MKe2+aPDIt2FMNE3n0CXyPP4H30b149uxB9v6wcoMBhGHQuH6d2vkL1M6fp3bhAnbRifZx79iB/6mnCDz/HL4nnqCKwcXVi5xeOs3ppdOM5Z3S15gnxpOpgzwaTjKiFmiWz1OrjWMjc0V9gQ+UX+es0QNIvBAL8rtdMY4lwnjuUU0gbIEwLIRuYZs22AJsgVivUpAUCRQZSZGQXAqyW7nt9PH7RbNZplT6nFLpM4qlzygWL2GaTravrISZ8b7McXMPl4wkSyLpOLTCJGBO8YjX4NdT3fxWej8h99Yy3YRpUrtwkfL771N+/z0nO1dR8O3fT/DVVwm++gpa59d7XW9HJZ9j7OxJbp4+wfy1dTHb3cuOJ59h5+Fnbi9m1ycab5Qnr0yOA85E46EnDjGy/xDdO3cjb2N68/3mxypg7weSJP0Z8AKQAFaAPwL+OeAGsus3Oy2E+IO73U9bwLZp06bN9vjn/+/n/NnZzQAYTZHY2RXk4ECcg4MxnhyOE/b+aohX+AHkwN5Pvos3yVqtxuTkJOPj44yPj1OpOHEcyWSy5c6m02nUO/SaCcvGWKhgbPTRTpcQDWdSqhxy4R4I4R5wXFqty09DbzAxMcHY2Bjj4+NUq1XAcYM3xGxPTw+2JVidKbM0XmB5osjSZBG96tyv26fSORimayhE12CY5GAIt/fr55et6NxYLnNjpczNlTI3lsvcXKlQ0Zut24Q8KoMdAYYSfgYTfgYSftIxH+mYj6hP+3GL22oWZk/B9AmYPg4rlylbUU743uD9xCucD3UzEXNhqhKKLdhlyLwQCvBbO5M8lLj9xOS7YRl11m5+zNriexSskxjuZQDUaoqCsYsJV5DPxBqfFS5jCQvNltmxprLnZp2d8zC6IhMZ2IH34Ydw79qFZ/du3Dt2ogS+u94H0WxiTE/TuHad+hef0/j8CxpXryIMZ8CY1p/Gt38//oMH8T35JFm/zReZL7i4cpELKxe4kb+BLWxcsovHO/exLzbIDrdFxLhOqXQJIQxk2Y0IvcBx5Q3+stLLoiHodKn83VScn3bH6fU4FQd2zaSZbdDMN7DyDZp5HatkYFcMrIqJXTEQxjeaL+egysgeBdmnoQQ0ZL+GHNBQwm7UiBsl5EaJulHC7ls2ir7V9RSCen2WYvECheIFCgVHuANYcoTL7qMcN/dwpZmiJDkluEpzlR4WeDHm5+/17+Xh+M4tvQ6FEDSuXqX83nuU330XY9zJnfU++ijBI0cIHj2Kq7dnS+ddLeQZO7MpZoWwifX0sfPJZ+5YZgxQzmaYOH+GiQtnWhONvcEQQ48fZPjAIQYe2Yfm+X5z636VBez9oi1g27Rp0+bbs1pu8My/+oDf2NfNv3h9Nz6XiusBbq7/GGgL2Nuxeh0SO+Ae/YdCCFZWVlpidnZ2Ftu20TSNwcHBlkMbj9+510vYguZqrTXl2JgqYpWcD/+SW8GVDuLuD+HqD6H2+lnJrd3WDR4aGmq5weFw2MljXamxPFliebLI8mSR3FLVGTglQSzlp3PQEbSdgyGiKf9t42aEECyXGoyvVhhfrTCxVmEqU2U6U2OhcGumaNCt0hP10hv10hv10RPx0hP10h3x0h3xkPC7tx1p851SyzlCdvJDZ+UmWbN6eSv2O/xN/CkuhqOshpyNgI664LDs4vWeGK/tSuL5Fpm05ewYq2Nvky1+SEW5jJBNJEtDLu1gkT4mNI0vrAWul24gEMhCYrDiZWTGYGhaZ3hJ0JsFd3cPrqEh3EODuAYH0Xp60FIptFTqW0X4CMvCyuUwl5YwZmYxZmYwZmbQx8cxxscRppPOIXk8eB56CO8jj+DZ+wiVh/uZkLPcyN3gSvYKX6x9wWp9FQC34mZvx14eifQz4oEuMUetdJ5mswxAIDQpvW4AACAASURBVLCbaPQZZj3P8bNiF3+dKWMKwbPRAP9lLMILugKrNZorNczVGs1MHbtya0qI5FVRwy7kgMsRnH4NyaM6rqpbQXIpSDIgSy3BJ6x1N7ZpIwwLW3eWqDexqyZW1cSumFhlA6Fbt14oRUKNelDjHtSEFzXpQ0v6UJM+FP+33xXVjQyFwlny+dPk86eo1Zze6ZwywgnlCGfMEWZFN7akIdl1guYNDgQs/nbPMC/3HsSnba3XVZ+cpPzOO5TeeQf96jUAPA8/TPDoEULHjuHq+/rgptuxIWZvnD7ecmadntln2PnkM8R707f//rUa059dYPzcaaY+PY9eraJoGumHH2X4iYMMPXGQYCyxpXO4n7QF7PZpC9g2bdq0+fb8L7+4zv/x0QQf/NMXGEj88g9o2gptAftVSkvwv+0CXwKGXoDhF2HoRQjf24nY6GXd6J/N5/MARKPRlpgdHBzE7b5zL6UQAqugOw7tdBFjpoy5sik8tS4/rnVBa3dozOTnGV//fhtucCKRYHh4mKGhIfr7+/GsOxh6vcnqVInlqSIr68cNl1ZzKyT7gyQHQnQOhOjoDxKMee7q5NQNi9lcbXNlq8zn6ywU6szn67c4t+D023aG3aTCXlJhT+vYFfbQFfKQCnuIB9w/3ODl/DRM/A2Mvw+TH2HrVS5pT/JXXX+H4+ERrkc9WIqE2xTsrUu8Eg7ym7s6SXd988mrllUnO3uCtfn3yddPo6tO2YhseqG0gyVSzGgaN6Uc1yo3qTYdZ94jVAZrAdJrgp6pMukFg1QOIlWQADkYRAmH11cIyetDUhQkVQFZQeg6dr2O3ahjV6o0sxmsbM4pud5AklBTXbiHhpF3DlMc6WS128e8r8F0ZZbJ4iQThQnyer71X9LBNA/HdzESiNKnGSSsWSqlS1iW85z1etJEo4eJxp7CFTrEz/MyfzKf4Wq1QQCJ3zBUfnvZom+udotQlX2qIxQ7fKgdXkc4xjwoETey58Fmkdp6E6toYBV1x/XNNmjmGjQzdZqZeqtVAEAOutBSfuf1m/Kj9QRQE95v5dg2Govk8ifJ5T4hlzuBaebQcfGFdpSP7ANcsQbQJT+IJm79OrtdOf5WVzc/GXiGLn/Xlr6HMTdH+e23Kf3ibRqXnXhuz8MPE3rtNULHjqL1bM2ZbZUZnzrR6pm91wAocIZALVy/wsSFs0xcOENxxalOSA4OM/zEIYb3HyI5sPU+4O3QFrDbpy1g27Rp0+bbUdGbPPkv3+fZ0QT/9qe//PE4W6UtYL+KXoZrf+0IlckPoeo4RiR2borZgWfAfW9Rks1mW2J2amoK0zSRZZl0Ot0qN+7s7LznhzC70cSYLTvTjmdLGLPllvsj+zVc6SCudJBa0GKqusj47CQzMzM0m00kSaKnp4fBwUEGBwfp6+tD0xw3SAhBcbXOylSRlekyK1NFMvOVVi+tJ6CR7A/S0RekIx0k0RcklLi7qN1ACEGp3mSxWGex4Kz5Qp3lYoOlQoPFYp2VUgPTuvU5psgSyaCbzpCHztDG0fOlrzlfD3u/53Jly4S5MzD2rrNWr5Anyl92/H3eTjzHuXCUiltGsgVDJcHTipufpOMc3t2B9i0yaBuNFbJzx8kunaDUuICuLgIg2SpaKU3R7GVZDTGrCmbIM1GbomgUW//fg0a3HaKr4SHakIlWIFKwCNQsvAZ4dIHbECiahvB4wO1C+Dw0Yn4aES/1kJtKQCXvF+RUnTU9y2JlkeXaMvaXYp9DrhBD4SEGQj30e4P0aBYJKUuzep1abYKNXFW/f5RIeD/hyH7C7sdQSjEuLxf501KZv5QNKjKMli1+e9bk2JKJX5ZRO31onY4I1Lp8aF1+5MB9eB4IAUYFzAbYTRCWc5Q1UN2guEDzgvINYm1s4QjbVcchNpeqmMtVzJWaM9QNkDQZrTuAqyeAqy+I1htAjX8zUSuETaVyjWz2ONncxxSLF7CExYT0CB/LRzlv76JExMmf1W+Slmb49WScn/Q/za7Yri1dO2N+3hGzb75F44ozD8j76KOE3nid4NFjaJ3JLZ3rbXtme/ocMXvoaRLpgduez0bf7Pj5M0xeOMvi2HUQgkAsztDjBxh+4hB9D+9Fc33zIWtboS1gt09bwLZp06bNt+PfHZ/kf/r5Nf7yHz/No32R7/t0fjC0BezdEAJWrqyL2Q9g5iQ0GyCr0HtwU9B27wPl7m5Ps9lkdna2JWhXVlYACAaDLXd2aGgI3xaiLYQtMJerGHNljNkyxmyJ5tp6Oa8EaocPrcdPxW8yb6xxPTPJ/JJTbqwoCr29vfT399Pf309fXx8u1+bUYsu0ySxUWJspsTpTZnWmTH6pir0+5MblVUn0Boj3Bkisr2jK34ry+SbYtiBXM1guNlgsOIJ2udRguaizUmq0VqnR/Nr/damyI3CDnpaw3XBzU2EPPVEvyaDnu3Nzi/Mw9g7cfBsmP8QydT4Jv8J/7PrbfBzqZ2G9hDRRsthflzkSD/HK7iQdPYFvJcAMI0N++Ry5xdOUKl9Q4wa23ADWRW2li4bRTUaKkNM8rCk2K1KNJXONjJ6hqBfv8R1uj0fxkPQlSXgTpPxJku4AHS6NuGIRk8qoxhz1+hSGsTmc1OVKEnDvIiDtwWfuwFMegpxGM9ugVKjzTgj+Y6+LyxEFlyV4uSj4qe3mQCyAq9OP2uVHjXm+XX+p2YDcel5wbgJKi84qL0FlDfSis2klttAbq3rBGwFvFLwxCHZtrlAPRAec5Y3CHX6mwrJprtUxFiqYCxXnuFhpubWSR3U2pPqCuNePX47fuhfNZpl8/hTZ7Mdksh/S0JeYE2lOKK9zWuwjS6IlZpPN67yRDPNr/c+yL7kPZQuDk4zZWUpv/YLSW2+hX78OkoTviSccMXvkCOpd2iW+zIaYHTv9SatnNprqZvTgU4weeprOoZE7vi5qxQKTF88xceEsM59fwtQbqC436UfWS433HbivET1tAbt92gK2TZs2bb45RtPm+f/1A/rjPv797z/5fZ/OD4q2gP0mmA2YOw0THziCdulzQIA7DIPPOiXHQy9AfOSOH2A3KJVKrd7ZyclJGo1Gyy0dHh5meHj4jlE9t8OumRjzFcehnStjzJex18uDUSSUpBc9KFijyGRlgbHcDE0sZFkmlUqRTqdJp9P09fV9Le+2aVrkFquszZZZm6uQnS+TWajS3OgBlCCc8BLr9jsr5Sea8hPt9KF+C2H7VeqGxWq5wWrZEbbLxQZr5Q2R6xyXig3q5q09iZoi0R3x0hf10R/3MZjw0x/3M9Thpz/mQ31Q+bdGDaY+hptvOYK2vMSkp5+f9fw+74T3ccXvw5Yl/A2b3RmL59xeXhtMMLo7jifw7XolhbCpVicpLl+klLlKtXqTmjWBqWZuuZ1iBFGNKNhhagTRZS+momEqGoaiOAOLZAlJlpAVCY8s8MgWHsnEjY5qFTCbWcxmlqZ9qwhWhB+PlcZt9OCqdeMu9KKtdqNUbn0+WRJcSHv4eY+L9wOChgSjqspPO6L89kCS+LcdAV/LwcJFWP7MiUxa/gJyU4DAtN3U7Ci61onu6cVwpTDUBLbix1J82LIHFBeSLCHJMrIsoypNNMVEkxq45AYekcMjMriba0j1LJSXndW8tRccdwjiw07VRmIUOnZC50MQGbhtX72whOPSzpdbm1KttgFATfpaffCugRBqfOtVENXqTTLZD8lmPqBYusi86OIT6WVO8hRrdKyL2avEjC94LRHi1wZe5EDqwJYyZ/XJKUpvvUnpzbcwJiZAUfAfOuSI2VdeQQmHt/BDcwTp+LnT3Dh9grkrnyNsm2C8g9GDTzJ68Cm6d915KnHTNJm/+gUTF84yefEspTWnYqZzaITnfvpfkX5475bO4W60Bez2aQvYNm3atPnm/OzCPP/0Lz7jT37vAC/u3Fq1068KbQG7HapZmPpofcjPB1CYdb4e6oXhFxx3dugF8N998IhlWSwuLrYE7eLiIkII3G53axjU8PAwsVhsy6fW6qWdrzgfjNedHru2LmolsMMqFY/OcjPPVHmBDCXqGMTiMfr6+lqro6MD+SsfvIUtKGbqZOcr5JaqZBcqZBeqFNfqrUgSJAjFPUS7/ES6fEQ7fY6w7fLdklV7PxBCUGo0HTe3WGch7/ThzudrzOVqzORqFGqbvZMuRWaow89oZ5BdXUH2dId4KBWiI3ifY4Js2xFUN96CG2/C8hcU1AA/7/k7/HX8GKd9ceqKhNoUDK6Y7G/IHEmG2bsrQddgGEXbnshuNsvUKtNUVsep5ieo1+YxjDVMO4cp5bDkKrbSAOnOr3XJUpEtL7LpQzXCKEYIVQ+hNmJo9Q60RgKt3oFiBlG861N6fRpKyIUSdCEHXUhBjc988Kal8/NSmRWjSVhV+PVkhN/pivFEyPfNrrsQjrM6fQLmzmLNnKewWqNgdVNsdlFSRylK/VStGNWGG8O4f5sVkuSU1/sjbvxhN/6gRNBTJ+jOE5IWCVmT+MqXkbI3nfzhDVwBSO6Brkeg+zFIPQbJ3bctTbb1JsZcBWO9bUCfKW9OKw9ot+RJa13+LbnTpllwnNnM+2RzHzHVjPAJz/OJ9Bw54iCauOqfETU+5fWOOG8MvsKh1KF7ilkhBPrNMUpvvknprbcwZ2dB0wg8/TSh118j8NLLW56MXa+UmbxwlrGzJ5n+7CKWaeINhRnZf4jRg0/R9/CjqNrtz0cIQXZuhomL55i8cJYX//4/omtkx5a+791oC9jt0xawbdq0abM1mpbNpbkC719b5T9cmCfud/GL//bZH3fKxwOgLWDvF0JAfmrdnf3QEbaNdYeqa+9muXH6SdDuHgtRq9Vaw6AmJiYormdoRiKR1nCmwcHBLZUb33qKjqg1FyoYixWnL2+xilXczH+1XFB26ayYOdasIgW5StVlEu/poLevl56eHrq7uwmFQrd9MVmmTWG1Rn65Rm6xQmGlRm65RmGlhvWloTYev0ak00e0y3fLMdThRXlAzmihZjCdrTGxWuHmapmxlQo3V8rM5zcdtETAxSM9Yfb2Rni0zzkmAvext64wty5mfw7TJzAEnOh8if8v9bu87xtkTXWcplSuyehqkyc1D8+kI6R3xOjoCyA/gGtjN22a1QpmuYjdMMEEYdoIE2ThRUZ1nt+ShKRKTi6qKiO7ZCS3iuxRnOm+HtXJUV1Ht21OFSq8kynxVqbIkm7iliVejoX4zc4or8ZD98xtbbH++rInPqZ07TMyk8tky2GyzTR5e4CimUSweV8ev0Yo4SEY8+CLuPGHXfhCbjwBDbdXweXVcHkUZEVGUR23GUlycmJtgW0LmoaNqVuYuoXRaKJXTeoVk0bFpFY2qBb01qqXb52ArLoVIkkvkYSLSKBKTFsgLq4RrpxDWf0M9JJzQ8UNqb3QewB690PPfoikv1bBIWxBc63mTCqfdjKlrYLzupV9qpMlPRTGPRpF7fDeu6/eNikUzpHJvM/q2ntc110cF89yUn6eMmEku4Grfp6Y8RlvJLt5bfAIB1MHtyRmG5evOGL2F7+gubSE5HYTeO45Qm+8TuD557ecXWw06kxdusD4uVNMXjyLUa/j8noZeGw/owcOM7hvP27fg5/G2Baw26ctYNu0adPm3lyazfMP/5/z5KoGqixxYCDGPzu2k33p6Pd9aj842gL2QWFbsPgpTP4NTHzoDPyxTVA90P+UI2aHX3JKC+/yYVMI0RoGNTk5yfT0NLrufHDt6upicHCQoaEh0un0Xacb3/VUaybGxpCZpWormuTLMSGGbJGnQlGqUpRq6B4bb1eQSH8HnX0purq6CIfDd/zgLGxBOdcgv1KjsFwjt1ylsFwjv1Kjvh4bBCDLEqEOL5Gkl3Cnj0jS5/w56SMQ2X7O5u0o1k2uL5W4ulTi8kKJLxYKjK9W2DCS0zEf+9IR9vVFeKI/xu5U8P6UH9cLMP4eXP85jL+H0EtcC+3mzcGf8rb/CS6rXoQk4dFt+teaDOctDni87E+H6R6KkOwPfauBUA8KIQTTdYPj+TIf5ct8mCtTtWy8ssTzsSC/loxyJB4ioG7tnI3VOXKXzpC5MU1moUam1kG22U9TOBtAkiSIJDRivVGiKad0fWMT5HY5x6WGyXxuY0p2jeVSg3zVoFAzKdRMKnoTvWmhN230po0QAkmSkCWQJQmPpuDRFLyajM+lEvZpRLwaYa9G1K0SkxQCTXDrAsom5bUGhdUa5UydjV+lsiIR7fKRSELCt0pCukFH7Tju1dNOfz1AoAvSh5zNrvRh6Hzktj32zYKOPllAnyiiTxQ2BW3IhWckgnskgmc0ihK8e7WDEIJK9QaZtXdZXnuP8xXBCZ7lDE/TkHzIVgl37TRx4zPeSI1ydPAIB7sOosp37/sXtk39008pvfkWpbd/gbWWQfL5CL74IqE33sD/zNPIrq1VYjRNk9nLnzJ+7jQT589QKxaQFZX0w3sZOfAkw/sPEYhuvULlm9AWsNvnB/Xe3KZNmzY/UH7nj08xnanyRz95iGd3JAh925aqXwHaAva7Qq/AzCeb/bNr152v+5Mw9PxmufE94no2yo0nJyeZmppibm4Oy3J6Wbu7uxkcHGRgYOBrw5m+KUI4U1TNlRrNtTrNtRrGag1jpYpUu7XXtIFBSapTU00IqbhiXvypCNF0B4nBLlz+uwtrvWY6wnZd3BZWHGFbXKvf4toqmky4w0u4w0sk6SOc3Dj68Edc97W8oqo3ubJY4rO5Ahdn81yczbNScgSCV1N4rC/CgYEo+wdiPN4fJeDeZmRL04DZk3DjF07vbH6aghrgo76f8G7nMT7RellaFwxe3aYn26Qnb7FH0tgfCzCYDtGRDhLr9j8wB/urCCEYr+mcK1U5W6hyolBmvuE4kSm3xqvxEK/GQzwTDeK9yzltuPa58RlyNyfJzRfJ5FyUzM3Se7eqE0/YJAY6SIz2Eu8NEEv5b9tj3bRsbqyU+WK+yPXlMuOrjtO+WtZvuZ1LkYn4NGJ+FxGfRsCt4dZk3KqzJElCCOdxNm2B3rRpmBYN06KiNynWTYo1k0LdxLK//ruyI+imN+olHfYy4HLRacv4dYGdNygsVqkVNzduwh0ekp02Sd8iSfsSyeLbqOWJ9RMNOkJ24GkYeM4pP75NT2gzW6cxUUAfd9ZGu4DW5cc96ohZ92AI6R4ZxfX6AmuZd1hcfZ+PihbHeZZLHMCSNNTmCq7qJyTNy7ze+xhHB46yv3P/PQdACcuidu48pTffpPzOO1iFAnIoRPDVVwi/8Qa+gweR1K29hmzbYvHmdSbOn2H87CkKK0sApEZ3MnLgSUYOHCbW3bul+9oKbQG7fX7Q781t2rRp8wPgzGSW3/0/T/NHP9nD7z09+H2fzg+etoD9viguOEJ28sP1uJ71qa3xEUfIDj7vDIby3r1swDRNZmdnmZqaYnp6moUFZ9qwLMv09PQwMDBwXwTtl7ENi2bWybvUVytUFvPoa1UoNXE1ZGRuFZKG1MT0CAioqBEPvkQAf2cYLepFCa73Sfq1r7mrwhZUCjqF1RrFlRqFtTrF1TrF1RrFTB27ufn83CjZjHaulySnnIFSkaQPRb0/gm6xUOfCTJ4LM3nOz+S4uljCFiBLsKc7xIGBGIcGY+wfiG2v7FgIyE7A+LuOQzt9ApoN5jwpTvb/Gp/En+a80s2UpCHWL1moatFZsEiVbAYVlZGgh4cSQXp7AkQ6fQRiHuRtuNemLZis61yt1LlaqXOlUudSqUa+6WxmRFSFJyMBno0GeDYaZMR3ay+x1bQp5xqUMw2KmTqF5TKFuVUKyzVKFRUhnJ+RhEVYXSUeqZHoDRLfNUz8oYcIJu5cFpuvGpybznF2KsfF2TxXFkvoTWfjw+dSGE0GGEkGGe0MkI756Il46Y54SQTuz6aHEIJi3WS1rLO6PlRssVBnLl9jPl9nNldjobDpwgKkwh52R/3scLnpsmW8ZYv6ap1qft1FlSXiKRed0SJd6lVStXcJFk87xRrusBPltbHxlRi9bcmxuVSlMZZHH8ujT5ec+B5Vxj0YwjMaxbMjitp5995j3ciQWXuXqdUPeKcgcVw8zTUeBknGZUzgqhynyxrjtfRTHBs8xr7kPmTp7q83YZpUT52i9POfU37vfexqFSUeJ3T0KKE3Xse7bx/SbYZd3enaZ+dnGT97ivHzp1mZHAcg1t3LS7/3B/TvfWxL93M32gJ2+/yo3pvbtGnT5nvg7/27M1xfLnPiD1/Ec4+N5jZtAfvDQAhYveq4s1MfwfQnYFYBCVKPwuBzjqBNH75n/qyu68zOzjI9Pc309HRrIJQkSXR3d5NOp+nv7yedTn/jHtotPRRb0Cw2yM2sUpjLUlstYebrUGni0mX8woOXrwtpIYHkU1BCbrSQGznoQgm4kAMaSlBDDjhCVwloSF4VIaCSa1BcrTvDe1Y2ndtyrtGa4CrLEuFOXyvyJ94boKMviC+0fTFf0Ztcms1zbirH2ekcl2YLLeE03OHn4GCcw0MxDg3G6Qrfve/5rpgNmD8LU8dh+jjMnwfbpKz4+LTnZT7vepZL7h1cJsysJGN/SY8Eazahuk2oYRNHpkPTCHtUx3kMuAj4NVS3iupVkF0KVdsmbzYpNC0yRpO5hsFMQ2exYbLhhasSjPo87A16edzj5RFVI9WU0Ssm9bJBreSsakGnktep5htUi8YtAk7BIKIuEFGWiPiKxFJ+YqP9RB5+HLXn4ds6jBsUayanJrOcnMhwaiLL2GoFcKKV9n6lf7k/5tuWcL9f6E2LuVyNybUq42sVxlecPuzx1QqN9SoDVZbYE/Wz1+elTyj4yha15Xpr2rcvpJLqqJFyX6e79i7x+klkyYZwn9OOMPyS02vv+fr0X9uwMKaKNMYKNMbyNFdqAMhBF57RCJ4dUdwjEZS7DFczzTxra+9zZeVj3iy4+Vg8w4LUjySauBuf46p8TI+0yLH+lzk2eIy9ib1byLhuUPn4Y0pvvkXlgw8Quo7a1dUSs55HHvlGmwylzBoT508zfu40z/8X/5DkwNCW/++daAvY7fOjfW9u06ZNm++Ai7N5fuvfnuRfvL6L339u+Ps+nR8FbQH7Q6RpwMIFJ4pl6mNHvFiGkz/b/bjjzA48A32HwHX3ISa6rjM3N8fMzAwzMzMsLCxgWc4H4o6OjlZ8Tm9vL7FY7IFOOTNNk0wmw9ryKrn5NaqrJfRsFats4LU1vLjxChcByYsfNy5LQeI256NIKAGtJXKVoAs55HKm3obd4FUp6xb5XIPcco3cQoXMQoVKbrOENBB105EOkuwP0TkYonMghOs2fZPfBKNp88VCkbNTOc5MZbkwnaesO2Wc/XEfhwfjPDnsrM7QdgRtHRYvwexpp7d6/jzUnLicuhZgOvU0E8kD3PDv5DqdLNoeVgRkEdS3uKmn2AK/CbGGIK4LYg1BomqTLFjE8k0w7M0YpdugqRYBVxm/tErAXiAgrxFSlgl7SgRTcQKDO5D69jvDi0I9d+0Db5gWF2bynBjP8Ml4hi8WigjhlHIfGHRc74ODMR7pCf/odi0tWzCTrXJ1qcS1pRJXF0t8sVAiU3GeqwrwRCTAXo+HTkNCzho01kuPNbdEqqNGj+sy3bW36BCfo8g4G10jr8DokTv22FtFncZYnsbN/Ga5sQRaTwDPjiienTFcfcE79pybZpFM5j1OLZ7lr0tBToinKUlRNFFFq57GVfmYtNbg2OBRjg0cY3ds9z1/t1iVKpUPPqD05ptUTpwA00Tr7iZ49CihY0fx7L23IH4QtAXs9vmleW9u06ZNmwfA7/3JWT6dK3DiD1/Cv92WtF8R2gL2x4BRc/JnNxy4hYsgLEfQ9jwB/U87/XF9h8AdvOtdmabJwsICc3NzzM7OMjs72xoK5fP56Ovro7d3c9qwx7MNobVFLMsil8uRzWbJZDKtY2Ytg11v4hUuvMKFX3IT90aIuIMEFR9e4cbVlJHrAlFrtlzXDSS3ghJxo8Y8qHEvIuSiagtyFZPl5Rprc86UZOfGEEv56RoM0TUcITUSJryFaa53fVy24OpiiTNTWU5P5jg7laW0HoUy1OHnqeE4Tw8neHI4TsS3DUdYCCfCafGi89xYuQzLl6G6unkbzQ/xIfToCJVQHwV3L6tSN1URwmy6MUwXpqmg6hIew0bRnd5UbKu1JExUdFTqKHYdlyjgtjK4jWXc+hze5hI+uYBXLqIpJsSGoGOXExWT3O3ExkQHb5uF+tXrdnmhyCcTGU6OZzk3nUNv2qiyxL50hKdHEjw9kuDR3giu+1Qe/kNCCMFKSeeLhSJfzBf4bL7IZ/OFVgxUp6JwOBhgCBVfycJY35jRXJCK5eiRLtCjv0uHNoEcTsHoqzB61Ck5vs2Gl7AF5kKFxk1H0BqzJRAgeVVHzO6I4tkZvaM7a5olltfe4+2FK/x1Jc4F8QSm5MZvrSCVP8JVPcGAL8DRAUfM7ojuuLeYLZUov/c+pbd/QfXkKTBN1FSK0JEjBI8exfvYo1suM94ubQG7fX5p35vbtGnTZpt8MV/kJ//mBP/D0Z384xdHvu/T+dHQFrA/RvSy475Nn3AGQ20IWkl2Inv6n9qcYBq4e/Cxbdusra0xNzfXWrlcrvXviUSC7u5uuru7SaWcacPfdtrxt6Fard4ibDf+nM/nW04ygM/joyfaRWcgTsITISIH8Fku5KpNM9egmW1Ac3MglORR0Tp9yHEPdVUhqzdZWG2wNFNCXx9+4w256B4OkxqJ0D0aId4b2FY56oagPTXplL6encpRNSwkCfakQjw55LizBwdjBO/H5LnKqlOanh13emqz45CbgvISGJXt378n4pSvhnscFzU25Kz4MEQHQN3a88SyBdeWSpyezHJmKseZyU2hv7MzyFMjcZ4dTXBoMP4ruzMphGAmW+PTLw0Vu7ZUxrIFfgGHAgF2Ky7CZQsz7zi06KZD0QAAIABJREFULs2mOzRPr3WcHuU8cfcy0tAzsOMY7DjqxPXcBrtm0hgv0LiRp3Ejh10xHXe2N4h3ZxTPrhhad+C27qxplpha+Rt+tjjFW9UUN9kFwibWHMMqfYBWO89gMMWxwWMcGzjGcOTepVJWsUj5gw8o/+Jtqp98gjBN1GSS4JEjhI4ewfv440jKg3Pe2wJ2+/zKvDe3adOmzTdgNlvjn/z5JSbXqpz4wxfvz2e/XxHaAvaXAb3ilBnPnILZUzB/bjOSIzYEfYedWI6+Q5DYeU8HrFarsbi4yMLCAgsLCywuLlKpbAqeeDxOV1dXa3V2dhIMBr/T8j7LsigUCptubSbD2toamUyGen0z19Xn85FMJkl1pegOJelwRfDrLpprdcwVJzJoY1orEqgJLyLhpSzLrFRMpufKlNYdLpdHoWs4Qs8OR9B29Ae3NfHXtGw+myvwyXiWU5MZLs4WMJo2sgQPdYfZPxDlwECM/f1RktspOb4djZIjZGs5J5O0UXKOtgXCdjZEhADN67h2mhdcAfAnwJcAX/yeecZ3om5YfDpX4MJMjvMzeS7O5FuCdaPU+qmROE8NJ+gIPrjNEmEJ7JqJVTawKyZ2vYkwrPUMXAuxsd/hRMQiaQqSS0FyycgeFdmvIftVFL92z8m+D4Ka0eTT2QJnp3Ocn3ZEbc2w8Nmw3+fjYdVNuGzRLDrOrcdl0Ou5Qi+n6HN9Rqi7A3Yegx2vQc/jt+07bg2Dup6jcSOHMVcG4fTOenfF8OyK4R6NIN9uGnSzzIXFj/jzxWXeqfeToQNNNIgbn9MovI+qX2NHdITXB1/n6MBR+oJ993zMVqVC5YMPHWf2+AmErqN0JAi9+irBI0fw7d+/5WnGW6UtYLfPr+x7c5s2bdrchore5N/8zTj/94kpVEXif/7NR/iNfXdPIWlzK20B+xWEECwt/QWh8D78vpHvpedq2zQNWPrMEbNzZ5xjLev8myfs9B32HoTe/U4Jsjdyz7sslUosLS2xtLTE8vIyy8vLFAqF1r97vV46OztJJpOt1dHRgdfrfVCP8o5UKhVWV1dba3l5mZWVlZZjq2ka3d3d9PT00NPdTU+4C3dJwlisYC5UMObK2FXnQ7+kychdfho+lTXdYmqpRmbFEciqSyY1HKZnZ5SeHdFtC9qGaXFxJs+pSads9tO5QmvAT0/Ey97eMI/2RdjbE2ZPd2h7ZcffEXrTYmylwhcLRT6fL/DpXJGbK+VW7MxIMsD+/iiH1odddUfu//PFbjQxFysYi1WamXprWUX9a2Xn3xbJo6CE3agRt3OMO2XrStyLGvfcVuDdb0zL6cE+M+n0YJ9bd/iDtsQBn5c9iptAoYlVdTYLwu48fcoZ0q6L9EQWce160RG0Qy/ecVicVTEcZ/Z6jsbNvJMVrUq4hyJ4dzuCVo1+fWPDMMu8M3eKv1jO87Gepo6XgJ0hol+kmn8PtbnE3sRejg0e40j/ETr9nfd8vFalSvXjjyi9/Q6Vjz5CNBoo0SiBl18idOQI/sOHke7D5PW2gN0+bQHbpk2bX1VWSg3+/NwclxeKVI0mFd1iNlslXzP5W4/38s+O7dzeXJRfUdoC9ivUalOcOv0KAJoWIxI5SDRykEj0EAH/DqR7RET8IBECcpOOmJ07A7Nn1nNo13++iR3OcKieJxwnpvPhLblr9XqdlZUVVldXWVlZYXl5mbW1NQxjM98yGAzS0dFBR0cHiUSidfT7/d+5Y5vJZFhaWmq5y8vLyy1RGw6HWwOt0uk0MTWEOV/GmCmjz5QwlypgAxIoXX70oIs1UzC1XGVtyemjVd0K3SOOoO3dGSXRF9xWybHRtLmyWOTCTJ5P5wp8Pl9kNldr/Xsq7GF3KsTOriAjHQFGkgGGk4HtZ9J+C2xbsFCoc3OlzNhqhZvLZa4ulRhfrdBcF6thr+aI8N4Ij/dHeDwdve8iXNiC5moNfaqIPl3CnC875ePrSB4VNeFBTXhRYx5nAFjA5Uy69qrrDquCpMlOiawAEI4pbVoIw0YYFnajiV01satNrKqJVdKxigZWUccqNLDXReIGStSNlvShdvrROn1o3QG0pBfpAWb2Ni2by4slTk1knU2RqRx1wyIuJJ70+RkVKq6sgd0UyJJNl+smae0cae8VEjv7kXa95rizodTtr3XTRp8uOWL2WrZ1nbUuH55dcTy7bz8IqqyX+dnMOX62WuW82YNAJmFN4a1foFb4AMUu83jn47w28Bqv9L9C3Bu/52O1azUqx09QfucdKh9+iF2tkvpX/5LIb/zGtq9jW8Bun7aAbdOmza8SFb3JhZk8//7sLO9cXcGyBTs6AwQ9Gn63Ssyn8Q+eHuSxvnsbSG1uT1vAfgUhBPX6LIXCWfKFMxQKZ2k0FgBQ1TCRyIGWoA0GdiNJP66ppy0aJWfoz/w5Z4rtwsXNoT+y6gzfST3q9NR2PQKde+6ZSQvrmZjFYsv9XFtbay3TNFu383q9JBIJ4vH4LSsajd63vNp70Ww2WV5ebvX+zs7OtkqlfT4f/f39DA4OMjQ0RDQQwZwvo0+V0CcLGLNlJ1dTllC7/TRCblZ0i8n5Crl1h9btU+kejdC7K0bvrijRrrtnbm6FXNXg8kKRa+tTa68tlZnMVDCtzddqR9BNf8xHOuYjHfeRCntIhjx0hTx0BN1EvBrqNxROpmWTqxosFxsslxqslBrM5WpMZ2vMZmvM5KottxggGXSzpzvEQ90h9qTCPNQdoj++/cd/O5rZ+vpE3QL6VBFRd8SjEnLh6gui9QZwdQfQugPIAe072TixG00nKzlbd8rVV2s0V2qYazXneQOgSmidflw9AVx9QVx9QdSk746Tf7eL0bT5dK7AyQlnmvOl2QLCEqSFwmGfj94GUFjv/1bKpLXzpN2X6Ou38D70Eux6wxnGdYfrZ67VaFzL0bieQ58ugg2yX8WzM4ZndxzPjgjyVzZXluol/nT6Ev8pYzLWTCAJi5R1FbV2gWrxE1QMDnQe4MjAEV7pf4WYJ3bPx2nrOtWTJ/E98QRKKLTt69YWsNunLWDbtGnzy4betDg5kWWp0KBYNynWTebzNa4ulpjKVhECoj6N39nfx989lKY/fvfUkDbfjLaA3QL1+gKFwhnyhbMUCmeo12cBUNXgukN7iGj0MIHA7h+nQwuOS1tacOJ7lj7bXNW1zduEep1YjuRuSO5xjokdW3JrN4Ttl/tVN/pXv9xfC45rG4vFiMViRKPRW44PsiRZCEE+n2dmZqaVo1ssFgEIhUIMDQ0xNDTE8PAwPpcHfaaMPlGgMV7AnHd6AyWXgpoOUvVrLFRNpiZLlNd7aP1hF727Y/TtdgStP3x/+jtNy2Y2V2N8tcL4aoWZbJWZbI3ZXI3lUoPbvYyDbpWQV8PvVnCpMpriLCEEpiVo2jZG06bcaFKsm9SMr0fmuFWZ/riP/rif/piPkWSA0c4AIx1Bwr4HN4hAmDb6ZIH69RyNG3msnOP8KRE37pEI7sEw7sEwStT9g2sBEJagma2vlzRXMBerGPNlRMO5vpJLwdUXwNUfwt0fwpUOIW8z3ulOVPUmZ6dyrXii68tl/DbskV3s0zxEiya2ASBIamP0uy6S7lgm+dgjyHt+4rQh3KGf3q43adzMUb/m/IxEvQmKhHsojHd33Ck1jt36e+NaqcCfzl7mr7KCZTuIKnR6mp9C9SLV0jkUyWJ/535e6X+Fl9Mvk/TdfUDd/aItYLdPW8C2adPmlwEhBJ/NF/nZhXn+6vPFVjoAgEuRSYbcPNQd4qHuMA/3hHhqOPGji/f7sfC9CFhJkjzAx4AbUIH/IIT4o6/cRgL+NfA6UAP+gRDi4t3u97t6k2w0lloObT5/mnp9BviSQxs9TDRymEBg549X0IIjasvLsHLFiWZZuez8OTMG9saLVoJovyNkEzsgMQrxEWcFOu+a77lBo9Egm82Sy+VuWfl8/mvi1uPxtMTsV1cgELjvgiWXyzE5Ocnk5CRTU1OtAVGpVIqRkRFGR0fp6elB0m30yaLjBI4XsLKbokruC1LUFGZzDWbGijTW+2vjvQHSe2L07YnRPRxB0e7/c0VvWqyVdVZKDZaLOmvlBsW6I0oLNYOaYWFYjlg1LGeAlKbIqLKES5UJeTRCXo2wVyPqd9G17uR2ht0k/O5tlUh/E6yyQeNajvq1LPp4AWHaSJqMeziCZ2cU92gUNe657c/fsmzqJZN62UCvNzHqTcxGE1O3sJoCy7KxLeFsQMggyRKyLKFqMqpLQXUpaB4Ft0/F49Nw+1XcPu2+PXZhO6LWmK9gzJYwZsuYixWnfFkCrcuPezCMazDkCPM7xNlsl9Vyg0/GMxwfcwTtalGny5J43OVhlwVy0Tkhj1yiz3WJ/tA46Ud78D56FAafA+X2mxbCEhgzRUfMXsvRzKz3kHf6nL7ZnTFc6RCS4lxPIQRnCwX+bPY6v8hLFIQHj6jSY56H6kWK5UtICB7reKwlZnuDvQ/kmkBbwN4P2gK2TZs2P3Ym1ir8j//pMicnsrhVmSMPdfFb+3rYlQoS8brwaPIPbtP8l5nvS8BKgF8IUZEkSQNOAP9ECHH6S7d5HfhvcATsIeBfCyEO3e1+v683yZagzZ92BG3DcWg1LXqLQ+v3j/64Be0GlunEsqxehbUbkLm5vsbA0jdv5wpCbHAzWiU25MSrRAcg2H3PacgAuq6Tz+dba0PY5nI5CoUCX36Out3uVinyRt9tR0cH0WgU5T7EbNi2zdLSEuPj40xMTDA3N4cQAo/Hw8jISGsFAgGnrHW8gH4zT2Oi4DhsEmjdAaykj6wtmFyssjhZwrYEqkumezRC37pDG+v+bnuEf2gIsT799lqO+vUc5lwZcDYEPOvDgjxDYSRNwbZsStkGheUahdUa5WyDUrZBOdugWtBbGwb3E0kCT9CFL+TCH3Lhj7gJxj0Eoh6CcQ/hDi+BiPtblwPbuoUxV8aYdnp5jZkSYr1EW+304R4K4xmO4B4KIz8At1sIwdhqheNjGT6+ucaZqSySbjNsqxzUVJIVC8tUAZukNk6//woDuwJ0HHoGaeQlZ2r1HTDXaut9s5ulxq3M2Z1O7uyGSG/ago9zef58fpx3CxI1oREUeXqNM4jaZ2TKl5GAXbFdvJx+mVf7X2UoPHRfXzttAbt92gK2TZs2P1YapsX//sE4f/zRJG5N5r97ZQf/+f5eQu3Im++V772EWJIkH46A/a+FEGe+9PU/Bj4UQvzZ+t9vAC8IIZbudF8/lDfJRmOxJWbzhdOtHlpnKNR6D23k0I/fof0qtgXF+a/kjk46qzAD9pcG2yguJ4cy0u84uJH+zb9H0k5cyz0+hFqWRbFYbLm3X86J3Sj9BVAUhY6ODjo7O2+J/9luOXK9XmdycpKxsTHGx8dbbnFPTw+jo6OMjo6SSqWQhIQxX0Zfd2eN2TLYAkmT0dJB6iE3S7Umk9Ml8uv9s96gRvdohJ4dUbpHI8RS/gfWG/lDwTYsx8VeFzdW0dkM0fqCLaeuqkrkF2tkFyvkFqtkF6sUV2uOg7qO6lYIxT2E4h78UQ++kCM0fUEXbp+Ky6vi8ipobhVZkZAVyZkeLTluqG0LxxU1bZqGhanbmLqFXjPRa00aVZNGxaRWNqgVDWpFnUpBp1Y0bnk8iioTSngIJ31EO31EunzEUn6iXT7c31B0CsvGWKigTxbRJwoY0+uCdj2f1TMSwTMacZxM9f7/TmmYFuen83w8tsZHN9a4sVymy5LYpyjsbTaxqx5AwisX6Pd8TnrQ5v9n782DI0nT875f1n3fuIHCffR9zbGzO3vN7L0rmhRpUjbpdYiW1nRIipD0B2kpgsGQw7JkS7JFhxhB0WGGSDkki6Qoytzlcu+5dmZnp3tm+prG2bgKV6GAuu/M/PzHV8gG+pwedDfQ3d8vIqO6C1mJLKBQXz35vu/z9L1wGs+Jz4M7eMfjmlVddi3clDnr6g3KixTjUStztmqYfH8rxx+vLPBKTtDATkyk6a2/BdVLrBan0YCB0ACf6/8cPzfycyRDt8+6vR+UgN0/h2VtVigUijshhODqaoE/e2+FN+e2KDd0qg2DYk2n2jT4uTM9/MOvHHmo8X6KD8+BCVhNuh9dAEaA3xFC/MZNX/8m8E+FEG+0/v8D4DeEEOdv2u8bwDcAksnkucXFxYd2zh+VajUlZ2izPyGb+ym1Wgp4wkyh7oWhQ34JsouQnYfteSlqs4vytprdu7/DC+FeiPTJ23Cyddu6L9Rzx5ZFkJXbnVnb3U7Ju1uSo9EoXV1dMk6np4fu7u6PbCBlmibr6+vMzMwwMzNDKiV/xz6fzxKzw8PDeL1ezLouhciMnJ/V09JZ2OZ3YO8NUnQ7WCk2WZjPU8pJUeTyOugYCNIxGKZjMERbMvjAZmgPCiGkY3BtJkdtapv6fB50geay4RwK02zzkXXY2MzU2EqV2Foto9dbs7gahBJeYl1+Yl0+Ih0+op1+wu1ePP47GzWZpkm1WqVUKlEul2k0GtTrdRqNBoZhIISwqvo2mw2n04nD4cDpdOL1evF4PHi9Xnw+3y2vFaNpUszWKG7XKGxWybe23EaFfLqKod8wuQpE3cS6/cS6AyR6/MR7g0Q7fdg/pPgUukljuSir/DNZGqmirGS67VZbtWc8iiPycKz51/JVXp3a5NXpTd6YyaBXdUYNjc/a6/jKHnTDjYZBp2ua/u48/ecGib/webRA4s7PyRQ0V0utzNnWcxJgCzrxjLWq7qMRbB4HBd3g25s5/uNqih8XDAxsdIgV+hpvQvUyy4UF/sUn/xGfG/q5fT9XJWD3jxKwCoXisNE0TK5vlpnaKHJtrcB3r64zt1nGadf42FCcmN+F12nH67LzhaOdvDB8b0d8xaPjMFRgI8B/Av6OEOLKrvu/BfyTmwTsrwshLtzpWI/LIikrtG+3XI5vmELZ7QEikWes6J5g8Dg221PSolArQG4J8svy1vr3sqzq7jgkW2gQ7JJi1qrkDtzYQj23bVEulUqsr69bcTpra2tWnq2mabS3t9PX12dt0Wj0I7Ujlstl5ubmrOpstVpF0zR6e3stQdvZ2YmmaRj5uhQiLUMosyBFqz3sxtYToOCyk640WV4ps71atkyZfGEXbX1BEn0B4j0BEr0Bwu2+RzaXer8IITC2a9LJeee5FuVzNQJOyn4naR2WtqoUczeqmW6fg0SvfI7x1m2sy4/TffuLPaZpks1m2djY2DNbvTNTbZrmbR93v7jdbgKBAMFgkFAoRDQaJRKJEI1Gicfje2ayTVNQ3KqSXauwvVa2KsjZtYolbG12jWiXn7beAIlk0Prdujz3NnEyazr12ZxVzTRysnrt6PBJF+DxKO6B0EOJ7WkaJu8uZnllepNXpjaZXC3QY8BLWon+pp1aTToHB2wZkolV+o/H6f30p3B19N/1uEapQW16pzqbRdR0sGm4B0JSzE7EcLR52W4afGszx5+urfF2UUeg0SsW+M2BEP/F0Ev7fn5KwO6fx2VtVigUTwc/uLbB3/n371kGlQ6bxtlklJ8908NXTnQ+8Ig/xYPnwAVs6yR+CygLIf75rvse2xbi+6VWXyeX3YnteYdKZQ4Am81LOHymJWifJRQ6jd3+lIYdN6tQWN0lanffLkJ+BcQup1yHB2LDcvY2MXbDOTk+Ao69b0ylUsnKhl1eXiaVSllZtoFAgP7+fmtra2vD9iFmd3djmiYrKytWdXZtTb6Eg8GgJWaHhoZwu92yKrlZpX49R30uT/16HrM1w2kLOHH0BmiG3OQM2CjU2Vgpk12vIFpZq3anjWinrEbGuvxEu3xE2n2E2rw4XY+2ui+aBrXlIqWZHPWFPOZqGVurgtrUIGPAek1nUxdUTdlyG+2S524J1p4A/ojrrhXVTCbDysoKKysrrK2tkU6n90Q2BQIBy8k6FAoRCAQIBAL4fD7cbjdutxuXy4XdbkfTNOt7GYaBruvouk6j0aBWq1GtVqlWq1QqFUqlEsVikWKxSD6fp1Ao7Dm3nZnstrY22tvb6ejooKOjY6+wNUxyG1UyK0W2UmUyqSKbyyWqrYsYaBBp99GWDNLeH6S9P0Rbf/Cuv8ud11BtSlYy6/N5MASa2y4rsxOyHdvufzgXxzYKNV6dltXZ16c3MSo6nzSyPIOBUWlDFx5sNOkOLtE/4qL/xdNEjpxAu8vflTAEjSWZOVud3EbfkF0L9pinVW2O4R4KsylM/nwzx39aT/OPR3s4Hd5/xp4SsPvncV2bFQrFk8cHqwV+4XffZDDh529+cojxziBDbX7cjie0A/IJ5aBMnNqAphAip2maF/gu8L8KIb65a5+vAn+bGyZO/6cQ4rm7HfdJWSQbjQy53PmWoD1PqXQNEGiak1DwOJHIs4QjzxAJn8PpVCHIgDSWyqcgu3Bj7nZrtjWHO39D3NocUtB2npQ5t12t211zeqZpkk6nrWzYxcVFS5x4vV4GBwetLR6P33eFtlgsMjs7y8zMDHNzc9TrdWw2G/39/ZagTSQSaJpmtdnuGPnUFwuWwzEaOBJeHF1+mj4nJSBT0clkqmyvVyht1/d8X3/YRajNSyDqIRBx44+48YVde5x1HS4bDqcNu8OGzWEDIRBmq3pqCPS6QbNhoDcMGlWdWlnOhNazNczNKlq2hr3UwF0z8BmCnWJwxRRs6YJtQ1Dx2HG0+Yh0+om0+wi3y1bgUJv3ntXjWq1GKpViaWmJ5eVlVlZWrIsNbrebrq4ua9a5vb2dRCKB2/1oWq11XadQKJDNZq1Z7J029mKxaO3n8/no6uqiu7vbamEPhUJ7XkflfJ3NpeKerZSVv0/NphHr9tMxGKJzMETnUJhIx50zds16qzo7maU6uYVZbM2ZJkN4j8qMVkeb9+Fk9BomF1M5q934SirPOT3Li6JEuBGm1OwAIOTM0N9bJfnMID0ffx6n9+6/Mz1Xk5XZyW3LlRqHhnsoYplBORIP5jkpAbt/npS1WaFQPN6kCzV+9nd+jAD+7G99go7QU1oUegI4KAF7EvgDwA7YgD8SQvxPmqb9GoAQ4ndbTsX/CvgSMkbnr988/3ozT+oi2WzmyecvkMu9Qy5/nkLhMkLICpPfP0o4fI5I+Bzh8Dm83uRT7V57W/S6FLLpazcigdYuQmlDfl2zQdsE9JyD3meg72NS5LYqQkIIcrkci4uLzM/PMz8/bwnanXzY4eFhBgcHCQQC93VqhmGwtLRkCdp0WrZKh8NhhoeHGR4eZmhoaI/hlFFq0FwpydiVVJHmWtlqGYVWFm3Cgy3qoemxUxMapYZBvtIkm2+Sz9cp5xp7ZjLvhR1w28Bn0/Bq4LVp+O0aARsEbBquXcKzoUHNZUcPuaHNi7MngL/TTyjhJRjz3FdcUD6fZ2lpydo2NuTvTNM0Ojo66Ovro7e3l+7ubuLx+H1Xxx8VlUqFjY0NNjY2rBb2dDptzdsGAgF6e3utraenB6dzb4W0nK+zuVhkY6HAxkKB9EKBekUao7n9DjoHw3QOh+kaDtM+ELptlXZnzlRG2mzRXC0D8kKI52gM79G4NIJ6SG3oW6U6b8xmeHV6k9emNwnmM7xsphkwPJTrfejCg11r0h3L0H80St+LzxBNtt31PU00TerzeVlxns6ib0ojtOgvjeM/s/+sWCVg98+TujYrFIrHh1rT4Jf+9VvMpEv88a+9wLHu8EGfkmIfHIoW4gfF07JIGkaNQuGiFLX5C+TzF9B1WeFxuRKEw2flFjpDMHgCu/3xNvt5aBQ3pJBduQAr5+XtjpmUJwLJj8lt4JOyStsyjRJCsL29zfz8vJURW6vJqmhnZycjIyMMDw/T19eHw3Hv+cXd5HI5qzI7Pz9PvV5H0zS6uroYGBhgYGCA/v7+W6qKZlWnuVGmuVZG36zSzFTRM1WMbE3mie7GBjavEzx2sNswQW5Cuu9i7GwmNt3E1jTRbvdW4HOgRT04El7cHT7cvQFcXQHswY82O2IYxp7K9/LysuUm7XK56O3tJZlMkkwm6enp2Vdl1TRNKpUK1WrVag3eMXLa2TRNw2azYbPZsNvtuFwuq+XY4/Hg9/s/sukXQLPZZGNjw2p/TqVSbG9vA9JAqquri2QySX9/P8lkEp/Pt+fxwhRkNyqsX8+zcT3P2vUC2bVy6/EaiWSQrpEw3SMRukbCeG+TH6vn6tSubVH9YIv6ddlqbAs48UzE8B5P4BmOoD2EfGKQc8EfrBV4bUaK2ZmFVT7bXOa43sTR7CKvdwMQcBdJ9kPfc+P0nhnGc4/WZ327Rm16G+/RBPbQ/ueYlIDdP0/L2qxQKA4OIQQ/nt3itZlN1vI11vNV0sU6uiHNGatNg1y1yb/+lXN84VjnQZ+uYp8oAfsEIIRJuTzTErPvks9fsIyhNM1JMHiUUOg04dBpQqGTeL39qkp7O4SQ8T/LP4Glt2DpJ7JyC+AKQN/zMPgpGPq0bEG2yQrXTj7s3NyclQ9rmiZOp5OBgQGGh4cZGRm573ZjwzBYWVmxxGwqlcI0zT3Vx50tEonc9tjCMDEKDYx8HSNXxyg0MKs6ZqWJWdERDQNhCIRhStFq09AcNjS7Bg4bNq8Dm9+J3efA5ndhj7hxRNzYw+59CZudiwA7ZlorKyusrq5as6uBQMASq8lkko6OjvvO8m00GntaeTOZDIVCgXw+T7FY5EG8vzkcDvx+v2XktLNFo1Frux+hXalUWF5etkT8ysoKhiHb39vb2+nv72dwcJD+/n78fv8tj6+Vm6xfz7M2l2dtNkd6oWhV2qOdPrpGInSPSkEbiu+NkTJrOrWpbapXt6RpUt1Ac8m5We+xOJ6JGLYPYSj1USnVdd6+vsXrMxnemFqnJ3OFF4wc7c0AW40xGsKPhkl7vELf8Q6Sz47RMRjG9hCMqXajBOz+eVrXZoVC8Wj46fw2//w7U/x0YRuXw0ZX2ENnyENHyIPTbkPTQAM+Pd7G107JgEI+AAAgAElEQVR2H/TpKh4ASsA+odQbGQr598jn3yNfeI9C4TKmKVvrnM4oweBxQqGThIInCYVO4HZ3HPAZH1JKaVh4AxZ/DPOvQ2ZK3u+Nysrs8Etyi95wVK3X68zPz1uCdqeqdre24A9Do9EglUqxsLBwy/yn1+ulu7vbmqvs6OggGo0eipbaSqVCJpMhnU5bcUbr6+vU67Lt2W6309nZuaeF9k6C/E40m01WV1ctZ+m1tTUymYwlUjVNIxwOE4lECIfDhEIhgsHgnmicHSOnnU0IgWmamKaJYRhW5E69XqdWq1Eul62tVCpZRk67DaQA/H4/iUSCRCJBW1ublUn8YdrNdV1nZWWFxcVF6/e+c/z29nYGBwcZGhqiv78fj+fWWR69aZBeLLI2m2NtVgrbRlW2HQdibrpHI3SPyLzhcPuNmVGhm9TnclQ/kNVZs9gEu4ZnJCIrs0fjD80Eaofl7QpvzGZ4fTrN2sxFPlGdY1QXNJtJ0s0RBHZcjia9Qy76zg6TPN5GKLG/bOfboQTs/lFrs0KheNCkCzV+MJnmzy+u8ubcFu1BN3/7pRF+6dk+Zcj0FKAE7FOCaeqUyzMUCu+TL1ykWLxMqTSNbB4Fl6uNYPCY3ALHCAaP4PH0omkHL4AOFcV1mH8Nrr8it8KKvD8+IoXsyOdg4EVw3aiObW9vW2J2d1twT0+PJWh7enruu8K403K7U71cXV1lY2PDEm1Op5O2tjYSiQTxeJx4PE40GiUcDuPz+R6YuG00GpaAy+Vy1ra1tcXW1hbVatXa1+Vy0d7eTmdnJ11dXXR1ddHe3n7frda7Z2NTqRQbGxtWPE4wGNxz7EQiQSwWw6mZUFyTbtaFVfm7rOWgmoNaHpoVaQZmNMDU5Wy03Ql2l7x1BcETAncIvBHwt0GgHQIdMtLJE5JtStUquVzOiu7Z3t62zJx2Ws1BCtvOzk7rZ9Hd3X3P2KadqvzCwgILCwssLS2h6/qHfj2ZpmB7tcTqTJ7VmRyrM1mqRSmIfSGXFLSjEbrHIsS6/NJIzJQOwNUrW1SvZjCyddDAPRjGeyKB91gce+jhjikYpuBiKscbMxkuXZuic/V1zukFgs0oK/WTlEw56xoOG/Sd6KTveCe9E9EPFUF0L5SA3T9qbVYoFA+C5e0Kf35ple9cWediSo4Y9Ua9fP2Ffv6bjw3gfcRpC4qDQwnYpxjDqFIsXqVQvEyp+AHF4lXKlVlEy7HXbg8QCEy0tnEC/jECgXEcjuA9jvyUIARkpmHuh3JbeEOKILsL+j8uxezI56FtHHZFs6RSKUvQrq6uIoTA5XJZ1bShoSHLhfh+aTabbG5uWoZBO1moN0e82Gw2QqEQfr/fqkB6PB4cDgd2u92a/dypPpqmSbPZpFarWRXInSiZnSrwbkKhEPF4nFgsRjweJ5FI0N7eTjgcvu/nJYQgk8mwuLjI4uIiS0tLe2Zje3p6rMptd3c3Qa8L0h/I+eb0JGzNyN9TbplbB4I18ISlIHX6we6Qvz+bA4TZErRNMOpQL0G9CI3iLecIyOPsZBLHBiE+Ki9sJMYg0IYQgnK5bFWhdyrR6XTaEt9ut5uenh5r6+3tvWulttlskkqlrFnsndeT2+1mcHDQmseORqN3/NnmNiotMSu3Hbdjb9BJ90iE7rEoPWMRYt3yokxztUz1SobqlYw0TNLA1R/CezyB93gCR+Thz9wXak3emtvip5NLNKa+z/HiNfp1O1uNI6QaJ9CFhxe/4uDUz3xq39/raRawmqb9PvA1IC2EON66Lwb8B2AAWAB+UQiRvdtx1NqsUCjuh1rTYDZdIl2skSk2WMvX+OFUmovLOQBO9UX4wtEOPnekg7GOgBqLewpRAlaxB8OoUS5PUyxdo1S8RrF0lVJpGsMoWfu43V0E/KP4rW0Yn28Yp/Mpd3Rr1uTs7Oz35bY5Ke8P98HIy1LMDn16T2RPtVq12o2vX79ONis/BwaDQQYHBxkYGGBwcPCOAuTD0mg0rGpgoVCwtnK5TK1WszZd1y3BKoRA0zSrpdbhcFgGRm63G7/fb2WqBgIBwuGw1Z57vxXV3RiGwfr6uiVWl5aWqFRk7udtZ2NLa3JeeektSL0DGx+A2WrjdfpaIrIlJiNJCHVDqEdWTt0hy236Q2MasmJbSkM5LW8LK5BburFtz0vRu4MvcSOLuOOonKFuPwpOD7quk06n98wD766kR6NRent76evrI5lM0t7efsfq+e7X0+zsrCX0E4kEIyMjjIyM0N/ff4vD8Q5CCAqZGqszWVamc6xMZ604Jm/QSfeobDfuGY8S7fShb1apXs5QvZyhuS4NpFzJoBSzJxI4oo8momB5u8Kb0+usXHmN2PIPOVbfpvrZ/4FPffGr+z72Uy5gPwWUgD/cJWD/N2BbCPFPNU37H4GoEOI37nYctTYrFIo7UWsaLG9XuJ4p8/5yjnfmt7mUytMw9iYlHOsO8VdOdfPVE130xXx3OJriaUEJWMU9EUJQq61SKk9SLk1RLs9SKs9Qqcximjeqb05nHL9vCJ9vEJ9vAJ9vEK93AK83id3+FGZt5ZZviNnrr8rKnc0J/S9IMTv6hT3VWZDtxtevX7fienaEWygUstxok8kkbW1tD32+dUfAPmwqlQqpVIrl5WVSqRSpVMqa84xGo5YTb39/P7FYDK241mrjfhUWXof8sjyQKyCjkLrPSNfo7tMQGbh/gfogMA15XluzsDkNm9dkjFP6GjRaF4NsDkiMy/PsPgM9Z6HjODjcNBoN1tfXrZ/J8vIypZJ8nNvt3vMz6e7uvm278E7lenZ2ltnZWRYWFjAMA6fTyeDgoJU5HIncPUu6kKmyMt0StFNZq0LrC7noGYvQMy4FrQ+oXW2J2VY8j7M3gO9EmxSzsUfzHmCagqurBQbiXlmN3ydPs4AF0DRtAPjmLgE7BXxGCLGmaVoX8IoQYvxux1Brs0Lx+FNtGBRrTZqmQDdMmoagaZg0dJOGYVJpGFTqOqW6Tq1pUNdNak2DatNgu9wgU2qQKdWp1A0MITBN6Qy8XqixIzccNo0TvWGeG4hxui9CR9hDW8BNIuBW7cGKPSgBq/jICGFQrS5TqVynXJmjXJ6jUpmnUpmn2dzas6/b3YnX24/Xm8Tr7cPr6cPr7cPj6cXl+mjtso8VegOW34bZ78HM92SLK0A4CaMtMTv4yT2zs0II0uk0CwsLVjVyR8Tsbp3dMW76KO25j5pyuczGxoZVbVxdXbWqzpqm0dnZaVUak8kkoVBIVjvnX78xd7w1Iw/mjcl54/5PyLijjuOyBfgwY5qQW4T1S7B2SbY5r74HlYz8us0JXSeh91m59T0H4T4EMmJpaWnJei1kMvIxLpeLZDJpVeu7urpue3Gj0WiwsLDAzMwMMzMz5HKyFau9vZ2xsTHGxsbo7e2964URWaGtsjKVIzWVZWU6SyUvL2IFom56x6P0TETpaveiLRWlmF2Rr1lnTwDviQS+Ewkc8QdvtvSwUAJWG2CvgM0JISK7vp4VQtzSIqJp2jeAbwAkk8lzi4uLj+aEFQrFA2Vmo8jv/3ieP313hfp95MfvYNMg5neTCLhIBNz43XYcNhs2m4bLbqMv5mUw4ac/7mesI4DPdcjXccWhQAlYxUOh2SxQrS5QqSxQrS5SqS5SrSxQraVoNDb37GuzufF4elpb965/9+Bxd+N2d2CzPWFvaPmUFLIz35OirFneNTv7edly3Daxpzq7Ez2zU5XbaTXdmZ/0er2WA3F7e7vleLufvNSPghCCYrFomTjtGBhtbGxYAhwgEolY4ruvr4/u7m6Zq6rXZSvw9VdklXXlAghDzqj2fxyGPiNbsduPHUx19UEjhKzWrrwLq+9C6rz8t94yvwr1SIHe9zH5/NuPgs1GqVSy3IkXFhbY3JR/V263m4GBgbvOU+9UZ6enp5mZmWFpaQnTNPF6vYyOjjI2NsbIyMhtnY1vPk5uo8LKVFYK2qkctbKsnofbvfSOR+ntDRCtG+gzWZrLcn7Y2eWXBlAnEjjbDncrmBKw2gAfQcDuRq3NCsXjRaWh88rUJn90fplXpjZxO2z81bM9HOsO47RrOGw2HHYNt8OG0y43v9uO3+3A73LgddnxOO24HTYcNu3QX1xXPH4oAat45BhGlWp1mVpthWotRa2WolpNUautUKut0Gxu3/QIG253x03ithvvLqFrtz8+FZ1b0Ouw+Oats7OBTinWhj8rK43h3lse2mg0LDOgnfiYzc1NdF239gkEAkSjUWKxGJFIhGAwaM2t+nw+XC4Xbrcbh8Nxx0Vmt4nTzrYTH1MqlSgWi5YLcT6ft7JLQWalJhIJOjo6LHHd1dV1I8dUb8gq5MLrMq5o6SfSDEuzyZbgoc/A0GdlRdKx/5bQxwKjKav0S2/LXOLFt6C4Kr/mjcqq88AnZdW+/ShoGqVSyWo9n5+f3zNPvSNmh4aGCAZvNWGr1WrMzs5agrZarWKz2ejv72dsbIzx8XFisdg9T1uYgq3VEqnJLCtTWVZmcjRr8rUQ6/bTPxCi26XhyVTRU/JihqPDZxlAOTt9h+6DjhKw2gCqhViheOLZKtV5bWaT71zZ4JXpNLWmSVvQzdc/1s8vf6yfmP8pWX8VjwVKwCoOHYZRoVZbpVZbo1Zfbf17pXW7Sr2+hhD6nsc4nfFWe3KvbE/29ePzDuD1DeByxg/dh+K7kluG6z+CuR/JKmS1JegjSeh/Uc7Q9j4r5ydvU4E0TZNsNks6nWZzc9Myb8pms7e4Ee9mx3nYZrOhafKKqWEY6LrO3d4LNE3D7/dbJk6RSIRIJGK5DweDwb1tqeUtWWFdflverlyQghWkGBt4UYrWgRels69CVmlzS9KoauF16XidXZBf87fLivTQZ+TWutCRzWYtd+Ld89QdHR0MDw8zMjJCMpm8xXDLNE1SqRRTU1NMT09bld22tjZLzN6r1dg6lmGyuVQiNbXNynSOtZkcelN2DHR1+RmOu4nWdWybstrsiHtuiNnew+EsqQSsNsBeAfvPgK1dJk4xIcSv3+0Yam1WKA4P1YbB4naZdKFOulhnPlPi9ZkMl1fyCAEdITdfOtbJl4538exAFIf9Ceh0UjxxKAGreOwQwqBeT1sV22otRa2aolpbplpNUa+vWlFAIOOApFPyEH7fMP7AGAH/BB5P96H4gHxXTBM2LssK7eKP5W2lNV/sCkrjn56z0rSo8yREB+/aVmsYhlU5LRaLVKtVGo0G9Xqder2OaZqWA7EQYo8DsdPpxOPxWJvP5yMQCOD3+28vZoSQ7rzpa3LWc+19WL0I+SX5dZsDOk9A73M3Zln98YfwQ3xCyS3tzSQut1rzE+Oyaj/8kpVJbJomGxsbljvxTsvwjqHTyMgIo6Ojt3W73t7eZnp6mqmpKRYXFzFNE5/Px9jYGBMTEwwNDcnW7w+BoZukFwqWw/H6XB69aeLWYDjhocdtw1tuogmwR9x4j8XxHkvgGgih2Q7mb/VpFrCapv174DNAAtgAfgv4M+CPgCSwBPyXQoib22b2oNZmheLgWdwq82/eXOCPz6co1W8UAWwanElG+cxYG58eb+N4dxjbAb3fKhQfFiVgFU8cptmkVku15m4XqVTmKVfmqFSuU6+vW/vt5NwGg0cJBo8RCp7A5xs+3PO2QkBmBlbOy1nJ1Duy1dRsLUbukIxraRuHttZtbEhG+Twsg6NGpRUdMyddd3ecd9PXoJ6/sV9sCLpOS9fdnmek867rcM8/PjYIIV8Hcz+SmcSLb8oZ2j2ZxJ+z5qrr9ToLCwuWQ/FOu3EikbBmYPv6+m6pzu60Gk9OTjIzM0O9XsfhcDA8PMzExARjY2M3WsM/BIZukl4ssjKdZXU6y9pcHq1p0unU6A84iQqBTYDmc+A7lsBzLI5nOILmfHQVgadZwD4o1NqsUDw6TFPwp++tcGExi2kKDCFIF+u8PrOJw6bx1RNdfO5oBx0h6fDbEfIoh1/FY4cSsIqnCl0vUi7PUCxNUipNUSp9QKk0iWHI9kqbzUMoeIJQ+BSh0GnC4TN43J0HfNb3oFmTMS1rl6S77U5cS3VXUUSzQ6QPIv0y/zTQLm+9UXAHZDXX5ZdV0R2EAY2ybO9tlHdln25COSOrq/nlGxXhHXwJmbtq5Z4ek7feu8e1KB4gzRosvQmzP5Db5jV5f6hXGoSNfl62G7uDCCHY2tqy3IkXFxcxDAOXy8Xw8DBjY2OMjo4SCAT2fAtd11lcXGRqaorJyUkKhQKappFMJhkfH2diYuJDzc3uxjBMNheLrM7kWJnOkZ7LETNMupw2Ol02HIBwaDiHI4TOtOOZiGHzPNwLTkrA7h+1NisUj4ZLqRy/+Z+vcnE5R8TnxO2wYdc0PC47XzvZza88n6Q99BTGGiqeOJSAVTz1CGFQqcxTKF6hWLhMvnCRYvEqQsh4EI+nh0j4GcLhc0Qiz+L3jx7+1mMhpMjcnJSzktl5eZtbhtKGFKI7Lrf3g2aTAtWfgFC3rOxG+mQcUHwIYsNKqB5G8ilpEDbzvZsyiT8uI5xGvyAvOrSqs9evX7cEbbEonYO7u7utuJ2urq49fwNCCNbW1picnGRqaoqNjQ1Azs1OTEwwPj5Od3f3fWcXm4ZJJlWS87PTWRrX8yRMk06nDY9NQwB6wov/RILY8504Ig/+g5kSsPtHrc0KxUdHCEGprpOrNMlWGuQqTXLVJvlqk1JNb2Wymixnq/zZ+yvE/W7+wZcn+Ktnew7/ZxWF4iOiBKxCcRtMs06xeI184T3yuQvk8uet+B+nM0Yk8izRyPNEoy88HoL2ZoSAehFqOaiXoNHazF0ZbxrgCsjKrMsv25O9sScjuuZpxmhKp+eZ70pBu1OdjQ7C+Jdh7IuQ/Dg4XAghWF9ft9yJU6kUIJ2NdwydBgcHcTqde75FNpu1xOzi4iJCCAKBgDU3e7vHfBiky3GZ1eltsle2sa0UaTMFAbv8+6u67Ii+IOFnOkicTGB/AOYjSsDuH7U2KxR3p6GbzGfKTG0UmdkosrxdYSVXZTVXY6NQQzfv/Xnc57Lz155N8nc/P0rIc//vrwrF44QSsArFh0AIQbW6RC73Drnc22Rzb1OrrQDSATkWfYFo9AVisY/j9SYP+GwVd8M0dRqNNPXGJnozR7OZR9cLGEYFUzQRZhNT6GhoaDYnNs2BZnNit/tx2AM4HEEcjiAuVxyXK4Hdfjjccj8y2UUpZqe/I02hjLq8WDHyMox9WbYb+2QrcKlUYmZmhunpaebm5mg0GtYM7E519uaYnkqlYs3Nzs7O0mg0cDqdDA8PMz4+ft9zs7vZyaFdfzdN9YNt3JkKO77V9bPtDP/iXdNdPhRKwO4ftTYrFJJa02Cr3CBTrDO9UeT95RwXUzkm14qWSLXbNLojHrrDXnoiXjrCHmI+FxGfk8jOrddJ2Ock4Hbgstuwq6xVxVOGErAKxUekWk2Rzf6EbPYttrNv0mikAfB4eolFP060JWrd7rYDPtOnDyFMKpUFypUZKuV5KpU5KpV5avU16vU0YN718ZrmAMQeN+s77+vC7W7D7e7E4+7C7emUUU7eJF5vEo+nB5vtMbka3ijLFuPpv5RbaUPOTydfgImvwsRXIDoAyBnYhYUFy6E4n5eGXT09PVZ1tqOjY8+Hqp3H7FRni8UimqbR19fH+Pg44+PjJBKJfT2F/FKRzFurtH+yh2B34N4PuAdKwO4ftTYrnhRMU9AwTArVJlvlBtlyg1y1SaHapFjTKdZ1GrpJQ5dtvYVak41CjXShzkahRrmxd00Juh2c7AtzsjfCRGeQsY4gQ21+3A5lqqRQ3A0lYG9CNBqk/t7fx3viON7Tp/EcP4E98NGqA4qnByEElcp1trNvks2+STb7E3RdZq76/aOyOht9gUjkeZxOlW36IJHV8UXy+fcoFC9RLF5tGXOVrX1crnZ8vkG8nh7cni4pNN0dOJxhnI4wDkcYu92HzeZE05yW6BLCRAgd02xgGBV0vYRhlGg2czSa2zQbWzQaGeqNNLXaGvX6GvX6OqbZsL63ptnxepP4fMP4/SP4/aMEAhP4fUPYbIc4GN40YfU9mPoLuaU/kPe3H2uJ2a/K+CZNQwjBxsaGlR27siK7E8LhsFVlHRgY2ONqvDM3OzU1xdTUFOvr0iE8Ho9bArivrw+7/WA/yCkBu3+UgFU8jixulfnO1XX+8so6V1YLNA2TD/Ox2OWw4bLbcNo1Ah4HnSEP7SEP7UE3iYCbuN9FzO9iqM3PUCKgImsUio+AErA30UitsPyNb9C4fl3eoWm4R0fxnjqJ99QpvKdO4RoeRlNzgIq7IIRBsXi1VZ19i1zuHUyzBmgEg8eIRj9GNPoCkfCzOBzqAsn9YJoNCsXL5HLnyefOky+8T7MpHZftdh+BwBEZjRQ4RiAwjs83iMMRvMdRHxxCmNQbaarVZarVRaqVBcqV65TLc1SrCwghI480zYHfN0xgV4xTIHDk8L4etq/D5F/A5Ldg+ScgTGniNfFVmPiarNK2opqKxaJVmb1+/Tq6ruNyuRgZGbFcjW9uG87lctZjFhYWMAwDj8djtSePjIx85Fbj/aAE7P5RAlZx2DFMweR6Qbb0Lud4dynHbLoEwPGeEB8bjON12XHYbDjsGiGvk7jfRdTnIup3EvQ4CXocBFwOJUgVikeAErB3wMjnqV66TPXiRarvv0/18mXMVoucLRDAe/IEnlOn8J6UwtZxn3ERiqcL06yTL1wim32LbPYt8vn3EaKBpjkIhU4SjUhBGw6fxW5XFve7MYwK+fx75HLvkM39lELhYutiAPh8Q4TDZwmHzhAOn8HvH0HTDm/rlWk2qVTmKZWnWjFO1ygWr1oGYWDD7x8hFDpFOCSjnAKBscP3nMoZmPq2FLNzP5Rzs94YjH8FjnxNRvQ4vQA0Gg3m5+ctcVoqldA0jd7eXqs629bWtqfVuF6vMzc3Z5lHlcuymt7b22sJ4M7Ozkcy86UE7P5RAlZxGBBCsFGoU6jJdt9SXeeD1QJvz29xYSFLsS4vLkZ9Tk71RXhxJMEXj3XSF1N55QrFYUMJ2A+JEILGwgLV9y9SvXSR6sWL1KemwZDzDM5kEu9pWaH1njyFZ3wMzXWI2wMVB4phVMnn32W7JWiLxcsIYWCzuQiHzhKJfoxo5HlCoVPY7e6DPt1HSrOZJZc7Lw2z8hcoFq+0qpY2gsEjRCLPEQk/SyRyDpdrf/OSh4V6PU2xeIVC4TKF4kUKhUs0m1kA7HY/4dBpQuEzRMLnCIfP4nDsf7bzgVEvyYieyW9JI6h6Hpx+aQI18TUY+4LMGwZM02Rtbc0Sszttw+Fw2DKBGhgY2ONQvPsxMzMzrK6uAtIJeXR0lLGxMYaGhnA9pPdbJWD3jxKwioOgaZgUazqXUjl+NJnmh1NplrdvjY8baQ/w/GCM5wZjnOmL0hfzKkMkheKQcyACVtO0PuAPgU6km8rvCSF++6Z9osDvA8NADfhVIcSVux33US+SZqVC7epVqpcuUX3/fSrvv4+xmQFAc7vxHD3aqtCexHPyFM6ebvWmqLgtul4klztvtRyXStcAgc3mIhS6IVzC4TM4nU9OzqoQBuXynIwryr9HPv8ulcocIM2RQqGTrciiZ1vC7dG1Ah8kO67XOz+XQv49iqVryLfLlpAPP0sk+hzRyHM4ndGDPmWJ3oDFN+DaN6WgLa2DzQH9n5BiduIrEO61di8UCkxPTzM9PW21GjudToaGhqxKaygU2vMtisUis7Oze5yQ7XY7AwMDjI6OMjo6Sjwef2BPSQnY/aMErOJBkinV+fblNQo1HcMUGKagWNNZyVVYzdVYy9co1Jo09BtmfR6njU8MJ3hxNEFb0I3f7SDgdjCY8JMIPF0XiRWKJ4GDErBdQJcQ4l1N04LABeBnhRAf7NrnnwElIcQ/0jRtAvgdIcTLdzvuQS+SQgj0tbWWoJVV2trVq4iGNHSxx+N4T5zAc/IE3pOn8J48gf2mD2cKBUCzmW9F9vyUbO6nlEofWI64Pt8IodAJQsEThEInCQSOPBZtx6apU60uUCxds6qNxeIVy2zJ4YgQDp8hEj5LOPIsoeDJp676fDd0vUShcHFXK/X7mGYdAL9/jGhU5hJHI88fjoscpgkrF2DqW1LMZqbl/V2nYLzlaNxxHFoX9ZrNpuVqPD09bbkad3Z2WsK0t7cX2y7/AV3XWVpasqqzW1tbAMRiMb7yla8wMjKy76ehBOz+Oei1WfH4Y5iC8wvb/D9vL/GXV9ZoGns/n/pcdrojMnamK+yRETMuBwGPg4GEnxeG4nich2wUQ6FQfGQORQuxpmn/GfhXQojv7brvW8A/EUK80fr/HPBxIcTGnY5zGBdJ0WhQm56hdvkS1YuXqF66dMMgCnANDsp52hMn8Z48gXtiAptqPVbchGFUKBQukctfIJ9/j2LxMo1GpvVVGz7fIIHABMHABH7/KD7fMF5v34HEt5hmnWo1RaUyR7kyT6U8R6k8Tbk8bQkuTXMRDB4hFDxJMHSccOgsPt+g6lC4D0yzIV8TrYscudx5TLMKaAQDR4nGXiAW/QSRyLPY7d6DPl3IzMDkN6URVOodQEAk2RKzX91jAiWEIJ1OW5mzy8vLCCHwer2MjIwwOjrK8PDwLaZO29vbzMzMMDs7y8svv0xnZ+e+T1sJ2P1zGNdmxeEgU6qzmquSrzbJV5tkyw3W8jXW87KSulWus1VqsF1pIAQEPQ5+/mwv//XzSfrjPuyapjJQFYqnkAMXsJqmDQCvAceFEIVd9/8vgEcI8fc1TXsOeBN4Xghx4abHfwP4BkAymTy3uLj40M95vxjFIrXLl2Wl9pK8NTItMeJ04pmYuFGpPXEC1+Cgcj1W7EEIQb2+TqF4iVLxGsXSNUqlSWq1lLWPpjlaWaR9eDy9eD09eDw9uFwJa3M4Qmjah3ttCWGi6wWazZyMkWlsUrsxQLgAACAASURBVK9vUK+vU69vUK2lqFaXqNc3gBvvHW5XBz7/MMHAEQKBCRkh4x853BEyjyE7gnanDT2ff69lFOYiHD5DLPYJYrEXCQWPH7wpVCktTaCm/gLmftQygYrC6BdlZXb4ZXDfmPOtVqvMzs4yOzvLzMwMlUoFkKZOO4K2q6trT3X2QaEE7P5RAvbpRjdMtisNKnWDUl0nXazx49ktfjybYXK9eMv+DptGR8hDR8hNe9BDPOAiHnAzmPDxpWNdeF2qkqpQPO0cqIDVNC0AvAr8YyHEn970tRDw28AZ4DIwAfwNIcTFOx3vcV0khRDo6+tUL16iduUy1UuXqV25gtn6kGbz+/EcO4bnxHG8x4/jOXECZ0+PuuKouAVdL1KuXKdSvm5VQGu1ZarVFXQ9d9vH2O0+7HY/drsXTXMAWmsD06xhGFVMs45hVJAzmHvRNDsuVxteTx8eb68UzZ4+/P7hRx5ho7iBYVTJ5c6znf0x2e03KZauAuBwhIlGXyAe+yTx+KfweLoP9kTrJelkPPktmPkOVLNgd0sn4/Evyy14o5Jqmiarq6uWmN3JnPX7/Xuqs17vg6k6KwG7fx7XtVmxP1ZzVf7d20v8v+8skSk19nzN5bDx7ECUT4wkGG0PEvY6CXudRH1O4gE3dhVFo1Ao7sKBCVhN05zAN4HvCCH+93vsqwHzwMndVdqbeZIWSWEYNK5fp3r5ihS1l69Qn5xENJsA2MNhPMeOSmF7VN46e3tVpVZxR3S9SK22RqORkVtzC72ZxzAq6EYZw6hYGaU7ae02uxubzYPd7sVu9+F0RnE6IjidEVyuBG53Jy5X7OAreop70mhssZ19k+3tN9jefoN6XToA+/2jxGOfIh7/NJHIM9hsBzh3bOiw9JaszE5+C3KtjpqeczKiZ/zL0H7UmpsFKJfLVnV2dnaWarWKpmn8/M//PMePH9/3KSkBu3+epLVZIWnoJotbZbKVJqW6jKUp1HTylQa5SpP5TJkfTaUBeGmig0+NJQi4HfjdDiJeJyd7I6qSqlAoPjIHZeKkAX8AbAsh/u4d9okAFSFEQ9O0vwl8Ugjx9bsd90lfJK152itXqF29Qu3qB9RmZqAlam2BAO6JcTwTR/AcmcA9PoF7ZBib5/Ab/CgUikeHEIJyeYat7dfY3nqdbO6nCNHAbvcRjX6cePzTJOKfxePpOsiThPQ1aQI19W1pCAVybnbsyzD+Jelu7LghuE3TZGVlhZmZGc6cOUM0un93ZiVg98+TvjY/6WwW61xbK3BtrcDkepFrawXmNku3GCnt4HXaSQRdfPVEN7/8fFLlqCoUigfOQQnYF4HXka3BOz2J/xBIAgghflfTtBeQUTsG8AHw3wkhsnc77tO4SJqNBvWZGWpXr1KfnKR2bZLa1BSi1X6M3Y5rYADP+BjusXHc42N4xsZwdKtIH4VCITGMCtnsT8hsvcrW1ivWLHUgMEEi/lniic8SDp0+2Ep7cR2m/xKm/hKu/wj0GrgCMPwSjH0JRr8AgbYH/m2VgN0/T+Pa/Dixlq+ymqtRquuU6zqZUp2ZjRIz6SKz6dKe9t/OkIcjXUEmukKMdwRJBNwEPA6CrS3sdeJ2qMqqQqF4uBy4idODRC2SEmEYNJeXqU1OUZuapD45RX16mmZrVgzkXK1rZBj3yAju4RHcw0O4hodxdnerNmSF4ilGCEG5MstW5kdktl4hnz+PEAZOZ0xWZhMvEY998mBnmxsVmH9NCtrp70BxFdCg56w0ghr7oozreQAX6ZSA3T9qbT5cCCG4ulrgux9s8N2r67c1Ugp6HIy2BxhpDzDRGWKiK8iRzhBRvzLfUygUB48SsE8RRrFIfWaG+vQ09ZlZ6rNyM1rZiQCax4NrYAD30CCugUFcg4O4BgZw9SdVZq1C8RTSbBbY3n6NTOaHZLZeRddzaJqDSOQ5EomXaEu8jNebPLgTFALWL8H0d6WgXbkACPjZ34XT/9W+D68E7P5Ra/PhoNLQ+bP3VvnDtxaYXC9i0+CZ/hifP9rBaEeAoEfOqMZ8LtqCbtWlpVAoDi1KwCrQs1ka169Tn5ujMXed+sI8jfkFmqkUmDdcZ+3RKK5kEmdvL86+Xly9vTh7enB2deHo7FSztgrFE45p6uQL75HJ/IBM5odUKnOANIJKJD5HW+IlQqFTB9tqXNqE2e/DyOceSEuxErD7R63NB4NpCq5nSry7lOPCQpa/uLJGsaZzpCvEr3wsyZeOdRIPHKBpm0KhUHxElIBV3BGz0aC5uEhjaYnGwiKN1r+bqRTNtTUwjD3722MxnJ2dODo7cXZ24OjYuZWbs6MDm0+ZOSgUTwqVyoKszGZ+QC7/TqvVOE4i8VnaEi8Ti72I3f54/80rAbt/1Nr8cCnXdV6f2eR7H6SZ2yxRaeiU6wb5apNSXTrLhzwOPjPeztdf6Odcf1RVVxUKxWPN3dZmx6M+GcXhwuZy4R4dxT06esvXhK7TXF+nmVqhub6Gvr5Oc2WV5sY6zVSKyoULmPn8rccMBHC0t7e2NhyJNhzxOI62BPZ4HEcigSMWwx6NojnUS1ChOMz4fAMkk79KMvmrNJt5trZeJbP1QzY3v8va2p9gs7mIRj9OIvEyicRLeNyd9z6oQqG4K0IIZtMl3pjN8Nr0Jj+e26Khm4S9Tk72hukMefC57IS8To51hziTjDKU8GNT2aoKheIpQKkHxR3RHA5cvbKN+E6YlQrNjQ30jTT6xjrNjTT65iZ6Oo2eTlO98C56JoOo12/zDTTskQj2WEwK2lgMeyyKIxpr3RfFHk/giMewx+PYIxF1RVmhOECczjCdnT9DZ+fPYJpNcvnzstV48wdMbb3C1NRvEgweb7Uav0wgcET9zSoU96Chm/zxhWXmN8tslxtkyg2m14usF2oADMR9/Mrz/Xz+aAfPDkRx2JUJo0KheLpRLcSKh44QArNcRt/cxNjaQt/aRt/KYGS20LPbGFvbGNvb6Nvy1sjnpWnLzTidsnrb1oajvQ1nRyfOrk7ZxtzdhbO3F0dbm3JYVigeMTuuxpnNH5DJfJ984X1A4HZ30Zb4HInEy0Sjz2GzHc5ZPNVCvH/U2vzRuJTK8et/conJ9SIep42430084KIv5uPFkQQvjiRUxqpCoXgqUS3EigNF0zTsgQD2QAAGB++5v9B1jHxeitqtLfStLSl8NzOysru5SXNxkcrbP8Us7o0G0FwunD09uJJJ6aw8OIBrYBD36AiOePwhPUOF4ulG0zQC/lEC/lEGBn6NRiNDJvMKm5nvsbr2J6RW/i12e4B47JOtVuPP4HRGD/q0FYoDo9Y0+D++P83/9dp12oJu/u//9hlePtJx0KelUCgUjwVKwCoOHZrDIWdm4/HbzubuxiiVZevy6irNVIpGKkVzOUVjaYny228jajVrX3sshntsDPfoKJ6JCTxHJnCNjGBzqcw7heJB4nIl6O7+Bbq7fwHDqJHNvsVm5vtkMj8gvfltwEY4fJa21tyszzesWo0VTw2pbIX//t9e4Opqgb/2bB//4CtHCHudB31aCoVC8digWogVTyzCNNE3Nqhfv05jdpbazIzMyJ2ZRVQqcieHA/fICJ5jR/EcO4b3+HHcExNK1D7hCCGg2ZT/sdvBZlMC6hEghEmxeIXNzA/IZH5AqXQNAK83SSL+EonES0Qiz2KzPdq/P9VCvH/U2vzheHM2w9/6d++im4J/+UunVdVVoVAo7oCK0VEodiFMk8biIvXJSWofXKN27Rq1K1cwcjkANKcT95EjeE+dwnvqFL6zZ3B2dx/wWSvuhRUJlUrJivzKKvr6OkYui57LYWRzmOUyol5HNBq3zlnbbNg8HjSfD5vXi83vlyZjkQj2SBhHPIGjox3nTmRUdzf2UOhgnuwTQq22SibzIzJbPySbfRPTbGC3+4lFP0488Vni8U8/EldjJWD3j1qb783vvzHP//ytDxhuC/B7X3+GwYT/oE9JoVAoDi1KwCoU90AIgb66SvXKVWqXL1G9eInqlSuIahUAR1cXvrNn8Z47i+/cM7hHR5RZ1AEhhEDf2KB29Sq1q1dlVX12jsbS0p7cYs3lwtHVKV2tIxHs0Si2QADN5cTmdqO1quzCNMEwEYaOqNYwq1W5lUoYuZzcslnrAsdubMEgzp4enL09uJL9uPpb2+AgjvY2VdW9Dwyjwnb2Lba2XiGT+RH1+hoAgcAR4vHPEI9/mnDoDDbbg598UQJ2/6i1+e787qtz/NNvT/LFYx38i188TcCtJrgUCoXibigBq1B8BISuU5+epvLue1TfvUDl/AX0dBoAWziM7+xZfM+cw3fuHJ6jRy1BpHiwmLUatStX5O/hvfeoXrqEsbUlv2iz4ervxz0ygmtkGPfQMK5kn6yOxuMP9CKDaDTQM5lWbNQGzdU1misrNFdWaKSWaS4ty8puC1s4jHtkBPfoCJ7xcdwTE3jGx7H5lKPovRBCUC5PSzG79Sr5/HmEMHA4QsRinyQR/wzx+KdwuRIP5PspAbt/1Np8Z/7gzQV+6/+7yl851c2//KXT2FVWq0KhUNwTJWAVigeAEIJmKkXl/AUqF85Tfec8jcVFADSPR7YbnzuH75lzeE+dwuZX7WEfBaNUpvreu1TeOU/lnXeoXrlizau6Bgfxnj6N59gxObc8MYHN6z3gM5YI00RfX6exsED9+jz1WTlvXZ+ZwSwU5E6ahmtgQJ7/cTlz7TlyRL1W7kGzWWA7+2O2tl5la+sVGo1NAI4f+206Or627+MrAXsrmqb9PeBvAAK4DPx1IUTtTvurtfn2/Id3lviN/3iZLxzt4Hd++SxOleGqUCgUHwolYBWKh4S+uUnlwrtU3r1A9fwFapOTYJpgt+M5cgTfubN4z8kqrYrxuT1GqUT1wgXKP/0plXfOU7t6VbYCOxx4jx3D9+wzeM+ew3vmNI7o4xe9IoRAX1uTs9bXJuXt1avo6+tyB5sN98gI3lMn8Zw8ie/0aVzDw6pF/Q4IYVIqXSOz9QpdnT+Hx7P/+XQlYPeiaVoP8AZwVAhR1TTtj4C/EEL8mzs95mlbm5uGyfXNMtfWCsymS2yV62yVGmyXG5QbBk3DpGmYLG1X+NRoG7/39XO4HfaDPm2FQqF4bFACVqH4/9u78yg5yvPe49+nZ5+ekRgtSKNl0IJWtI1WhKQRBuxLMAEc24ADBttJcBLfe3F8cmKwncSOzb02cZzY4ZjFxgYbbJNggjG+dkTsaDUSWtE2EtoRkpAEWmc0m2ae+0fVSK3Z0Eij6arp3+ecPtNdXV39zNPVXfXU+9Zb3aSxqoqateuCFtpVq6nZsAGvqwMgd9gwCqZPo3DadAqnTyNnyJCMPEfy9JEjnFq9mppVq4KCtbnoz8mhYNIkCmdMJzlzJgVTpvTo7ranDx+mZtMmatdvoGb9emo2bKDp+HEAEr16UVA+hcLyoJt6/sSJJPLy0hxxz6UC9lxhAbscmAycAF4EvuPuC9p7TU/fNp+sbWDV7qMs3/Uuy3ceoXL/CeobmwDIShglhbn0TebSJ5lLMi+b3GwjNyvBwN4F3H/9KApyVbyKiHSGCliRNGmqr6d246agoF29hlNr1pzpTprdvz8F5eUUTC2nsLycvHHjetzle7ypifrde6h5/fUz5xHX79oFgOXlUTBlCoXTg4K+YMqUyHQHTgd3p373bmrWrgu6UK9ZS/2OHUAwMnb+pElBF/UZ0ykoLyerqCjNEfccKmBbM7P7gYeAGmCBu9/V0fw9Zdvs7hw+WcfmAyeoPHCSTfuPs/nACXa9U4075GQZU4ZextSyEsaV9mJcaS9G9E+qa7CISBdTASsSEd7URN227UFBu3YdNWvW0LBvHxBevmfsWAomTiR/wgTyx48jb8SI2AwO5e407NtPbeVm6iorg5GcN2w4U7AnevemsLw8KFanTqVgwoTY/G/pcvroUWrWrj1z3nXtps1w+jQkEuSPHUvhjOkUzpxJ4fTpZPXune5wY0sF7LnMrAT4OXAHcAz4d+B5d3+mxXz3AfcBlJWVTdsTjgkQF01NztaDJ6k8cILN+09Q+XZQtB6pPjsY25CSAsaX9mL8oF7MHNaH8rIStaaKiHQDFbAiEdZw8BA169YFl+9Zv4HajRtpOnUqeDInh7yRI8kfM5rcESPJGzmC3JEjyR0yBMvJSUu83tTE6UOHgsGKtu+gbsd26rdtp3bbtjNdYEkkyBs1KryW7iQKJk3SeZ1doOnUKWpefz0oaFetombduqCLuhl548aSnDmL5DWzKZw2TQNDdYIK2HOZ2UeBG939T8LH9wBXu/tftveauG2bT9Q28Jln17Bk2zsA5GUnGDOwmHEDezGutJhxpb0YO7AXvQvT8zsrIpLpVMCKxIg3NlK/eze1W7ZQt2Vr8HfbtrOD/gBkZZEzaBC5ZWXklA0lp3QQOaUDyRk4kOwBA85e87ST59h6UxONx48H1z09coTThw7RcOAADQfeDv6++Sb1e/fitWcHI00UF4eXixlF/vhx5I8fT97o0STy87sqJdKOpvp6atevDwbAWvEaNWvXBpfyyc6mYMpkkrNnk7zmGgomTsSydd3J9qiAPZeZzQJ+AMwg6EL8FLDK3f+1vdfEadu871gNn/rhSnYcruLzN47lfWP7M6xvkmx1AxYRiQwVsCI9QGNVFfU7d1K3Yyf1e3bT8OZe6sOC8kzLZ6rsbLJ69yZRlCSRl4/l5Z3tstvYiHsTNJymqaYmuJ06RVNVVTCgUguJoqKgQB5aRm5ZGblXlJF7xRXkjryS7Mv7Z+RgVFHUVFtLzZo1VL+6nOrf/57azZvBnURxMcmrZ5GcM4fknDnkDh2a7lAjRQVsa2b2FYIuxKeBtcCfuntde/PHZdu8cd9xPvXUSmrqG3n07mnMHdU11xIWEZGulZYC1syGAj8CBgJNwBPu/u0W8/QGngHKgGzgm+7+w46WG5eNpEh3aqqupuHgQRoOHOD0wUM0HjsW3I4epenUKby+jqa6OrwuOLfLshKQyIKsBInCwvCWJFGUJLukD1klJWSVlJB9eX9ySkvJKi5O838oF+L00aOcWrGC6mW/p3rZMhr27wcgp6yMorlzSM6dR/LqWT16tOfzoQL24kV52+zurNpzlKeW7eY3m95mYK98fvjJGYweoN81EZGoSlcBWwqUuvsaMysGVgO3ufvmlHm+APR298+bWX9gKzDQ3evbXmq0N5IiIlHVPMpx9bLfU710KdUrVuA1NVhOTnDpornzKJpfQe6IERnXoq4C9uJFddu8cOsh/vE/t7Jp/wl65Wdzx4yh3Fcxkv7FuiyViEiUdbRtvmQnRbn7AeBAeP+kmVUCg4HNqbMBxRbsLRUBRwi6K4mISBcyM/KGDydv+HD63H0XTfX11KxeTdXiJVQtWcyhhx/m0MMPkzNoEMmKeRRVzFfrrMTW0ep6vvryZl5Yu48R/ZL8nw9N5LbyQRTm6lxwEZG465ZfcjMbBpQDK1o89QjwErAfKAbucPfWJ+CJiEiXSuTmBoM8zZ7NgM//DQ3794fF7BKOv/RLjv3subOts/MqgtbZ4cMzrnVW4sPdefPIKZZse4d/fuUNjtc08L+vu5LPXHcledm69I2ISE9xyQdxMrMiYBHwkLu/0OK5jwBzgM8BI4FXgMnufqLFfLG+1pyISJy0bJ2t374DgJzBg8PW2QqSs3pO66y6EF+8dHUhdncWbj3Mcyv3smrPUd6pCsaZmjSkN9/48CTGlfbq9phEROTipW0UYjPLAV4G/tPdv9XG878Cvu7uS8LHvwMecPfX2ltmVM+zERHpqRr27aNqyVKqFi+mevly/NQpLDeXwlmzKKqooOja+bEe2VgF7MXr7m1zQ2MTL6/fz+OLdrLl7ZMM6JXHnJH9mDashGlXlDD68mISCfUWEBGJq7ScAxue1/okUNlW8Rp6E7geWGJmA4AxwM5LFZOIiHRezuDBlNx5ByV33hG0zq5aRdWiRVQtXMTBhx7i4EMPkTtiBEXz51M0v4LCqVPPXrJJpIttefsE/+sna9l2qIoxA4r51u2T+cPJg8jRdVxFRDLCpRyFeC6wBNhAcBkdgC8QXDIHd3/MzAYRXCC9FDCC1thnOlquWmBFRKKjfvduqhYvpmrhIk6tXIk3NJBIJknOmUPR/AqS8+aRc/nl6Q6zQ2qBvXjdsW12d55d8SZffXkzvQpy+NptE/jA+AE6L1tEpAdK1yjESwmK0o7m2Q984FLFICIil1busGH0GTaMPvfcQ1N1NdWvvkrVosVULV7MyQULAMgfP57k/AqK588nf+JELEsD6kjn1DY08lfPrePXG9+mYnR/vnX7ZPoV6VI4IiKZSOPJi4hIl0gkkxTfcAPFN9yAu1P3xhtULVxE1eLFvPv4E7z76GNklZSQnDc36G48dy5ZvXunO2yJgUd+t51fb3ybB/5gLPfNG6HzW0VEMpgKWBER6XJmRv6YMeSPGUO/T99H47FjVC1dRtXiRVQvXsKJl34JWVkUTJ5M0by5JOdVkD9+HJbQeYxyru2Hqnh88Q7+qHwwfz5/ZLrDERGRNFMBKyIil1zWZZfR++YP0vvmD+KNjdRu2MDJRUExe/jb3+Hwt79DVt++JK+5hqK5c0hecw3Z/funO2xJM3fnb1/cSEFOFg/eNC7d4YiISASogBURkW5lWVkUTJlCwZQpcP/9nH73XaqXLqVqyVKqly3jxC9/CUDemDEk586haM4cCqZNI5Gncx4zzS/W7efVne/y1dsm0L9Yn7+IiKiAFRGRNMvu25fet95K71tvxZuaqK2spHrpMqqXLePIj37MkSd/gOXnUzhjRtA6O3cuuSNGaPTZHu54TQNf+1Ulk4f05o9nlqU7HBERiQgVsCIiEhmWSFBw1VUUXHUV/T59XzCy8cqVQUG7dCkH/+/XAcguLaX0K1+mqKIizRFLVzlwvIaPP/kaTU1OozvVdac5Ul3PU5+cQZYGbRIRkZAKWBERiaxEMknxtddSfO21ANS/tY/qZUExmz1gYHqDky6Vm5Vg9IAishIJsgwSCaNiVH8mDNZI1SIicpYKWBERiY3cIYPJveN2Su64Pd2hSBfrW5THd++alu4wREQk4nS9AhEREREREYkFFbAiIiIiIiISCypgRUREREREJBZUwIqIiIiIiEgsqIAVERERERGRWFABKyIiIiIiIrGgAlZERERERERiQQWsiIiIiIiIxIK5e7pj6BQzOwzs6aLF9QPe6aJldae4xg3xjT2ucYNiT4e4xg3xjf1i4r7C3ft3ZTCZRtvmS045aU05aU05OZfy0VqcctLutjl2BWxXMrNV7j493XF0VlzjhvjGHte4QbGnQ1zjhvjGHte4pTV9lq0pJ60pJ60pJ+dSPlrrKTlRF2IRERERERGJBRWwIiIiIiIiEguZXsA+ke4ALlBc44b4xh7XuEGxp0Nc44b4xh7XuKU1fZatKSetKSetKSfnUj5a6xE5yehzYEVERERERCQ+Mr0FVkRERERERGJCBayIiIiIiIjEQo8uYM3sr8xsk5ltNLOfmlm+mX3ZzPaZ2brwdlPK/A+a2XYz22pm/yNicT+XEvNuM1sXzjvMzGpSnnssXXGH8dwfxr3JzD4bTutjZq+Y2bbwb0nK/FHJeVtx/6OZbTGz9Wb2H2Z2WTg9DjmP/HoextJW7JFc183sB2Z2yMw2pkzr9LptZtPMbEP43HfMzKISt5m938xWh/GtNrPrUl6zMPxfmvN/+aWM+wJib3f96O6cy4UxsxvDdWy7mT2Q7njSwcyGmtl/m1ll+Lt4fzi93d+aTGFmWWa21sxeDh9ndE7M7DIzez7cT6k0s9nKSZv7zxmVk67aV4k8d++RN2AwsAsoCB//G/AJ4MvAX7cx/3jgdSAPGA7sALKiEneLef4J+Lvw/jBgY7rzHcYyAdgIFALZwH8Bo4CHgQfCeR4AvhGxnLcX9weA7HCeb6TEHYecR3o97yj2FvNEZl0HKoCpqTFcyLoNvAbMBgz4NfAHEYq7HBiU8vnsS3nNQmB6hHPe7vrR3TnX7YI+66zwezICyA2/P+PTHVca8lAKTA3vFwNvhL8nba73mXQDPgf8BHg5fJzROQGeBv40vJ8LXJbJOaH9/f6MyklX7atE/dajW2AJdooLzCybYCd5fwfz3gr8zN3r3H0XsB2Y2Q0xtqXduMOWg9uBn6Ypto6MA5a7+yl3Pw0sAj5EkNunw3meBm4L70cl523G7e4LwscAy4EhaYjtvbSX8/ZEJefwHrFHbV1398XAkRaTO7Vum1kp0MvdX/Vg6/GjlNekPW53X+vuzb83m4B8M8u7lPF1pJM5b1M6ci4XZCaw3d13uns98DOCzzqjuPsBd18T3j8JVBLsmHdqve9pzGwI8EHg+ymTMzYnZtaLoFB5EsDd6939GBmck1Bb+88ZlZOu2FfplkAvUo8tYN19H/BN4E3gAHDc3ReET/9PC7qF/iClGX0wsDdlEW+F07rVe8QNMA846O7bUqYND7vVLDKzed0YbksbgQoz62tmhcBNwFBggLsfgGDjDDR3PYxEzmk/7lSfImi5aRb1nEOE1/PQe+U9yut6s86u24PD+y2nd7f24k71YWCtu9elTPth2D33b9PYDbej2NtaP6KSc+lYlH6bIsHMhhH0iljB+X1ne7J/Af4GaEqZlsk5GQEcJvhNXmtm3zezJBmckw72nzM2Jymivh/eaT22gA132G8laBIfBCTN7G7gUWAkMIVgBf+n5pe0sZhuv8ZQB3E3+xjntkgdAMrcvZywe014ZK7buXslQVfbV4DfEHRLON3BSyKR8/eK28y+GD5+NpwUh5xHej2H81pfIruun4f28hyZ/HfEzK4i+Gw+nTL5LnefSHBgYR7w8XTE1oH21o9Y5Fz0OaUysyLg58Bn3f1EuuNJJzO7GTjk7qvTHUuEZBN0E300/M2rTHCXEQAACElJREFUJugamrHOY/9ZWovt726PLWCBG4Bd7n7Y3RuAF4Br3P2guze6exPwPc42lb/Fua0/Q+i4y/Gl0mbcAGGXiD8CnmueOWz2fze8v5qg//robo/6bDxPuvtUd68g6MKwDTgYduNr7s53KJw9KjlvL27M7F7gZoKddw/njXzOY7CeAx3mPfLreqiz6/ZbnNsVPV35by/u5q56/wHc4+47mqeHR7ebuzX+hPR1M2oz9g7Wj6jkXDoWqd+mdDKzHILi9Vl3fyGc3O53NgPMAW4xs90EXcuvM7NnyOycvAW85e4rwsfPExS0mZyT9vafMzknzSK/H95ZPbmAfRO42swKw65u1wOVzR9g6EME3RgBXgLuNLM8MxtOMBDOa90acaDNuMPnbgC2uPuZ7nBm1t/MssL7Iwji3tnNMZ9h4cikZlZGUID8lCC394az3Av8IrwflZy3GbeZ3Qh8HrjF3U+lzBv5nMdgPQfaXV8gBut6qFPrdth156SZXR1+v+9JeU13ajNuC0ba/hXwoLsva57ZzLLNrF94P4fgoM5G0qO92NtcPyKUc+nYSmCUmQ03s1zgToLPOqOE6+iTQKW7fyvlqfZ+a3o8d3/Q3Ye4+zCC9eJ37n43mZ2Tt4G9ZjYmnHQ9sJkMzgnt7z9nck6aRX4/vNM8AiNJXaob8BVgC8GO1o8JRtn6MbABWE/wwZWmzP9FgqP2W0njKJVtxR1Ofwr48xbzfphgsJXXgTXAH6Y550sIfkRfB64Pp/UFfkvQuvZboE8Ec95W3NsJzg1YF94ei1HOI7+etxd7OD1y6zpBcX0AaCA4avknF7JuA9PD7/YO4BHAohI38CWCrmjrUm6XA0lgdbg+bQK+TTeMVNjJ2NtdP7o757pd8Od9E8GouzuAL6Y7njTlYC5BF771Kd/Bmzr6rcmkG3AtZ0chzuicEJwitCpcV14ESpSTNvf7MyonXbWvEvWbhcGLiIiIiIiIRFpP7kIsIiIiIiIiPYgKWBEREREREYkFFbAiIiIiIiISCypgRUREREREJBZUwIqIiIiIiEgsqIAVERERkR7DzL5oZpvMbL2ZrTOzWZf4/Raa2fRL+R7nGccnzOyRdMchcqllpzsAEREREZGuYGazgZuBqe5eZ2b9gNw0hxULZpbl7o3pjkPkvagFViTCzGyime0xs7/oYJ4CM1tkZlld8H65ZrbYzHRwS0RE4qgUeMfd6wDc/R133w9gZn9nZivNbKOZPWFmFk5faGb/HG7/Ks1shpm9YGbbzOxr4TzDzGyLmT0dtuw+b2aFLd/czD5gZq+a2Roz+3czK2pjnoVm9g0ze83M3jCzeeH0c1pQzexlM7s2vF8Vvma1mf2Xmc0Ml7PTzG5JWfxQM/uNmW01s79PWdbd4futM7PHm/cZwuX+g5mtAGZfbPJFuoMKWJEIc/cNwJ3APR3M9ingha44auru9cBvgTsudlkiIiJpsICgiHvDzL5rZvNTnnvE3We4+wSggKCltlm9u1cAjwG/AD4DTAA+YWZ9w3nGAE+4+yTgBPCXqW8ctvZ+CbjB3acCq4DPtRNntrvPBD4L/H0786RKAgvdfRpwEvga8H7gQ8A/pMw3E7gLmAJ81Mymm9k4gu36HHefAjSG8zQvd6O7z3L3pecRh0jaqYAVib5DwFUdPH8Xwca2+QhxpZl9Lzz/Z0HYQtt85Pj74ZHnZ83sBjNbFh5hnpmyvBc5u2ETERGJDXevAqYB9wGHgefM7BPh0+8zsxVmtgG4jnO3rS+FfzcAm9z9QNiKuxMYGj63192XhfefAea2ePurgfHAMjNbB9wLXNFOqC+Ef1cDw87jX6sHfpMS4yJ3bwjvp77+FXd/191rwveYC1xPkJOVYVzXAyPC+RuBn5/H+4tEhroJikTf14E8M7vC3fekPmFmucAId9+dMnkU8DF3/zMz+zfgw8BS4ErgowQb9ZXAHxNs2G4BvgDcFr5+IzDj0v07IiIil07YI2khsDAsVu81s58B3wWmu/teM/sykJ/ysrrwb1PK/ebHzfvL3vKtWjw2ggLyY+cRZvN7NKYs/zTnNi6lxtfg7s3vdyZGd29qcdpPWzEa8LS7P9hGHLU671XiRi2wIhFmZjcSdO/5FW23wvYDjrWYtsvd14X3U4/s7nL3De7eBGwCfhtuDM85ehtuyOrNrLir/g8REZHuYGZjzGxUyqQpwB7OFoPvhOelfuQCFl8WDhIF8DGCg8OplgNzzOzKMJZCMxvdieXvBqaYWcLMhhJ0B+6s95tZHzMrIDgwvYzg1KCPmNnlYVx9zKy9lmGRyFMLrEhEmVk+8DBBC+knCc7F+X8tZqvh3CO0cO6R40aC83xaTk89wpx6dLlZHlB7QYGLiIikTxHwr2Z2GUGL5nbgPnc/ZmbfIzhou5ugJ1JnVRK05j4ObAMeTX3S3Q+H3ZV/amZ54eQvAW+c5/KXAbvCGDcCay4gxqXAjwl6Xf3E3VcBmNmXgAVmlgAaCM7x3dPuUkQizM72RhCRKAlHPjzm7t80s48At7h7q8GczGwvMMrda81sGPByOEAFZvbXBBvzp1pMb378fBuv6Qssdfdxl/Y/FBERiYeW20oRSR91IRaJIDMbQzC64L+EkzYQtMC2ZQGtB5K4GO+jdUuviIiIiEjaqQVWJObMrBz4nLt/vIuW9wLwoLtv7YrliYiIiIh0FbXAisScu68F/rv5ouQXIxzV+EUVryIiIiISRWqBFRERERERkVhQC6yIiIiIiIjEggpYERERERERiQUVsCIiIiIiIhILKmBFREREREQkFlTAioiIiIiISCyogBUREREREZFY+P933/eLyXJfgwAAAABJRU5ErkJggg==\n",
"text/plain": [
""
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
}
],
"source": [
"plt.figure(figsize=(16,4))\n",
"plt.subplot(1,2,1)\n",
"plt.plot(x_scale,x_test[:20,:].T)\n",
"plt.title('First 20 spectra fromthe test sample')\n",
"plt.xlabel(r'$\\lambda$ (nm)')\n",
"plt.ylabel('X')\n",
"plt.subplot(1,2,2)\n",
"plt.title('test Y labels')\n",
"plt.plot(y_test)\n",
"plt.xlabel('Sample number')\n",
"plt.ylabel('Y')\n",
"plt.show()"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Check the distributions of the train and test Y values. In this case train and test samples have similar distributions, which should be a good thing for the predictions. Some times train and test subsets have very different distributions and that difficults/degrades the models performance."
]
},
{
"cell_type": "code",
"execution_count": 6,
"metadata": {
"ExecuteTime": {
"end_time": "2020-07-19T10:10:12.912092Z",
"start_time": "2020-07-19T10:10:12.712878Z"
}
},
"outputs": [
{
"data": {
"image/png": "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\n",
"text/plain": [
""
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
}
],
"source": [
"plt.title('Distributions of Y train and test data')\n",
"sns.distplot(y_train,label='train Y')\n",
"sns.distplot(y_test,label='test Y')\n",
"plt.legend()\n",
"plt.show()"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"We can see that the both distributions are very similar. This is a preliminary indication that predictions model might work well for this data. Many times, specially in datasets of biological tissues spectra, the training and test distributions are quite different rendering the predictions less accurate."
]
},
{
"cell_type": "markdown",
"metadata": {
"ExecuteTime": {
"end_time": "2020-07-01T08:15:01.049570Z",
"start_time": "2020-07-01T08:15:00.838033Z"
},
"heading_collapsed": true
},
"source": [
"## PLS model (baseline)"
]
},
{
"cell_type": "markdown",
"metadata": {
"hidden": true
},
"source": [
"### Data pre-prossessing for the PLSR model (polynomial detrend + SNV)"
]
},
{
"cell_type": "markdown",
"metadata": {
"hidden": true
},
"source": [
"Before building the PLSR model (that will serve as baseline), the authors preprocessed the data by applying a second order polynomial detrend and Standard Normal Variate (SNV). For the detrend we will use the detrend function implemented by the [obspy.signal.detrend](https://docs.obspy.org/packages/autogen/obspy.signal.detrend.polynomial.html) package that we import individually here for convenience."
]
},
{
"cell_type": "code",
"execution_count": 7,
"metadata": {
"ExecuteTime": {
"end_time": "2020-07-20T05:58:00.475768Z",
"start_time": "2020-07-20T05:58:00.139660Z"
},
"hidden": true,
"scrolled": false
},
"outputs": [
{
"data": {
"image/png": "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\n",
"text/plain": [
""
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
}
],
"source": [
"from detrend import polynomial\n",
"\n",
"## NOTE: This function performs detrend on the data directly (in place). In order to not modify the original\n",
"## data we have to use copy(). Example of detreind a spectrum:\n",
"x_sample_detrended = polynomial(x_test[0].copy(),order=2,plot=True) \n",
"\n",
"## Note: In some machines (due to packages versions combinations), you might get a LinAlgError. \n",
"## If that happens just try to run the cell again."
]
},
{
"cell_type": "markdown",
"metadata": {
"ExecuteTime": {
"end_time": "2020-06-30T08:40:58.494567Z",
"start_time": "2020-06-30T08:40:58.484582Z"
},
"hidden": true
},
"source": [
"So we have to go sample by sample and perform detrend and followed by SNV correction. We create a couple of functions that loop through all samples."
]
},
{
"cell_type": "code",
"execution_count": 8,
"metadata": {
"ExecuteTime": {
"end_time": "2020-07-20T05:58:06.427859Z",
"start_time": "2020-07-20T05:58:06.418872Z"
},
"hidden": true
},
"outputs": [],
"source": [
"## Define a function to perform Standard Normal Variate on all spectra \n",
"def snv(input_data):\n",
" # Define a new array and populate it with the corrected data \n",
" data_snv = np.zeros_like(input_data)\n",
" for i in range(input_data.shape[0]):\n",
" # Apply correction\n",
" data_snv[i,:] = (input_data[i,:] - np.mean(input_data[i,:])) / np.std(input_data[i,:])\n",
" return data_snv\n",
"\n",
"## Define a function detrend all spectra\n",
"def det(input_data):\n",
" data_det = np.zeros_like(input_data)\n",
" for i in range(input_data.shape[0]):\n",
" # Apply detrend\n",
" data_det[i,:] = polynomial(input_data[i,:].copy(), order=2, plot=False)\n",
" return data_det"
]
},
{
"cell_type": "code",
"execution_count": 9,
"metadata": {
"ExecuteTime": {
"end_time": "2020-07-20T05:58:10.803947Z",
"start_time": "2020-07-20T05:58:09.843830Z"
},
"hidden": true,
"scrolled": false
},
"outputs": [
{
"data": {
"image/png": "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\n",
"text/plain": [
""
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
}
],
"source": [
"## Apply detrend and SNV to train and test datasets\n",
"x_train_prep = snv(det(x_train)) # apply SNV to detrended signal \n",
"x_test_prep = snv(det(x_test))\n",
"\n",
"## Plot detrent+SNV data\n",
"plt.figure(figsize=(12,4))\n",
"plt.subplot(1,2,1)\n",
"plt.title('Preprocessed test set spectra: Detrend + SNV')\n",
"# plt.plot(x_scale, x_train_prep.T, 'k')\n",
"plt.plot(x_scale, x_test_prep.T)\n",
"plt.xlabel(r'$\\lambda$ (nm)')\n",
"\n",
"## Plot SNV+detrent data for educational purposes\n",
"plt.subplot(1,2,2)\n",
"plt.title('Preprocessed test set spectra: SNV + Detrend')\n",
"plt.plot(x_scale,det(snv(x_train)).T)\n",
"\n",
"plt.show()"
]
},
{
"cell_type": "markdown",
"metadata": {
"hidden": true
},
"source": [
"In the previous figures, on the left panel, we implemented detrend followed by SNV (like the authors suggest) and on the right panel, SNV followed by detrend (for discussion purposes). This is just to highlight that the order in which the preprocessing methods are applied is important. Many authors do not specify the order of preprocessing when several methods are applied and that becomes an inconvenient. In case you are not familiar with these preprocessing steps, here is a quick summary of what they are doing. The \"detrend\" process fits a curve (in this case a 2nd order polynomial) to the spectra and subtracts it from the spectra. This usually centers the spectra around zero and eliminates small/known spectral distortions. SNV stands for Standard Normal Variate and consists in finding the mean value for each spectrum and subtract it from the spectrum followed by a division by the standard deviation of that spectrum (i.e. these operations are done on a single spectrum (single row), not over many samples. SNV transforms the data in each spectrum as being centered on zero with std = 1. "
]
},
{
"cell_type": "markdown",
"metadata": {
"hidden": true
},
"source": [
"### PLS optimization"
]
},
{
"cell_type": "code",
"execution_count": 6,
"metadata": {
"ExecuteTime": {
"end_time": "2020-07-20T05:57:54.803475Z",
"start_time": "2020-07-20T05:57:54.768507Z"
},
"hidden": true
},
"outputs": [],
"source": [
"## Help functions to help tune the PLS models, etc... \n",
"\n",
"\n",
"## Function to help find the best number of components of the PLS based on 10 CV MSE Loss\n",
"def pls_optimization(X, Y, plot_components=False):\n",
" \"\"\"\n",
" This function finds the optimal number of PLS components (a.k.a Latent Variables, #LV) that best models the data\n",
" based on mean squared error (MSE) loss and 10 Cross Validation (CV)\n",
" X - The training data X\n",
" Y - The training data Y\n",
" plot_components - Plot the model's optimization and optimized model\n",
" \"\"\"\n",
" ## Run PLS for a variable number of components (LVs), up to nmax, and compute mean of 10 CV error metrics\n",
" cv_mse=[]\n",
" nmax=30\n",
" \n",
" print('Computing optimal number of LVs for PLS model in the range 1 to {}...\\n'.format(nmax))\n",
" component = np.arange(1, nmax)\n",
" for i in component:\n",
" pls = PLSRegression(n_components=i)\n",
" cv_score=cross_val_score(pls, X, Y, cv=KFold(10, shuffle = True),\\\n",
" n_jobs=-1, scoring='neg_mean_squared_error')\n",
" ## This scikit-learn scorer returns a negative MSE so we multiply it by -1 when saving its mean value\n",
" cv_mse.append(-np.mean(cv_score))\n",
" ## Trick to update computation status on the same line (like a progress bar / counter)\n",
" counter = 100*(i+1)/nmax\n",
" stdout.write(\"\\r%d%% completed\" % counter)\n",
" stdout.flush()\n",
" stdout.write(\"\\n\")\n",
" \n",
" ## Calculate and print the position of minimum (indices start on 0, LV starts on 1)\n",
" cv_msemin = np.argmin(cv_mse)\n",
" print(\"Suggested number of components based in Mean of 10 CV MSE loss: \", cv_msemin+1)\n",
" print('Minimum found in Mean of 10 CV MSE: {}'.format(np.min(cv_mse)))\n",
" stdout.write(\"\\n\")\n",
" \n",
" ## Define PLS with suggested optimal number of components and fit train data\n",
" pls = PLSRegression(n_components=cv_msemin+1)\n",
" pls.fit(X, Y)\n",
" \n",
" ## Get predictions for calibration (a.k.a. train) set\n",
" Y_pred = pls.predict(X) \n",
" \n",
" # Calculate and print error scores\n",
" R2_p = r2_score(Y, Y_pred)\n",
" mse_p = mean_squared_error(Y, Y_pred)\n",
" rmse_p = np.sqrt(mse_p)\n",
" sep = np.std(Y - Y_pred)\n",
" \n",
" print('\\nError metrics for best PLS model:')\n",
" print('R2: %5.3f' % R2_p)\n",
" print('Root Mean Squared Error (RMSE): %5.3f' % rmse_p)\n",
" print('Standard Deviation of Prediction Error (SEP): %5.3f' % sep)\n",
" \n",
" ## Plots: MSE vs. PLS LV and regression for best PLS model \n",
" rangey = max(Y) - min(Y)\n",
" rangex = max(Y_pred) - min(Y_pred)\n",
" \n",
" if plot_components is True:\n",
" plt.figure(figsize=(15,5))\n",
" ax1=plt.subplot(1,2,1)\n",
" ax1.plot(component, np.array(cv_mse), '-v', color = 'blue', mfc='blue')\n",
" ax1.plot(component[cv_msemin], np.array(cv_mse)[cv_msemin], 'P', ms=10, mfc='red')\n",
" plt.xlabel('Number of PLS components')\n",
" plt.ylabel('Mean of 10 CV MSE loss')\n",
" plt.title('# PLS components')\n",
" plt.xlim(left=-1)\n",
" \n",
" ## linear fit to predicted data\n",
" z = np.polyfit(np.ravel(Y), np.ravel(Y_pred), 1) \n",
" ax2 = plt.subplot(1,2,2, aspect=1)\n",
" ax2.scatter(Y,Y_pred,c='k',s=2)\n",
" ax2.plot(Y, z[1]+z[0]*Y, c='blue', linewidth=2,label='linear fit')\n",
" ax2.plot(Y, Y, color='orange', linewidth=1.5, label='y=x')\n",
" plt.ylabel('Predicted')\n",
" plt.xlabel('Measured')\n",
" plt.title('Prediction based on best PLS model')\n",
" plt.legend(loc=4)\n",
" \n",
" # Print the scores on the plot for convenience\n",
" plt.text(min(Y_pred)+0.02*rangex, max(Y)-0.1*rangey, 'R$^{2}=$ %5.3f' % R2_p)\n",
" plt.text(min(Y_pred)+0.02*rangex, max(Y)-0.15*rangey, 'RMSE: %5.3f' % rmse_p)\n",
" plt.show() \n",
" return \n",
"\n",
"\n",
"\n",
"## Function that computes the PLS model and metrics for a train(calib) / test(valid) set pair and a given number of LV\n",
"def pls_prediction(X_calib, Y_calib, X_valid, Y_valid, components, plot_components=False):\n",
" \"\"\"\n",
" Very similar to the previous function but without the CV optimization part.\n",
" This function is simply used to compute the PLS model on train data and the error metrics on both\n",
" train and test data. \n",
" NOTE: Check if the Y data is in its original scale before interpreting the error metrics.\n",
" \"\"\"\n",
" i=components\n",
" pls = PLSRegression(n_components=i)\n",
" ## Fit PLS model to train/calib data\n",
" pls.fit(X_calib, Y_calib)\n",
" ## Predict test/validation data\n",
" Y_valid_pred = pls.predict(X_valid)\n",
" ## Predict train/calibration data (for metric purposes)\n",
" Y_calib_pred = pls.predict(X_calib)\n",
" \n",
" ## Compute test error scores\n",
" score_p = r2_score(Y_valid, Y_valid_pred)\n",
" mse_p = mean_squared_error(Y_valid, Y_valid_pred)\n",
" rmse_p = np.sqrt(mse_p)\n",
" sep = np.std(Y_valid-Y_valid_pred)\n",
" \n",
" ## Compute train error scores \n",
" score_p0 = r2_score(Y_calib, Y_calib_pred)\n",
" mse_p0 = mean_squared_error(Y_calib, Y_calib_pred)\n",
" rmse_p0 = np.sqrt(mse_p0)\n",
" sep0 = np.std(Y_calib-Y_calib_pred)\n",
" \n",
" print('ERROR METRICS: \\t TRAIN \\t\\t TEST')\n",
" print('--------------------------------------')\n",
" print('R2: \\t\\t %5.3f \\t\\t %5.3f' % (score_p0, score_p ))\n",
"# print('Mean Squared Error (MSE): %5.3f' % mse_p)\n",
" print('RMSE: \\t\\t %5.3f \\t\\t %5.3f' % (rmse_p0, rmse_p))\n",
" print('SEP : \\t\\t %5.3f \\t\\t %5.3f' % (sep0, sep))\n",
" \n",
" ## Plot regression for PLS predicted data\n",
" rangey = max(Y_valid) - min(Y_valid)\n",
" rangex = max(Y_valid_pred) - min(Y_valid_pred)\n",
"\n",
" if plot_components is True:\n",
" plt.figure(figsize=(5,5))\n",
" z = np.polyfit(np.ravel(Y_valid), np.ravel(Y_valid_pred), 1)\n",
" ax = plt.subplot(aspect=1)\n",
" ax.scatter(Y_valid,Y_valid_pred,c='k',marker='o',s=20, alpha=0.6)\n",
" ax.plot(Y_valid, z[1]+z[0]*Y_valid, c='blue', linewidth=2,label='linear fit')\n",
" ax.plot(Y_valid, Y_valid, 'k--', linewidth=1.5, label='y=x')\n",
" plt.ylabel('Predicted')\n",
" plt.xlabel('Measured')\n",
" plt.title('Prediction from PLS')\n",
" plt.legend(loc=4)\n",
" \n",
" # Print the scores on the plot\n",
" plt.text(min(Y_valid_pred)+0.02*rangex, max(Y_valid)-0.1*rangey, 'R$^{2}=$ %5.3f' % score_p)\n",
" plt.text(min(Y_valid_pred)+0.02*rangex, max(Y_valid)-0.15*rangey, 'RMSE: %5.3f' % rmse_p)\n",
" plt.text(min(Y_valid_pred)+0.02*rangex, max(Y_valid)-0.2*rangey, 'SEP: %5.3f' % sep)\n",
" plt.show() \n",
" \n",
" return"
]
},
{
"cell_type": "markdown",
"metadata": {
"hidden": true
},
"source": [
"We first run the pls_optimization() to find the number of PLS components a.k.a latent variables (LV) that best models the preprocessed train data. This is done by sampling the train dataset using a 10 fold cross-validation (CV) procedure and looking for the lowest MSE (Mean Squared Error). Please note that this method is not always the best because sometimes the best CV MSE on the train set is achieved with a high number of LVs. In these cases we effectively get a low MSE on the training set but the model does not generalizes well when applied to the test set. This is a somehow black magic art of people working in this research field. Over the year they develop an intuition about the best LV to choose. One of the expert tips is to choose the LV number where the monotonicity of the MSE curve first changes, i.e., where MSE has its first 'bump' or flattens out. "
]
},
{
"cell_type": "code",
"execution_count": 21,
"metadata": {
"ExecuteTime": {
"end_time": "2020-07-16T07:21:06.889278Z",
"start_time": "2020-07-16T07:21:05.759311Z"
},
"hidden": true
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Computing optimal number of LVs for PLS model in the range 1 to 30...\n",
"\n",
"100% completed\n",
"Suggested number of components based in Mean of 10 CV MSE loss: 10\n",
"Minimum found in Mean of 10 CV MSE: 0.2153663280312077\n",
"\n",
"\n",
"Error metrics for best PLS model:\n",
"R2: 0.925\n",
"Root Mean Squared Error (RMSE): 0.428\n",
"Standard Deviation of Prediction Error (SEP): 0.428\n"
]
},
{
"data": {
"image/png": "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\n",
"text/plain": [
""
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
}
],
"source": [
"## Find the number of PLS LV that best simulates dataset 1\n",
"pls_optimization(x_train_prep, y_train,plot_components=True)"
]
},
{
"cell_type": "markdown",
"metadata": {
"hidden": true
},
"source": [
"We get 10 LV as the optimal number for the PLS model (this is where the 10CV MSE is lower). However we can see from the left panel that around LV=8 or 9 the MSE stabilizes. As mentioned before, the experts advice is that in practice, choosing the values of LV where the MSE on the train set stabilizes usually produce better error metrics in the test set. In this case, the authors also choose to use LV=8 for their PLS model. You can change the LV bellow and see what I'm talking about (marginal gains though). This is one of those things in chemometrics that are very difficult to automate because it depends on the intuition of the researcher for that dataset. These automated methods of PLS optimization like the one I'm presenting here usually do a good job, but they don't always do the best job possible."
]
},
{
"cell_type": "code",
"execution_count": 40,
"metadata": {
"ExecuteTime": {
"end_time": "2020-07-16T07:31:10.165027Z",
"start_time": "2020-07-16T07:31:09.987820Z"
},
"hidden": true,
"run_control": {
"marked": false
}
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Optimal PLS model applyed to full dataset from instrument 1\n",
"\n",
"ERROR METRICS: \t TRAIN \t\t TEST\n",
"--------------------------------------\n",
"R2: \t\t 0.919 \t\t 0.940\n",
"RMSE: \t\t 0.445 \t\t 0.425\n",
"SEP : \t\t 0.445 \t\t 0.422\n"
]
},
{
"data": {
"image/png": "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\n",
"text/plain": [
""
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
}
],
"source": [
"## Applying the PLS model with the best number of latent variables to the full datasets\n",
"## and compute error metrics fror train and test sets\n",
"\n",
"## Visualize the prediction of the PLS model and error metrics\n",
"print('Optimal PLS model applyed to full dataset from instrument 1\\n') \n",
"pls_prediction(x_train_prep,y_train, x_test_prep , y_test, components=8, plot_components=True) "
]
},
{
"cell_type": "markdown",
"metadata": {
"hidden": true
},
"source": [
"These results for the test set are consistent with those found by the authors. These error metrics will be used as baseline for comparison with the CNN model."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## CNN model"
]
},
{
"cell_type": "markdown",
"metadata": {
"ExecuteTime": {
"end_time": "2020-07-04T07:55:50.189990Z",
"start_time": "2020-07-04T07:55:50.065861Z"
}
},
"source": [
"The CNN model that Cui, Fearn 2018 propose is composed by an input layer with the same size of the input data (as usual), followed by a Conv1d layer and 3 Fully Connected layers (FC). However, the exact CNN hyperparameters they implemented in the model used to analyze this data set (dataset3) are not very clearly laid out in the paper (at least for me!)."
]
},
{
"cell_type": "markdown",
"metadata": {
"ExecuteTime": {
"end_time": "2020-07-16T09:53:07.537080Z",
"start_time": "2020-07-16T09:53:07.514338Z"
}
},
"source": [
"### Data pre-prossessing for the CNN model"
]
},
{
"cell_type": "markdown",
"metadata": {
"ExecuteTime": {
"end_time": "2020-07-04T07:55:50.189990Z",
"start_time": "2020-07-04T07:55:50.065861Z"
}
},
"source": [
"By reading the paper I get that the preprocessing method the authors proposed for dataset 3 is the standard normalization (standardization) of the spectral samples (on columns). However, as we will see, this type of preprocessing does not lead to the best results in our model. In order to squeeze the last drops of performance of this model, we need to first standardize the data on rows (procedure analogous to SNV) and after that standardize it again on columns. I thank C.Cui for the useful tip regarding this issue. \n",
"\n",
"That being said, our first step is to standardize the input features to have zero mean and standard deviation of one ( column axis, and row+column axis )."
]
},
{
"cell_type": "code",
"execution_count": 10,
"metadata": {
"ExecuteTime": {
"end_time": "2020-07-20T05:58:23.601012Z",
"start_time": "2020-07-20T05:58:23.589029Z"
}
},
"outputs": [],
"source": [
"## Define a couple of help functions\n",
"## Since the test set is unknown (we are not suppose to have access to it during the\n",
"## optimization of the model) the scalling process should take this into account. We\n",
"## have to define a scaler based only on the train data, and apply it to the test data.\n",
"\n",
"def standardize_row(X_train, X_test):\n",
" scaler = StandardScaler()\n",
" ## rows are scaled individually\n",
" X_train_scaled = scaler.fit_transform(X_train.T)\n",
" X_test_scaled = scaler.fit_transform(X_test.T)\n",
" return [X_train_scaled.T, X_test_scaled.T]\n",
"\n",
"def standardize_column(X_train, X_test):\n",
" scaler = StandardScaler().fit(X_train)\n",
" ## for columns we fit the scaler to the train set and apply it to the test set\n",
" X_train_scaled = scaler.transform(X_train)\n",
" X_test_scaled = scaler.transform(X_test)\n",
" return [X_train_scaled, X_test_scaled]"
]
},
{
"cell_type": "code",
"execution_count": 30,
"metadata": {
"ExecuteTime": {
"end_time": "2020-07-20T06:16:28.257407Z",
"start_time": "2020-07-20T06:16:28.242426Z"
}
},
"outputs": [],
"source": [
"## Standardize on columns\n",
"x_train_scaled_col, x_test_scaled_col = standardize_column(x_train, x_test)\n",
"\n",
"## Standardize on rows and columns\n",
"x_train_scaled_row, x_test_scaled_row = standardize_row(x_train, x_test)\n",
"x_train_scaled_rowcol, x_test_scaled_rowcol = standardize_column(x_train_scaled_row, x_test_scaled_row)"
]
},
{
"cell_type": "code",
"execution_count": 12,
"metadata": {
"ExecuteTime": {
"end_time": "2020-07-20T05:58:29.973587Z",
"start_time": "2020-07-20T05:58:29.220159Z"
}
},
"outputs": [
{
"data": {
"image/png": "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\n",
"text/plain": [
""
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
}
],
"source": [
"plt.figure(figsize=(15,3))\n",
"plt.subplot(131)\n",
"plt.title('Original test data')\n",
"plt.plot(x_test.T)\n",
"plt.subplot(132)\n",
"plt.title('Scaled on column')\n",
"plt.plot(x_test_scaled_col.T)\n",
"plt.subplot(133)\n",
"plt.title('Scaled on row followed by column')\n",
"plt.plot(x_test_scaled_rowcol.T)\n",
"plt.show()"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### CNN model implementation"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Next we implement the CNN model. Some remarks before the implementation done here:\n",
" \n",
"\n",
"- **Weights initialization:** The authors mention that \"*the weights are initialized by a zero-mean Gaussian distribution whose standard deviation is $\\sqrt{\\frac{2}{n_i}}$, where $n_i$ is the number of input neurons*\". This corresponds to he_normal() initializer in tf.keras.\n",
"\n",
"\n",
"\n",
"- **Loss function:** The loss function used by *Cui, Fearn 2018* is $Loss = MSE + \\frac{1}{2} \\lambda \\sum{w_i^2}$, i.e. the mean squared error (MSE) plus an L2 penalty $\\lambda$ on the weights $w_i$. Instead of crafting a custom loss function we can achieve the same result by implementing L2 regularization. According to [Keras documentation](https://keras.io/api/layers/regularizers/) \"*Regularizers allow you to apply penalties on layer parameters (e.g. weights) during optimization. These penalties are summed into the loss function that the network optimizes*\". Therefore we will choose the loss function as 'mse' and apply an L2 regularization to the weights (by using tf.keras.regularizers.l2(beta)). The regularization hyperparameter $\\lambda$ mentioned in the paper is related to keras $\\beta$ as $\\beta=\\frac{\\lambda}{2}$ (see [tf.keras documentation](https://www.tensorflow.org/api_docs/python/tf/keras/regularizers/l2)). \n",
"\n",
"\n",
"- **Optimizer:** We use the Adam optimizer here. In the paper, the authors use the Adam optimizer for dataset 1 and 2 but for dataset 3 they used the L-BFGS second-order optimizer for full batch optimization. Since Tensorflow does not provide a native L-BFGS optimizer we proceed with Adam. In the next section of this notebook I'll show a possible alternative implementation of L-BFGS optimization."
]
},
{
"cell_type": "code",
"execution_count": 31,
"metadata": {
"ExecuteTime": {
"end_time": "2020-07-20T06:16:33.147126Z",
"start_time": "2020-07-20T06:16:33.061187Z"
}
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Model: \"sequential_1\"\n",
"_________________________________________________________________\n",
"Layer (type) Output Shape Param # \n",
"=================================================================\n",
"reshape_1 (Reshape) (None, 100, 1) 0 \n",
"_________________________________________________________________\n",
"conv1d_1 (Conv1D) (None, 100, 1) 6 \n",
"_________________________________________________________________\n",
"flatten_1 (Flatten) (None, 100) 0 \n",
"_________________________________________________________________\n",
"dense_4 (Dense) (None, 36) 3636 \n",
"_________________________________________________________________\n",
"dense_5 (Dense) (None, 18) 666 \n",
"_________________________________________________________________\n",
"dense_6 (Dense) (None, 12) 228 \n",
"_________________________________________________________________\n",
"dense_7 (Dense) (None, 1) 13 \n",
"=================================================================\n",
"Total params: 4,549\n",
"Trainable params: 4,549\n",
"Non-trainable params: 0\n",
"_________________________________________________________________\n",
"None\n"
]
}
],
"source": [
"## Make computations reproducible\n",
"reproducible_comp()\n",
"\n",
"\n",
"## Layers dimensions\n",
"INPUT_DIMS = np.shape(x_train)[1]\n",
"CONV1D_DIMS = INPUT_DIMS\n",
"K_NUMBER = 1\n",
"K_WIDTH = 5\n",
"K_STRIDE = 1\n",
"FC1_DIMS = 36\n",
"FC2_DIMS = 18\n",
"FC3_DIMS = 12\n",
"OUT_DIMS = 1\n",
"\n",
"## L2 regularizer parameter\n",
"beta= 0.003/2.\n",
"\n",
"## For the sake of simplicity we do the weights initialization for multiple layers here\n",
"## Due to this, we might have to re-run this cell before each experiment to ensure proper weight initialization each time\n",
"K_INIT = tf.keras.initializers.he_normal(seed=42)\n",
"\n",
"## Weights L2 regularization \n",
"K_REG = tf.keras.regularizers.l2(beta)\n",
"\n",
"\n",
"model_cnn = keras.Sequential([ keras.layers.Reshape((INPUT_DIMS, 1),input_shape=(INPUT_DIMS,)), \\\n",
" keras.layers.Conv1D(filters=K_NUMBER, \\\n",
" kernel_size=K_WIDTH, \\\n",
" strides=K_STRIDE, \\\n",
" padding='same', \\\n",
" kernel_initializer=K_INIT,\\\n",
" kernel_regularizer=K_REG,\\\n",
" activation='elu',\\\n",
" input_shape=(CONV1D_DIMS,1)), \\\n",
" keras.layers.Flatten(),\n",
" keras.layers.Dense(FC1_DIMS, \\\n",
" kernel_initializer=K_INIT, \\\n",
" kernel_regularizer=K_REG, \\\n",
" activation='elu'),\n",
" keras.layers.Dense(FC2_DIMS, \\\n",
" kernel_initializer=K_INIT,\\\n",
" kernel_regularizer=K_REG,\\\n",
" activation='elu'),\n",
" keras.layers.Dense(FC3_DIMS, \\\n",
" kernel_initializer=K_INIT, \\\n",
" kernel_regularizer=K_REG, \\\n",
" activation='elu'),\n",
" keras.layers.Dense(1, kernel_initializer=K_INIT, \\\n",
" kernel_regularizer=K_REG,\\\n",
" activation='linear'),\n",
" ])\n",
"\n",
"print(model_cnn.summary())"
]
},
{
"cell_type": "code",
"execution_count": 36,
"metadata": {
"ExecuteTime": {
"end_time": "2020-07-19T08:45:07.696913Z",
"start_time": "2020-07-19T08:45:07.483504Z"
}
},
"outputs": [
{
"data": {
"image/png": "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\n",
"text/plain": [
""
]
},
"execution_count": 36,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"plot_model(model_cnn, to_file='Cui_cnn.png', show_shapes=True, show_layer_names=True, \\\n",
" rankdir='TB', expand_nested=False, dpi=96)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"#### Apply CNN to data standardized on columns"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Note: the number of training epochs chosen is based on figure 5 of the paper. Alternatively we can make the training stop automatically using early_stop (included below in a comment line). On a first try there is no problem in overfitting the model, as long as we deal with it afterwards. In this case I'm using plot_losses to check/visualize the performance on the test set (passed into the model as validation_data). Note that 'validation_data' in model.fit() is \"*data on which to evaluate the loss and any model metrics at the end of each epoch. The model will not be trained on this data.*\""
]
},
{
"cell_type": "code",
"execution_count": 51,
"metadata": {
"ExecuteTime": {
"end_time": "2020-07-16T10:09:11.311214Z",
"start_time": "2020-07-16T10:00:58.958335Z"
},
"scrolled": false
},
"outputs": [
{
"data": {
"image/png": "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\n",
"text/plain": [
""
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Loss\n",
"\ttraining \t (min: 1.215, max: 87.975, cur: 1.245)\n",
"\tvalidation \t (min: 1.692, max: 57.006, cur: 1.730)\n",
"Mean Squared Error\n",
"\ttraining \t (min: 1.140, max: 87.775, cur: 1.171)\n",
"\tvalidation \t (min: 1.618, max: 56.804, cur: 1.656)\n",
"\n",
"Epoch 01500: val_loss did not improve from 1.69244\n",
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"2/2 [==============================] - 0s 155ms/step - loss: 1.2451 - mse: 1.1715 - val_loss: 1.7297 - val_mse: 1.6561\n"
]
}
],
"source": [
"## plot losses in real time during training process\n",
"plot_losses = PlotLossesKerasTF()\n",
"\n",
"##### TRAIN MODEL WITH ADAM OPTIMIZER #########\n",
"BATCH=256 \n",
"LR=0.01*BATCH/256.\n",
"\n",
"print('Adam learning rate = {}'.format(LR))\n",
"model_cnn.compile(optimizer=tf.keras.optimizers.Adam(learning_rate=LR), loss='mse', metrics=['mse']) \n",
"\n",
"#### We can play around with different the following methods to check for improvements. In this base notebook only checkpoint is used! \n",
"### 1) Stop the training if it does not improve\n",
"# early_stop = keras.callbacks.EarlyStopping(monitor='val_loss', min_delta=1e-3, patience=50, mode='auto', restore_best_weights=True)\n",
"#\n",
"### 2) Reduce learning rate dynamically\n",
"# rdlr = ReduceLROnPlateau(patience=25, factor=0.5, min_lr=1e-6, monitor='val_loss', verbose=0)\n",
"### 3) Save the best weights into file\n",
"checkpointer= keras.callbacks.ModelCheckpoint(filepath=\"Cui_cnn1.h5\", verbose=1, save_best_only=True)\n",
"\n",
"## Train the model\n",
"h1=model_cnn.fit(x_train_scaled_col, y_train, batch_size=BATCH, epochs=1500, \\\n",
" validation_data=(x_test_scaled_col, y_test), \\\n",
" callbacks=[plot_losses, checkpointer],verbose=1)\n",
"\n",
"tf.keras.backend.clear_session()"
]
},
{
"cell_type": "code",
"execution_count": 53,
"metadata": {
"ExecuteTime": {
"end_time": "2020-07-16T10:10:47.592066Z",
"start_time": "2020-07-16T10:10:47.296047Z"
}
},
"outputs": [
{
"data": {
"image/png": "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\n",
"text/plain": [
""
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
}
],
"source": [
"## If you used Livelossplot, you can skip this. Otherwise we can take a look at the training process by plotting the\n",
"## models history.\n",
"plt.plot(h1.history['loss'], label='loss')\n",
"plt.plot(h1.history['val_loss'], label='val_loss')\n",
"plt.yscale('log')\n",
"plt.ylabel('Loss')\n",
"plt.xlabel('Epochs')\n",
"plt.legend()\n",
"## In case you used ReduceLROnPlateau() you can plot the lr as well\n",
"# ax2 = plt.gca().twinx()\n",
"# ax2.plot(h1.history['lr'], color='r')\n",
"# ax2.set_ylabel('lr',color='r')\n",
"plt.show()"
]
},
{
"cell_type": "code",
"execution_count": 55,
"metadata": {
"ExecuteTime": {
"end_time": "2020-07-20T06:43:36.820735Z",
"start_time": "2020-07-20T06:43:36.726721Z"
}
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Train| RMSE: 1.083 \t SEP: 1.079\n",
"Test | RMSE: 1.272 \t SEP: 1.271\n"
]
}
],
"source": [
"## Load saved best parameters and compute prediction metrics\n",
"model_cnn.load_weights('Cui_cnn1.h5')\n",
"y_pred_train1 = model_cnn.predict(x_train_scaled_col)\n",
"y_pred_test1 = model_cnn.predict(x_test_scaled_col)\n",
"## Compute SEP and RMSE\n",
"sep_train1 = np.std(y_train - y_pred_train1)\n",
"sep_test1 = np.std(y_test - y_pred_test1)\n",
"rmse_train1 = np.sqrt(mean_squared_error(y_train,y_pred_train1))\n",
"rmse_test1 = np.sqrt(mean_squared_error(y_test,y_pred_test1))\n",
"## Print metrics\n",
"print('Train| RMSE: {:2.3f} \\t SEP: {:2.3f}'.format(rmse_train1, sep_train1))\n",
"print('Test | RMSE: {:2.3f} \\t SEP: {:2.3f}'.format(rmse_test1, sep_test1))"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"The error metrics obtained are more than double of those obtained with the PLS model. This means that we should try other type of scaling of the X data..."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"#### Apply CNN to data standardized on rows and columns"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"It is useful to run again the cell where we define the model before running a new experiment. This ensures that the weights are properly initialized! So... go up, re-run that cell and comeback here."
]
},
{
"cell_type": "code",
"execution_count": 58,
"metadata": {
"ExecuteTime": {
"end_time": "2020-07-16T10:26:11.441407Z",
"start_time": "2020-07-16T10:18:07.794382Z"
},
"scrolled": false
},
"outputs": [
{
"data": {
"image/png": "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\n",
"text/plain": [
""
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Loss\n",
"\ttraining \t (min: 0.256, max: 89.637, cur: 0.257)\n",
"\tvalidation \t (min: 0.353, max: 54.058, cur: 0.425)\n",
"Mean Squared Error\n",
"\ttraining \t (min: 0.154, max: 89.438, cur: 0.156)\n",
"\tvalidation \t (min: 0.229, max: 53.856, cur: 0.324)\n",
"\n",
"Epoch 01500: val_loss did not improve from 0.35279\n",
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"2/2 [==============================] - 0s 153ms/step - loss: 0.2574 - mse: 0.1563 - val_loss: 0.4246 - val_mse: 0.3235\n"
]
}
],
"source": [
"## plot losses in real time during training process\n",
"plot_losses = PlotLossesKerasTF()\n",
"\n",
"##### TRAIN MODEL WITH ADAM OPTIMIZER #########\n",
"BATCH=256 \n",
"LR=0.01*BATCH/256.\n",
"\n",
"print('Adam learning rate = {}'.format(LR))\n",
"model_cnn.compile(optimizer=tf.keras.optimizers.Adam(learning_rate=LR), loss='mse', metrics=['mse']) \n",
"\n",
"## Save the best weights into file\n",
"checkpointer= keras.callbacks.ModelCheckpoint(filepath=\"Cui_cnn2.h5\", verbose=1, save_best_only=True)\n",
"\n",
"## Train the model\n",
"h2=model_cnn.fit(x_train_scaled_rowcol, y_train, batch_size=BATCH, epochs=1500, \\\n",
" validation_data=(x_test_scaled_rowcol, y_test), \\\n",
" callbacks=[plot_losses, checkpointer],verbose=1)"
]
},
{
"cell_type": "code",
"execution_count": 59,
"metadata": {
"ExecuteTime": {
"end_time": "2020-07-16T10:27:25.920144Z",
"start_time": "2020-07-16T10:27:25.617711Z"
}
},
"outputs": [
{
"data": {
"image/png": "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\n",
"text/plain": [
""
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
}
],
"source": [
"## If you used Livelossplot, you can skip this. Otherwise we can take a look at the training process by plotting the\n",
"## models history.\n",
"plt.plot(h2.history['loss'], label='loss')\n",
"plt.plot(h2.history['val_loss'], label='val_loss')\n",
"plt.yscale('log')\n",
"plt.ylabel('Loss')\n",
"plt.xlabel('Epochs')\n",
"plt.legend()\n",
"plt.show()"
]
},
{
"cell_type": "code",
"execution_count": 54,
"metadata": {
"ExecuteTime": {
"end_time": "2020-07-20T06:43:15.637435Z",
"start_time": "2020-07-20T06:43:15.548455Z"
}
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Train| RMSE: 0.458 \t SEP: 0.438\n",
"Test | RMSE: 0.484 \t SEP: 0.481\n"
]
}
],
"source": [
"## Load best weights and compute error metrics\n",
"model_cnn.load_weights('Cui_cnn2.h5')\n",
"y_pred_train2 = model_cnn.predict(x_train_scaled_rowcol)\n",
"y_pred_test2 = model_cnn.predict(x_test_scaled_rowcol)\n",
"## Compute SEP and RMSE\n",
"sep_train2 = np.std(y_train - y_pred_train2)\n",
"sep_test2 = np.std(y_test - y_pred_test2)\n",
"rmse_train2 = np.sqrt(mean_squared_error(y_train,y_pred_train2))\n",
"rmse_test2 = np.sqrt(mean_squared_error(y_test,y_pred_test2))\n",
"## Print metrics\n",
"print('Train| RMSE: {:2.3f} \\t SEP: {:2.3f}'.format(rmse_train2, sep_train2))\n",
"print('Test | RMSE: {:2.3f} \\t SEP: {:2.3f}'.format(rmse_test2, sep_test2))"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"We get better error metrics (lower) with the row + col standardized data. This approaches the values obtained with the PLS model. The model we used here uses default hyperparameters that in principle can be optimized even further (e.g. using RandomGridSearch, Bayesian Opt, TPE, etc. (check other notebooks in this series). Also we can try to extend the training for a longer number of epochs and see if the RMSE goes downs even more.\n",
"\n",
"**Personal comments** \n",
"The results we obtain with this implementation of the CNN are not as good as the author's present in their paper (although by a very small margin). From this experiment we can see that the method they propose is valid but raises the question of 'results reproducibility'. One of the points of the article is that this type of CNN can get results on par or even better than PLS (considered nowadays the standard in Chemometrics). The authors show evidence for that claim using 3 different datasets. In their defense, on our experiment here, we only tested one of the datasets (arguably the worst case scenario where we have a small number of samples) and the CNN implementation is not the same as theirs (different packages, API, etc). That being said, the claim that this CNN works better than the PLS cannot be drawn from these computations. This is, in my opinion, one of the main difficulties of these modeling methods, i.e., its hard to exactly reproduce results of models (CNN or other type of NN architectures) that involve some degree of randomness in their computation. The PLS on the other hand is much more straightforward. When the results between models is as small as the ones we see here, a simple analysis like we did is not enough to say which model is better. In the paper, the authors provide further evidence (e.g. prediction noise analysis, etc.) that favor CNNs. Some possible solutions for improving results reproducibility are for researchers to give more details about their implementation of the models (provide info about package versions, random(seed), etc, etc. in an separate appendix) at the cost of making the publication longer, or alternatively just simply share their codes (which sometimes is not possible due to PI issues)."
]
},
{
"cell_type": "markdown",
"metadata": {
"heading_collapsed": true
},
"source": [
"### Implementation of the L-BFGS optimizer for model training"
]
},
{
"cell_type": "markdown",
"metadata": {
"hidden": true
},
"source": [
"Just for the sake of comparison I tried to implement the model optimization using the L-BFGS optimizer the authors mentioned in the paper. Since tf.keras does not implement this optimizer directly in the API we have to craft some workaround. After some online searching I found this [tutorial blog](https://pychao.com/2019/11/02/optimize-tensorflow-keras-models-with-l-bfgs-from-tensorflow-probability/) by Pi-Yueh Chuang. The code below is ported from the corresponding GitHub.\n",
"\n",
"The method consists in crafting a utility function that carries the right parameters into the tfp.optimizer.lbfgs_minimize() optimizer that is included in the tensorflow_probability package. I modified the original script in order for it to compute the metrics on the test data as well (like before, the training process does not use test data). I confess that I don't fully understand the inner works of the script but it seems to work!!"
]
},
{
"cell_type": "code",
"execution_count": 73,
"metadata": {
"ExecuteTime": {
"end_time": "2020-07-16T11:24:28.122626Z",
"start_time": "2020-07-16T11:24:28.103217Z"
},
"hidden": true
},
"outputs": [],
"source": [
"import tensorflow_probability as tfp\n",
"\n",
"def function_factory(model, loss, train_x, train_y, test_x, test_y):\n",
" \"\"\"A factory to create a function required by tfp.optimizer.lbfgs_minimize.\n",
" Args:\n",
" model [in]: an instance of `tf.keras.Model` or its subclasses.\n",
" loss [in]: a function with signature loss_value = loss(pred_y, true_y).\n",
" train_x [in]: the input part of training data.\n",
" train_y [in]: the output part of training data.\n",
" ## DP addition:\n",
" test_x [in]: the input part of test data (just for metric purposes)\n",
" test_y [in]: the output part of test data (just for metric purposes)\n",
" Returns:\n",
" A function that has a signature of:\n",
" loss_value, gradients = f(model_parameters).\n",
" \"\"\"\n",
"\n",
" # obtain the shapes of all trainable parameters in the model\n",
" shapes = tf.shape_n(model.trainable_variables)\n",
" n_tensors = len(shapes)\n",
"\n",
" # we'll use tf.dynamic_stitch and tf.dynamic_partition later, so we need to\n",
" # prepare required information first\n",
" count = 0\n",
" idx = [] # stitch indices\n",
" part = [] # partition indices\n",
"\n",
" for i, shape in enumerate(shapes):\n",
" n = np.product(shape)\n",
" idx.append(tf.reshape(tf.range(count, count+n, dtype=tf.int32), shape))\n",
" part.extend([i]*n)\n",
" count += n\n",
"\n",
" part = tf.constant(part)\n",
"\n",
" @tf.function\n",
" def assign_new_model_parameters(params_1d):\n",
" \"\"\"A function updating the model's parameters with a 1D tf.Tensor.\n",
" Args:\n",
" params_1d [in]: a 1D tf.Tensor representing the model's trainable parameters.\n",
" \"\"\"\n",
"\n",
" params = tf.dynamic_partition(params_1d, part, n_tensors)\n",
" for i, (shape, param) in enumerate(zip(shapes, params)):\n",
" model.trainable_variables[i].assign(tf.reshape(param, shape))\n",
"\n",
" # now create a function that will be returned by this factory\n",
" @tf.function\n",
" def f(params_1d):\n",
" \"\"\"A function that can be used by tfp.optimizer.lbfgs_minimize.\n",
" This function is created by function_factory.\n",
" Args:\n",
" params_1d [in]: a 1D tf.Tensor.\n",
" Returns:\n",
" A scalar loss and the gradients w.r.t. the `params_1d`.\n",
" \"\"\"\n",
"\n",
" # use GradientTape so that we can calculate the gradient of loss w.r.t. parameters\n",
" with tf.GradientTape() as tape:\n",
" # update the parameters in the model\n",
" assign_new_model_parameters(params_1d)\n",
" # calculate the loss\n",
" loss_value = loss(model(train_x, training=True), train_y)\n",
" \n",
"\n",
" # calculate gradients and convert to 1D tf.Tensor\n",
" grads = tape.gradient(loss_value, model.trainable_variables)\n",
" grads = tf.dynamic_stitch(idx, grads)\n",
"\n",
" # print out iteration & loss\n",
" f.iter.assign_add(1)\n",
" test_loss=loss(model(test_x), test_y)\n",
" tf.print(\"Iter:\", f.iter, \"loss:\", loss_value, \"test loss:\", test_loss)\n",
"\n",
" # store loss and test_loss values so we can retrieve later\n",
" tf.py_function(f.history.append, inp=[[loss_value, test_loss]], Tout=[])\n",
"\n",
" return loss_value, grads\n",
"\n",
" # store these information as members so we can use them outside the scope\n",
" f.iter = tf.Variable(0)\n",
" f.idx = idx\n",
" f.part = part\n",
" f.shapes = shapes\n",
" f.assign_new_model_parameters = assign_new_model_parameters\n",
" f.history = []\n",
"\n",
" return f"
]
},
{
"cell_type": "markdown",
"metadata": {
"hidden": true
},
"source": [
"It is useful to run again the cell where we define the model before running a new experiment. This ensures that the weights are properly initialized! So... go up, re-run that cell and comeback here."
]
},
{
"cell_type": "code",
"execution_count": 74,
"metadata": {
"ExecuteTime": {
"end_time": "2020-07-16T11:24:53.889195Z",
"start_time": "2020-07-16T11:24:33.584769Z"
},
"hidden": true,
"scrolled": true
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Iter: 1 loss: 93.8331451 test loss: 92.6674957\n",
"Iter: 2 loss: 3.21259821e+13 test loss: 4.43607e+13\n",
"Iter: 3 loss: 93.8331451 test loss: 92.6674957\n",
"Iter: 4 loss: 5.98535127e+10 test loss: 8.31036129e+10\n",
"Iter: 5 loss: 93.8331451 test loss: 92.6674881\n",
"Iter: 6 loss: 181676960 test loss: 254870032\n",
"Iter: 7 loss: 93.8320618 test loss: 92.6661148\n",
"Iter: 8 loss: 1126717.25 test loss: 1591306.5\n",
"Iter: 9 loss: 93.7817612 test loss: 92.602\n",
"Iter: 10 loss: 15302.2324 test loss: 21282.6953\n",
"Iter: 11 loss: 92.675827 test loss: 91.181427\n",
"Iter: 12 loss: 385.263519 test loss: 492.785736\n",
"Iter: 13 loss: 81.434166 test loss: 75.976265\n",
"Iter: 14 loss: 38.4782181 test loss: 23.1704197\n",
"Iter: 15 loss: 40.6020393 test loss: 23.8449745\n",
"Iter: 16 loss: 4805.7124 test loss: 7246.19336\n",
"Iter: 17 loss: 39.8924026 test loss: 23.2348175\n",
"Iter: 18 loss: 41.8566628 test loss: 44.9027328\n",
"Iter: 19 loss: 33.7247429 test loss: 19.8295383\n",
"Iter: 20 loss: 27.4857807 test loss: 48.8987045\n",
"Iter: 21 loss: 24.6595345 test loss: 24.5543327\n",
"Iter: 22 loss: 19.5282974 test loss: 32.9685745\n",
"Iter: 23 loss: 19.2960091 test loss: 30.1464558\n",
"Iter: 24 loss: 16.2156315 test loss: 21.3848629\n",
"Iter: 25 loss: 22.7144279 test loss: 22.186388\n",
"Iter: 26 loss: 16.0191708 test loss: 17.1008873\n",
"Iter: 27 loss: 12.7870293 test loss: 15.9347334\n",
"Iter: 28 loss: 14.8996468 test loss: 21.1701698\n",
"Iter: 29 loss: 10.226697 test loss: 15.6686182\n",
"Iter: 30 loss: 6.69337225 test loss: 10.1015244\n",
"Iter: 31 loss: 72.1409378 test loss: 45.0425873\n",
"Iter: 32 loss: 6.552773 test loss: 9.65249062\n",
"Iter: 33 loss: 4.28206635 test loss: 5.24486732\n",
"Iter: 34 loss: 8.51044178 test loss: 7.40180779\n",
"Iter: 35 loss: 4.0603323 test loss: 3.7492218\n",
"Iter: 36 loss: 3.28519893 test loss: 3.21383119\n",
"Iter: 37 loss: 6.5196352 test loss: 7.14282\n",
"Iter: 38 loss: 3.13800406 test loss: 3.21103787\n",
"Iter: 39 loss: 2.81339312 test loss: 3.36845255\n",
"Iter: 40 loss: 3.48453569 test loss: 8.10411\n",
"Iter: 41 loss: 2.65744925 test loss: 3.98648429\n",
"Iter: 42 loss: 2.39764237 test loss: 4.16604042\n",
"Iter: 43 loss: 3.29496503 test loss: 4.68803596\n",
"Iter: 44 loss: 2.39484477 test loss: 4.32439899\n",
"Iter: 45 loss: 2.25829911 test loss: 4.0947752\n",
"Iter: 46 loss: 2.56358457 test loss: 3.90336609\n",
"Iter: 47 loss: 2.20098877 test loss: 3.9302814\n",
"Iter: 48 loss: 2.08942366 test loss: 3.34455419\n",
"Iter: 49 loss: 1.98096716 test loss: 2.5984695\n",
"Iter: 50 loss: 1.96858764 test loss: 2.59270334\n",
"Iter: 51 loss: 1.86936581 test loss: 2.6287334\n",
"Iter: 52 loss: 2.03309345 test loss: 3.24284649\n",
"Iter: 53 loss: 1.83270454 test loss: 2.7617321\n",
"Iter: 54 loss: 1.79786909 test loss: 2.67947555\n",
"Iter: 55 loss: 1.99681675 test loss: 2.78299403\n",
"Iter: 56 loss: 1.79367924 test loss: 2.65500283\n",
"Iter: 57 loss: 1.78079188 test loss: 2.66381049\n",
"Iter: 58 loss: 1.77117014 test loss: 2.71731567\n",
"Iter: 59 loss: 1.76797915 test loss: 2.70009804\n",
"Iter: 60 loss: 1.74849606 test loss: 2.68033457\n",
"Iter: 61 loss: 1.70292294 test loss: 2.62311816\n",
"Iter: 62 loss: 2.76308894 test loss: 3.25176644\n",
"Iter: 63 loss: 1.70131159 test loss: 2.61795783\n",
"Iter: 64 loss: 1.63727009 test loss: 2.53031516\n",
"Iter: 65 loss: 1.60921252 test loss: 2.40966392\n",
"Iter: 66 loss: 1.57012892 test loss: 2.40382791\n",
"Iter: 67 loss: 1.50892174 test loss: 2.32446504\n",
"Iter: 68 loss: 1.78544545 test loss: 2.54397702\n",
"Iter: 69 loss: 1.49933624 test loss: 2.3006959\n",
"Iter: 70 loss: 1.47631133 test loss: 2.30054736\n",
"Iter: 71 loss: 1.46017826 test loss: 2.37513065\n",
"Iter: 72 loss: 1.45234597 test loss: 2.33513474\n",
"Iter: 73 loss: 1.41823494 test loss: 2.24918699\n",
"Iter: 74 loss: 1.42244422 test loss: 1.95358944\n",
"Iter: 75 loss: 1.38549984 test loss: 2.11374378\n",
"Iter: 76 loss: 1.34879899 test loss: 2.08464408\n",
"Iter: 77 loss: 1.41715598 test loss: 2.08551145\n",
"Iter: 78 loss: 1.33334756 test loss: 2.06961513\n",
"Iter: 79 loss: 1.31057382 test loss: 2.09137893\n",
"Iter: 80 loss: 1.33436394 test loss: 2.22618699\n",
"Iter: 81 loss: 1.29714787 test loss: 2.13703513\n",
"Iter: 82 loss: 1.27350485 test loss: 2.41424298\n",
"Iter: 83 loss: 1.27288771 test loss: 2.37584376\n",
"Iter: 84 loss: 1.29644775 test loss: 2.07728624\n",
"Iter: 85 loss: 1.26271176 test loss: 2.29469204\n",
"Iter: 86 loss: 1.2515862 test loss: 2.23737907\n",
"Iter: 87 loss: 1.45731211 test loss: 1.90592897\n",
"Iter: 88 loss: 1.25101447 test loss: 2.22774243\n",
"Iter: 89 loss: 1.23489845 test loss: 2.23608303\n",
"Iter: 90 loss: 1.25135779 test loss: 2.37917233\n",
"Iter: 91 loss: 1.22356057 test loss: 2.26257348\n",
"Iter: 92 loss: 1.18075931 test loss: 2.21681881\n",
"Iter: 93 loss: 1.41127574 test loss: 2.06820965\n",
"Iter: 94 loss: 1.17282557 test loss: 2.18430138\n",
"Iter: 95 loss: 1.13602555 test loss: 2.0242703\n",
"Iter: 96 loss: 1.25477016 test loss: 1.53594828\n",
"Iter: 97 loss: 1.12133634 test loss: 1.91927767\n",
"Iter: 98 loss: 1.07260871 test loss: 1.65742326\n",
"Iter: 99 loss: 1.07019091 test loss: 1.71658862\n",
"Iter: 100 loss: 1.2874248 test loss: 2.01089597\n",
"Iter: 101 loss: 1.05172372 test loss: 1.71683073\n",
"Iter: 102 loss: 1.03481686 test loss: 1.60412705\n",
"Iter: 103 loss: 1.39727342 test loss: 1.84833598\n",
"Iter: 104 loss: 1.03468943 test loss: 1.59920812\n",
"Iter: 105 loss: 1.01206124 test loss: 1.47402751\n",
"Iter: 106 loss: 1.04784095 test loss: 1.10469723\n",
"Iter: 107 loss: 0.999708951 test loss: 1.33304358\n",
"Iter: 108 loss: 0.972761571 test loss: 1.23861873\n",
"Iter: 109 loss: 1.3201735 test loss: 1.32450283\n",
"Iter: 110 loss: 0.971292138 test loss: 1.22401607\n",
"Iter: 111 loss: 0.922978401 test loss: 1.27433646\n",
"Iter: 112 loss: 1.27505219 test loss: 2.01837301\n",
"Iter: 113 loss: 0.919084907 test loss: 1.31325471\n",
"Iter: 114 loss: 0.899356604 test loss: 1.23260224\n",
"Iter: 115 loss: 1.06564021 test loss: 1.17539072\n",
"Iter: 116 loss: 0.896744907 test loss: 1.20823193\n",
"Iter: 117 loss: 0.87214458 test loss: 1.2085923\n",
"Iter: 118 loss: 0.894710183 test loss: 1.38545585\n",
"Iter: 119 loss: 0.857124627 test loss: 1.23198414\n",
"Iter: 120 loss: 0.922030807 test loss: 1.27211678\n",
"Iter: 121 loss: 0.843807578 test loss: 1.17341042\n",
"Iter: 122 loss: 0.802464843 test loss: 1.10062277\n",
"Iter: 123 loss: 0.799426436 test loss: 1.09575737\n",
"Iter: 124 loss: 0.750646651 test loss: 1.16375136\n",
"Iter: 125 loss: 3.65977097 test loss: 6.3120513\n",
"Iter: 126 loss: 0.750578225 test loss: 1.16545343\n",
"Iter: 127 loss: 0.719988346 test loss: 1.22853041\n",
"Iter: 128 loss: 0.715022504 test loss: 1.19814837\n",
"Iter: 129 loss: 0.681797206 test loss: 1.30789363\n",
"Iter: 130 loss: 0.680363 test loss: 1.28159308\n",
"Iter: 131 loss: 0.650022089 test loss: 1.34260929\n",
"Iter: 132 loss: 0.941815615 test loss: 2.7343049\n",
"Iter: 133 loss: 0.649958551 test loss: 1.3537184\n",
"Iter: 134 loss: 0.636671126 test loss: 1.20825636\n",
"Iter: 135 loss: 0.636653602 test loss: 1.2124809\n",
"Iter: 136 loss: 0.627342045 test loss: 1.17763507\n",
"Iter: 137 loss: 0.706966877 test loss: 1.30261\n",
"Iter: 138 loss: 0.627248108 test loss: 1.17481506\n",
"Iter: 139 loss: 0.61136 test loss: 1.30188704\n",
"Iter: 140 loss: 0.690670907 test loss: 2.36813951\n",
"Iter: 141 loss: 0.606505334 test loss: 1.40202892\n",
"Iter: 142 loss: 0.596535921 test loss: 1.46005881\n",
"Iter: 143 loss: 0.647137523 test loss: 1.83735967\n",
"Iter: 144 loss: 0.593934596 test loss: 1.5043329\n",
"Iter: 145 loss: 0.578133881 test loss: 1.34851968\n",
"Iter: 146 loss: 0.577456236 test loss: 1.38178968\n",
"Iter: 147 loss: 0.564790487 test loss: 1.43904698\n",
"Iter: 148 loss: 0.564678192 test loss: 1.43188465\n",
"Iter: 149 loss: 0.557574332 test loss: 1.29598629\n",
"Iter: 150 loss: 0.69476819 test loss: 0.74783659\n",
"Iter: 151 loss: 0.557471156 test loss: 1.28463197\n",
"Iter: 152 loss: 0.550506055 test loss: 1.28260565\n",
"Iter: 153 loss: 0.556453 test loss: 1.28520668\n",
"Iter: 154 loss: 0.546034157 test loss: 1.28131557\n",
"Iter: 155 loss: 0.539222538 test loss: 1.30475235\n",
"Iter: 156 loss: 0.529209077 test loss: 1.46145391\n",
"Iter: 157 loss: 0.529111922 test loss: 1.44418728\n",
"Iter: 158 loss: 0.513408363 test loss: 1.38287461\n",
"Iter: 159 loss: 0.698388219 test loss: 1.42754698\n",
"Iter: 160 loss: 0.513033628 test loss: 1.37563121\n",
"Iter: 161 loss: 0.507754803 test loss: 1.43139267\n",
"Iter: 162 loss: 0.555881739 test loss: 1.68884885\n",
"Iter: 163 loss: 0.507561564 test loss: 1.44549954\n",
"Iter: 164 loss: 0.504144132 test loss: 1.43444109\n",
"Iter: 165 loss: 0.542894304 test loss: 1.33617389\n",
"Iter: 166 loss: 0.503910184 test loss: 1.43108034\n",
"Iter: 167 loss: 0.501909614 test loss: 1.5015043\n",
"Iter: 168 loss: 0.507537484 test loss: 1.82842529\n",
"Iter: 169 loss: 0.501206219 test loss: 1.5731324\n",
"Iter: 170 loss: 0.504269063 test loss: 1.5375272\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Iter: 171 loss: 0.497225702 test loss: 1.56587577\n",
"Iter: 172 loss: 0.494278461 test loss: 1.66946137\n",
"Iter: 173 loss: 0.509341478 test loss: 2.20641327\n",
"Iter: 174 loss: 0.4937585 test loss: 1.7372359\n",
"Iter: 175 loss: 0.491932958 test loss: 1.74287009\n",
"Iter: 176 loss: 0.496476829 test loss: 1.73851073\n",
"Iter: 177 loss: 0.491133481 test loss: 1.74593496\n",
"Iter: 178 loss: 0.487799078 test loss: 1.79185057\n",
"Iter: 179 loss: 0.481849402 test loss: 1.9968276\n",
"Iter: 180 loss: 0.481845349 test loss: 1.99106765\n",
"Iter: 181 loss: 0.477854252 test loss: 1.89713919\n",
"Iter: 182 loss: 0.477539212 test loss: 1.91650558\n",
"Iter: 183 loss: 0.473561138 test loss: 1.95968413\n",
"Iter: 184 loss: 0.504108548 test loss: 2.15139794\n",
"Iter: 185 loss: 0.473266035 test loss: 1.9755671\n",
"Iter: 186 loss: 0.471224934 test loss: 1.97346163\n",
"Iter: 187 loss: 0.472804785 test loss: 1.92775655\n",
"Iter: 188 loss: 0.469723105 test loss: 1.96449757\n",
"Iter: 189 loss: 0.465464056 test loss: 1.96495831\n",
"Iter: 190 loss: 0.460234642 test loss: 1.96322048\n",
"Iter: 191 loss: 0.459809184 test loss: 1.964077\n",
"Iter: 192 loss: 0.459947199 test loss: 2.18532133\n",
"Iter: 193 loss: 0.45174709 test loss: 2.05766\n",
"Iter: 194 loss: 0.431583405 test loss: 1.68342876\n",
"Iter: 195 loss: 0.475677222 test loss: 0.690426588\n",
"Iter: 196 loss: 0.426028222 test loss: 1.28386855\n",
"Iter: 197 loss: 0.415602088 test loss: 1.58826244\n",
"Iter: 198 loss: 0.414554268 test loss: 1.50737476\n",
"Iter: 199 loss: 0.420673907 test loss: 1.32302606\n",
"Iter: 200 loss: 0.409823328 test loss: 1.44835126\n",
"Iter: 201 loss: 0.405927241 test loss: 1.56683636\n",
"Iter: 202 loss: 0.473747075 test loss: 2.48519278\n",
"Iter: 203 loss: 0.405871212 test loss: 1.57959974\n",
"Iter: 204 loss: 0.401212096 test loss: 1.52732158\n",
"Iter: 205 loss: 0.410523832 test loss: 1.2966696\n",
"Iter: 206 loss: 0.40047273 test loss: 1.46095526\n",
"Iter: 207 loss: 0.398841441 test loss: 1.38275242\n",
"Iter: 208 loss: 0.418342322 test loss: 1.12669086\n",
"Iter: 209 loss: 0.398816496 test loss: 1.37263107\n",
"Iter: 210 loss: 0.398203373 test loss: 1.4155786\n",
"Iter: 211 loss: 0.398202181 test loss: 1.4137404\n",
"Iter: 212 loss: 0.397445202 test loss: 1.47251523\n",
"Iter: 213 loss: 0.399251252 test loss: 1.74785089\n",
"Iter: 214 loss: 0.397138357 test loss: 1.53629243\n",
"Iter: 215 loss: 0.399821132 test loss: 1.66474235\n",
"Iter: 216 loss: 0.394748747 test loss: 1.58705807\n",
"Iter: 217 loss: 0.390158057 test loss: 1.59063745\n",
"Iter: 218 loss: 0.446692407 test loss: 1.37440825\n",
"Iter: 219 loss: 0.38975 test loss: 1.58835649\n",
"Iter: 220 loss: 0.384367853 test loss: 1.50827229\n",
"Iter: 221 loss: 0.431244731 test loss: 1.23468888\n",
"Iter: 222 loss: 0.383979827 test loss: 1.48425782\n",
"Iter: 223 loss: 0.382300168 test loss: 1.46787798\n",
"Iter: 224 loss: 0.388050228 test loss: 1.36254334\n",
"Iter: 225 loss: 0.381787956 test loss: 1.45190454\n",
"Iter: 226 loss: 0.379660398 test loss: 1.49585199\n",
"Iter: 227 loss: 0.381265104 test loss: 1.64659059\n",
"Iter: 228 loss: 0.378505945 test loss: 1.56287444\n",
"Iter: 229 loss: 0.375116199 test loss: 1.46821737\n",
"Iter: 230 loss: 0.376612425 test loss: 1.13021231\n",
"Iter: 231 loss: 0.372682661 test loss: 1.31545568\n",
"Iter: 232 loss: 0.369187087 test loss: 1.34251094\n",
"Iter: 233 loss: 0.383422971 test loss: 1.42376196\n",
"Iter: 234 loss: 0.368223727 test loss: 1.36022007\n",
"Iter: 235 loss: 0.365748972 test loss: 1.59298027\n",
"Iter: 236 loss: 0.363449514 test loss: 1.49301374\n",
"Iter: 237 loss: 0.361560494 test loss: 1.4912709\n",
"Iter: 238 loss: 0.379759908 test loss: 1.51688981\n",
"Iter: 239 loss: 0.361526191 test loss: 1.49131322\n",
"Iter: 240 loss: 0.360534936 test loss: 1.46293354\n",
"Iter: 241 loss: 0.361075908 test loss: 1.3543601\n",
"Iter: 242 loss: 0.359858751 test loss: 1.41656661\n",
"Iter: 243 loss: 0.358830303 test loss: 1.41981781\n",
"Iter: 244 loss: 0.357903391 test loss: 1.43620384\n",
"Iter: 245 loss: 0.357654899 test loss: 1.43051529\n",
"Iter: 246 loss: 0.356062233 test loss: 1.37876916\n",
"Iter: 247 loss: 0.382490516 test loss: 1.19338202\n",
"Iter: 248 loss: 0.356062263 test loss: 1.37803721\n",
"Iter: 249 loss: 0.354942203 test loss: 1.36531675\n",
"Iter: 250 loss: 0.356979668 test loss: 1.30121553\n",
"Iter: 251 loss: 0.354482651 test loss: 1.34813857\n",
"Iter: 252 loss: 0.35433659 test loss: 1.36967683\n",
"Iter: 253 loss: 0.353558213 test loss: 1.36060584\n",
"Iter: 254 loss: 0.35228771 test loss: 1.38303852\n",
"Iter: 255 loss: 0.352210313 test loss: 1.37898254\n",
"Iter: 256 loss: 0.351281583 test loss: 1.47074354\n",
"Iter: 257 loss: 0.35877949 test loss: 1.86627984\n",
"Iter: 258 loss: 0.351228386 test loss: 1.5006752\n",
"Iter: 259 loss: 0.350913078 test loss: 1.49834\n",
"Iter: 260 loss: 0.351231843 test loss: 1.49301481\n",
"Iter: 261 loss: 0.35073936 test loss: 1.49564433\n",
"Iter: 262 loss: 0.349939585 test loss: 1.45529962\n",
"Iter: 263 loss: 0.349939197 test loss: 1.45612121\n",
"Iter: 264 loss: 0.349257827 test loss: 1.47006881\n",
"Iter: 265 loss: 0.349718422 test loss: 1.53158867\n",
"Iter: 266 loss: 0.348833174 test loss: 1.49422598\n",
"Iter: 267 loss: 0.347910315 test loss: 1.46653736\n",
"Iter: 268 loss: 0.346970648 test loss: 1.35561347\n",
"Iter: 269 loss: 0.346781969 test loss: 1.38821018\n",
"Iter: 270 loss: 0.346322238 test loss: 1.35615993\n",
"Iter: 271 loss: 0.349424362 test loss: 1.23362756\n",
"Iter: 272 loss: 0.346271366 test loss: 1.34220719\n",
"Iter: 273 loss: 0.346086264 test loss: 1.34488356\n",
"Iter: 274 loss: 0.345704108 test loss: 1.35644841\n",
"Iter: 275 loss: 0.351252288 test loss: 1.43515956\n",
"Iter: 276 loss: 0.34569186 test loss: 1.35918295\n",
"Iter: 277 loss: 0.345018387 test loss: 1.33835447\n",
"Iter: 278 loss: 0.344647735 test loss: 1.34597576\n",
"Iter: 279 loss: 0.342616767 test loss: 1.27735007\n",
"Iter: 280 loss: 0.343993723 test loss: 1.01918566\n",
"Iter: 281 loss: 0.341360688 test loss: 1.16735172\n",
"Iter: 282 loss: 0.340536624 test loss: 1.21159685\n",
"Iter: 283 loss: 0.348690838 test loss: 1.40067\n",
"Iter: 284 loss: 0.340510368 test loss: 1.22114801\n",
"Iter: 285 loss: 0.34041813 test loss: 1.22305667\n",
"Iter: 286 loss: 0.340329021 test loss: 1.2292825\n",
"Iter: 287 loss: 0.340307593 test loss: 1.22761703\n",
"Iter: 288 loss: 0.339884192 test loss: 1.21363294\n",
"Iter: 289 loss: 0.33984521 test loss: 1.16178811\n",
"Iter: 290 loss: 0.339536101 test loss: 1.18598723\n",
"Iter: 291 loss: 0.340407044 test loss: 1.17907012\n",
"Iter: 292 loss: 0.339034677 test loss: 1.18277788\n",
"Iter: 293 loss: 0.338883 test loss: 1.19184518\n",
"Iter: 294 loss: 0.339204669 test loss: 1.22597647\n",
"Iter: 295 loss: 0.338822693 test loss: 1.20185685\n",
"Iter: 296 loss: 0.338563144 test loss: 1.11384535\n",
"Iter: 297 loss: 0.338219494 test loss: 1.15007544\n",
"Iter: 298 loss: 0.337619275 test loss: 1.16911912\n",
"Iter: 299 loss: 0.33711198 test loss: 1.24122369\n",
"Iter: 300 loss: 0.336925924 test loss: 1.21649325\n",
"Iter: 301 loss: 0.335723907 test loss: 1.20210397\n",
"Iter: 302 loss: 0.343518138 test loss: 1.12810552\n",
"Iter: 303 loss: 0.335553527 test loss: 1.19447708\n",
"Iter: 304 loss: 0.334300578 test loss: 1.21577585\n",
"Iter: 305 loss: 0.33993122 test loss: 1.31220222\n",
"Iter: 306 loss: 0.334104538 test loss: 1.23113763\n",
"Iter: 307 loss: 0.333238244 test loss: 1.22338772\n",
"Iter: 308 loss: 0.332646906 test loss: 1.2020551\n",
"Iter: 309 loss: 0.332299173 test loss: 1.2086134\n",
"Iter: 310 loss: 0.330788881 test loss: 1.08592248\n",
"Iter: 311 loss: 0.33067733 test loss: 1.11180723\n",
"Iter: 312 loss: 0.328952402 test loss: 1.11871266\n",
"Iter: 313 loss: 0.340634942 test loss: 1.04327166\n",
"Iter: 314 loss: 0.328864634 test loss: 1.11866856\n",
"Iter: 315 loss: 0.32801792 test loss: 1.10918915\n",
"Iter: 316 loss: 0.331508934 test loss: 1.1447314\n",
"Iter: 317 loss: 0.327918082 test loss: 1.10716081\n",
"Iter: 318 loss: 0.327349633 test loss: 1.24710417\n",
"Iter: 319 loss: 0.326680958 test loss: 1.19010055\n",
"Iter: 320 loss: 0.324696183 test loss: 1.18727648\n",
"Iter: 321 loss: 0.333232194 test loss: 1.13277125\n",
"Iter: 322 loss: 0.324317127 test loss: 1.18233597\n",
"Iter: 323 loss: 0.322243303 test loss: 1.2774359\n",
"Iter: 324 loss: 0.322204113 test loss: 1.263991\n",
"Iter: 325 loss: 0.320235431 test loss: 1.23172653\n",
"Iter: 326 loss: 0.351060659 test loss: 1.07689011\n",
"Iter: 327 loss: 0.320229501 test loss: 1.22992909\n",
"Iter: 328 loss: 0.319889933 test loss: 1.1936065\n",
"Iter: 329 loss: 0.321935 test loss: 1.05771852\n",
"Iter: 330 loss: 0.319842875 test loss: 1.17523432\n",
"Iter: 331 loss: 0.319558114 test loss: 1.18268466\n",
"Iter: 332 loss: 0.319045365 test loss: 1.21423817\n",
"Iter: 333 loss: 0.329461128 test loss: 1.41241586\n",
"Iter: 334 loss: 0.319044799 test loss: 1.21571422\n",
"Iter: 335 loss: 0.318514794 test loss: 1.22001505\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Iter: 336 loss: 0.32046774 test loss: 1.22688735\n",
"Iter: 337 loss: 0.318375915 test loss: 1.22284913\n",
"Iter: 338 loss: 0.318141341 test loss: 1.22907877\n",
"Iter: 339 loss: 0.318488955 test loss: 1.24958277\n",
"Iter: 340 loss: 0.318043441 test loss: 1.23697078\n",
"Iter: 341 loss: 0.317716509 test loss: 1.22195\n",
"Iter: 342 loss: 0.317071229 test loss: 1.16336906\n",
"Iter: 343 loss: 0.330243 test loss: 0.908215344\n",
"Iter: 344 loss: 0.317065865 test loss: 1.1579802\n",
"Iter: 345 loss: 0.316503614 test loss: 1.04823446\n",
"Iter: 346 loss: 0.316333383 test loss: 1.08992052\n",
"Iter: 347 loss: 0.315664262 test loss: 1.06755\n",
"Iter: 348 loss: 0.316603661 test loss: 0.981705129\n",
"Iter: 349 loss: 0.315316439 test loss: 1.038499\n",
"Iter: 350 loss: 0.314569384 test loss: 1.03078604\n",
"Iter: 351 loss: 0.313796431 test loss: 0.995620191\n",
"Iter: 352 loss: 0.31365484 test loss: 1.00638056\n",
"Iter: 353 loss: 0.312325418 test loss: 1.00741422\n",
"Iter: 354 loss: 0.315688163 test loss: 1.02404475\n",
"Iter: 355 loss: 0.311882228 test loss: 1.00973308\n",
"Iter: 356 loss: 0.310596585 test loss: 0.923007846\n",
"Iter: 357 loss: 0.370350122 test loss: 0.55466342\n",
"Iter: 358 loss: 0.310596257 test loss: 0.922383904\n",
"Iter: 359 loss: 0.3097803 test loss: 0.906656325\n",
"Iter: 360 loss: 0.311849743 test loss: 0.824603796\n",
"Iter: 361 loss: 0.309541881 test loss: 0.888798416\n",
"Iter: 362 loss: 0.310016394 test loss: 0.98504293\n",
"Iter: 363 loss: 0.309162945 test loss: 0.927339494\n",
"Iter: 364 loss: 0.308645934 test loss: 0.894730449\n",
"Iter: 365 loss: 0.315177351 test loss: 0.743919909\n",
"Iter: 366 loss: 0.308630109 test loss: 0.889407456\n",
"Iter: 367 loss: 0.307959735 test loss: 0.869333327\n",
"Iter: 368 loss: 0.312318116 test loss: 0.746492445\n",
"Iter: 369 loss: 0.307766289 test loss: 0.858948588\n",
"Iter: 370 loss: 0.306531191 test loss: 0.821767569\n",
"Iter: 371 loss: 0.306466311 test loss: 0.844724\n",
"Iter: 372 loss: 0.306391418 test loss: 0.860924\n",
"Iter: 373 loss: 0.305102825 test loss: 0.85666275\n",
"Iter: 374 loss: 0.394247919 test loss: 1.27684963\n",
"Iter: 375 loss: 0.304701626 test loss: 0.871658266\n",
"Iter: 376 loss: 0.302611619 test loss: 0.862175822\n",
"Iter: 377 loss: 0.322251111 test loss: 0.803146\n",
"Iter: 378 loss: 0.302507073 test loss: 0.859366834\n",
"Iter: 379 loss: 0.301495463 test loss: 0.863470137\n",
"Iter: 380 loss: 0.301474482 test loss: 0.862544894\n",
"Iter: 381 loss: 0.300746 test loss: 0.796356082\n",
"Iter: 382 loss: 0.300745666 test loss: 0.798170567\n",
"Iter: 383 loss: 0.299814016 test loss: 0.838993549\n",
"Iter: 384 loss: 0.309383154 test loss: 1.00392485\n",
"Iter: 385 loss: 0.299785942 test loss: 0.846922815\n",
"Iter: 386 loss: 0.29801175 test loss: 0.841316879\n",
"Iter: 387 loss: 0.303297102 test loss: 0.845159\n",
"Iter: 388 loss: 0.297515571 test loss: 0.838411\n",
"Iter: 389 loss: 0.296980649 test loss: 0.833912\n",
"Iter: 390 loss: 0.295947164 test loss: 0.813900709\n",
"Iter: 391 loss: 0.328226805 test loss: 0.664879143\n",
"Iter: 392 loss: 0.295943469 test loss: 0.813062549\n",
"Iter: 393 loss: 0.294265151 test loss: 0.783598065\n",
"Iter: 394 loss: 0.29319635 test loss: 0.681847155\n",
"Iter: 395 loss: 0.292520314 test loss: 0.717699707\n",
"Iter: 396 loss: 0.29105559 test loss: 0.6880427\n",
"Iter: 397 loss: 0.306455761 test loss: 0.558878779\n",
"Iter: 398 loss: 0.290989071 test loss: 0.681196749\n",
"Iter: 399 loss: 0.289973676 test loss: 0.688324511\n",
"Iter: 400 loss: 0.292298764 test loss: 0.725289524\n",
"Iter: 401 loss: 0.289592624 test loss: 0.69723016\n",
"Iter: 402 loss: 0.288350403 test loss: 0.676607072\n",
"Iter: 403 loss: 0.289102435 test loss: 0.611367702\n",
"Iter: 404 loss: 0.287558913 test loss: 0.645861924\n",
"Iter: 405 loss: 0.287751973 test loss: 0.535085797\n",
"Iter: 406 loss: 0.286379546 test loss: 0.586594343\n",
"Iter: 407 loss: 0.285215706 test loss: 0.591051757\n",
"Iter: 408 loss: 0.287153274 test loss: 0.612898648\n",
"Iter: 409 loss: 0.284675539 test loss: 0.597056627\n",
"Iter: 410 loss: 0.28270337 test loss: 0.601569116\n",
"Iter: 411 loss: 0.297773182 test loss: 0.660860837\n",
"Iter: 412 loss: 0.282408774 test loss: 0.604520261\n",
"Iter: 413 loss: 0.279983252 test loss: 0.616852522\n",
"Iter: 414 loss: 0.277378917 test loss: 0.66390568\n",
"Iter: 415 loss: 0.277039319 test loss: 0.653191864\n",
"Iter: 416 loss: 0.274561167 test loss: 0.62888068\n",
"Iter: 417 loss: 0.274548233 test loss: 0.630379736\n",
"Iter: 418 loss: 0.272896379 test loss: 0.586807966\n",
"Iter: 419 loss: 0.314313293 test loss: 0.46561563\n",
"Iter: 420 loss: 0.272885799 test loss: 0.584854066\n",
"Iter: 421 loss: 0.291350722 test loss: 0.787159622\n",
"Iter: 422 loss: 0.272363454 test loss: 0.607035041\n",
"Iter: 423 loss: 0.271979094 test loss: 0.602814\n",
"Iter: 424 loss: 0.274882555 test loss: 0.585820496\n",
"Iter: 425 loss: 0.271956176 test loss: 0.601390362\n",
"Iter: 426 loss: 0.271858513 test loss: 0.599044263\n",
"Iter: 427 loss: 0.272303879 test loss: 0.592443526\n",
"Iter: 428 loss: 0.271840155 test loss: 0.597628295\n",
"Iter: 429 loss: 0.27172792 test loss: 0.595114231\n",
"Iter: 430 loss: 0.271563321 test loss: 0.585172355\n",
"Iter: 431 loss: 0.271558762 test loss: 0.586614132\n",
"Iter: 432 loss: 0.271130562 test loss: 0.579758584\n",
"Iter: 433 loss: 0.271543831 test loss: 0.550624\n",
"Iter: 434 loss: 0.27088958 test loss: 0.568844378\n",
"Iter: 435 loss: 0.270296127 test loss: 0.544526398\n",
"Iter: 436 loss: 0.271682888 test loss: 0.467395395\n",
"Iter: 437 loss: 0.270070523 test loss: 0.520478725\n",
"Iter: 438 loss: 0.269122899 test loss: 0.532173216\n",
"Iter: 439 loss: 0.290017396 test loss: 0.675673902\n",
"Iter: 440 loss: 0.269119352 test loss: 0.532833099\n",
"Iter: 441 loss: 0.268697 test loss: 0.530027151\n",
"Iter: 442 loss: 0.268790931 test loss: 0.520446718\n",
"Iter: 443 loss: 0.26837489 test loss: 0.525316417\n",
"Iter: 444 loss: 0.267855 test loss: 0.529674\n",
"Iter: 445 loss: 0.268384 test loss: 0.551667035\n",
"Iter: 446 loss: 0.267567843 test loss: 0.537189066\n",
"Iter: 447 loss: 0.427762538 test loss: 0.615002096\n",
"Iter: 448 loss: 0.267406821 test loss: 0.526822686\n",
"Iter: 449 loss: 0.267181456 test loss: 0.538587391\n",
"Iter: 450 loss: 0.267406911 test loss: 0.587688863\n",
"Iter: 451 loss: 0.267056286 test loss: 0.556753516\n",
"Iter: 452 loss: 0.266761899 test loss: 0.554437399\n",
"Iter: 453 loss: 0.266377658 test loss: 0.545820117\n",
"Iter: 454 loss: 0.266352803 test loss: 0.547453761\n",
"Iter: 455 loss: 0.266638815 test loss: 0.568100631\n",
"Iter: 456 loss: 0.266183853 test loss: 0.55485332\n",
"Iter: 457 loss: 0.265970051 test loss: 0.552553356\n",
"Iter: 458 loss: 0.266475767 test loss: 0.546054602\n",
"Iter: 459 loss: 0.265889317 test loss: 0.550385296\n",
"Iter: 460 loss: 0.265737683 test loss: 0.54723388\n",
"Iter: 461 loss: 0.266855985 test loss: 0.536658704\n",
"Iter: 462 loss: 0.265725851 test loss: 0.546096861\n",
"Iter: 463 loss: 0.265599728 test loss: 0.542624295\n",
"Iter: 464 loss: 0.266221583 test loss: 0.528819799\n",
"Iter: 465 loss: 0.265578598 test loss: 0.540363073\n",
"Iter: 466 loss: 0.265406132 test loss: 0.543692946\n",
"Iter: 467 loss: 0.265362829 test loss: 0.557797611\n",
"Iter: 468 loss: 0.265256763 test loss: 0.551248968\n",
"Iter: 469 loss: 0.264836 test loss: 0.533820093\n",
"Iter: 470 loss: 0.264827788 test loss: 0.466775686\n",
"Iter: 471 loss: 0.264480948 test loss: 0.501269042\n",
"Iter: 472 loss: 0.26418376 test loss: 0.502913237\n",
"Iter: 473 loss: 0.265928149 test loss: 0.517273605\n",
"Iter: 474 loss: 0.264145076 test loss: 0.504059911\n",
"Iter: 475 loss: 0.264003754 test loss: 0.503215313\n",
"Iter: 476 loss: 0.264333159 test loss: 0.500696421\n",
"Iter: 477 loss: 0.263952553 test loss: 0.502396703\n",
"Iter: 478 loss: 0.26378566 test loss: 0.517827272\n",
"Iter: 479 loss: 0.263736188 test loss: 0.512587845\n",
"Iter: 480 loss: 0.263692647 test loss: 0.513675153\n",
"Iter: 481 loss: 0.263765067 test loss: 0.519013703\n",
"Iter: 482 loss: 0.263673753 test loss: 0.515195966\n",
"Iter: 483 loss: 0.263462305 test loss: 0.487808853\n",
"Iter: 484 loss: 0.263431907 test loss: 0.494708061\n",
"Iter: 485 loss: 0.263114423 test loss: 0.490219474\n",
"Iter: 486 loss: 0.263570756 test loss: 0.479640722\n",
"Iter: 487 loss: 0.262979269 test loss: 0.485237688\n",
"Iter: 488 loss: 0.262849092 test loss: 0.490940452\n",
"Iter: 489 loss: 0.263037264 test loss: 0.513816297\n",
"Iter: 490 loss: 0.262789726 test loss: 0.498605341\n",
"Iter: 491 loss: 0.262668133 test loss: 0.493838459\n",
"Iter: 492 loss: 0.262987077 test loss: 0.477181196\n",
"Iter: 493 loss: 0.262626618 test loss: 0.489251\n",
"Iter: 494 loss: 0.262447208 test loss: 0.488534302\n",
"Iter: 495 loss: 0.261867046 test loss: 0.485926032\n",
"Iter: 496 loss: 0.26205942 test loss: 0.478472084\n",
"Iter: 497 loss: 0.261339277 test loss: 0.481255114\n",
"Iter: 498 loss: 0.260053962 test loss: 0.488174647\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Iter: 499 loss: 0.268257886 test loss: 0.555103779\n",
"Iter: 500 loss: 0.259845048 test loss: 0.492889553\n",
"Iter: 501 loss: 0.259259731 test loss: 0.472179919\n",
"Iter: 502 loss: 0.267518461 test loss: 0.386591315\n",
"Iter: 503 loss: 0.259230882 test loss: 0.46852386\n",
"Iter: 504 loss: 0.258954674 test loss: 0.464262575\n",
"Iter: 505 loss: 0.263092816 test loss: 0.456531405\n",
"Iter: 506 loss: 0.258953393 test loss: 0.463998228\n",
"Iter: 507 loss: 0.258851409 test loss: 0.466861576\n",
"Iter: 508 loss: 0.260130316 test loss: 0.482112437\n",
"Iter: 509 loss: 0.258850247 test loss: 0.467161804\n",
"Iter: 510 loss: 0.258736044 test loss: 0.465383381\n",
"Iter: 511 loss: 0.258567 test loss: 0.458570331\n",
"Iter: 512 loss: 0.258562237 test loss: 0.459475398\n",
"Iter: 513 loss: 0.258355677 test loss: 0.453324497\n",
"Iter: 514 loss: 0.259568393 test loss: 0.432581276\n",
"Iter: 515 loss: 0.258329034 test loss: 0.450284362\n",
"Iter: 516 loss: 0.258248746 test loss: 0.446551859\n",
"Iter: 517 loss: 0.258649796 test loss: 0.433416\n",
"Iter: 518 loss: 0.258234739 test loss: 0.444338351\n",
"Iter: 519 loss: 0.258174509 test loss: 0.44646588\n",
"Iter: 520 loss: 0.258588821 test loss: 0.455744952\n",
"Iter: 521 loss: 0.25816834 test loss: 0.447384\n",
"Iter: 522 loss: 0.25806284 test loss: 0.44621855\n",
"Iter: 523 loss: 0.258001298 test loss: 0.442493409\n",
"Iter: 524 loss: 0.257956654 test loss: 0.443772644\n",
"Iter: 525 loss: 0.257719964 test loss: 0.449552178\n",
"Iter: 526 loss: 0.257752597 test loss: 0.472480506\n",
"Iter: 527 loss: 0.257535756 test loss: 0.46032843\n",
"Iter: 528 loss: 0.257288575 test loss: 0.466440529\n",
"Iter: 529 loss: 0.25882411 test loss: 0.49231407\n",
"Iter: 530 loss: 0.257256448 test loss: 0.469405174\n",
"Iter: 531 loss: 0.257172078 test loss: 0.468729228\n",
"Iter: 532 loss: 0.256980389 test loss: 0.466076583\n",
"Iter: 533 loss: 0.25973627 test loss: 0.454282165\n",
"Iter: 534 loss: 0.256969064 test loss: 0.465339601\n",
"Iter: 535 loss: 0.25704 test loss: 0.489804834\n",
"Iter: 536 loss: 0.256871045 test loss: 0.475683898\n",
"Iter: 537 loss: 0.256615728 test loss: 0.463952392\n",
"Iter: 538 loss: 0.258858949 test loss: 0.41859\n",
"Iter: 539 loss: 0.256597281 test loss: 0.460342556\n",
"Iter: 540 loss: 0.256531835 test loss: 0.458361357\n",
"Iter: 541 loss: 0.256585836 test loss: 0.450562418\n",
"Iter: 542 loss: 0.25649181 test loss: 0.455318332\n",
"Iter: 543 loss: 0.256418824 test loss: 0.45679149\n",
"Iter: 544 loss: 0.256234765 test loss: 0.462861508\n",
"Iter: 545 loss: 0.257946 test loss: 0.497876108\n",
"Iter: 546 loss: 0.256207556 test loss: 0.466303855\n",
"Iter: 547 loss: 0.256018311 test loss: 0.475577533\n",
"Iter: 548 loss: 0.25672248 test loss: 0.513740301\n",
"Iter: 549 loss: 0.255972415 test loss: 0.483069301\n",
"Iter: 550 loss: 0.255911767 test loss: 0.486584723\n",
"Iter: 551 loss: 0.256496429 test loss: 0.502635658\n",
"Iter: 552 loss: 0.255908817 test loss: 0.487472266\n",
"Iter: 553 loss: 0.255809397 test loss: 0.486558676\n",
"Iter: 554 loss: 0.255661368 test loss: 0.483588338\n",
"Iter: 555 loss: 0.255658925 test loss: 0.483878344\n",
"Iter: 556 loss: 0.255384207 test loss: 0.490399957\n",
"Iter: 557 loss: 0.255229264 test loss: 0.487060517\n",
"Iter: 558 loss: 0.25469926 test loss: 0.478601456\n",
"Iter: 559 loss: 0.257493556 test loss: 0.457953423\n",
"Iter: 560 loss: 0.254610956 test loss: 0.474236429\n",
"Iter: 561 loss: 0.254488796 test loss: 0.464408934\n",
"Iter: 562 loss: 0.255414099 test loss: 0.426716357\n",
"Iter: 563 loss: 0.254481256 test loss: 0.461072654\n",
"Iter: 564 loss: 0.254318357 test loss: 0.45234412\n",
"Iter: 565 loss: 0.254282415 test loss: 0.419691354\n",
"Iter: 566 loss: 0.254175663 test loss: 0.434336096\n",
"Iter: 567 loss: 0.254023492 test loss: 0.433948487\n",
"Iter: 568 loss: 0.255411625 test loss: 0.434120238\n",
"Iter: 569 loss: 0.254017979 test loss: 0.43387872\n",
"Iter: 570 loss: 0.253801823 test loss: 0.429311424\n",
"Iter: 571 loss: 0.253827572 test loss: 0.411715299\n",
"Iter: 572 loss: 0.253635705 test loss: 0.42073378\n",
"Iter: 573 loss: 0.252999187 test loss: 0.419805795\n",
"Iter: 574 loss: 0.251792789 test loss: 0.416118026\n",
"Iter: 575 loss: 0.265043885 test loss: 0.39318794\n",
"Iter: 576 loss: 0.251791745 test loss: 0.415273875\n",
"Iter: 577 loss: 0.250686228 test loss: 0.379147291\n",
"Iter: 578 loss: 0.269749969 test loss: 0.272780448\n",
"Iter: 579 loss: 0.250674129 test loss: 0.37644276\n",
"Iter: 580 loss: 0.250547 test loss: 0.374012619\n",
"Iter: 581 loss: 0.251935959 test loss: 0.364785641\n",
"Iter: 582 loss: 0.250545084 test loss: 0.373634696\n",
"Iter: 583 loss: 0.250519276 test loss: 0.371608466\n",
"Iter: 584 loss: 0.250651389 test loss: 0.363725036\n",
"Iter: 585 loss: 0.250515312 test loss: 0.370433182\n",
"Iter: 586 loss: 0.250351399 test loss: 0.370312423\n",
"Iter: 587 loss: 0.249833629 test loss: 0.370411128\n",
"Iter: 588 loss: 0.250435382 test loss: 0.384056717\n",
"Iter: 589 loss: 0.24943018 test loss: 0.372955501\n",
"Iter: 590 loss: 0.255981088 test loss: 0.389964789\n",
"Iter: 591 loss: 0.249072284 test loss: 0.366894513\n",
"Iter: 592 loss: 0.248736203 test loss: 0.373412\n",
"Iter: 593 loss: 0.248720169 test loss: 0.372043282\n",
"Iter: 594 loss: 0.24859643 test loss: 0.366337776\n",
"Iter: 595 loss: 0.249536633 test loss: 0.345930547\n",
"Iter: 596 loss: 0.248584703 test loss: 0.364244819\n",
"Iter: 597 loss: 0.248667404 test loss: 0.36405009\n",
"Iter: 598 loss: 0.248559564 test loss: 0.364169866\n",
"Iter: 599 loss: 0.248520151 test loss: 0.363259941\n",
"Iter: 600 loss: 0.248540565 test loss: 0.360687256\n",
"Iter: 601 loss: 0.248493016 test loss: 0.361969918\n",
"Iter: 602 loss: 0.248439208 test loss: 0.351979256\n",
"Iter: 603 loss: 0.249239594 test loss: 0.315788627\n",
"Iter: 604 loss: 0.248439029 test loss: 0.351537704\n",
"Iter: 605 loss: 0.248421967 test loss: 0.351670504\n",
"Iter: 606 loss: 0.248407051 test loss: 0.352318168\n",
"Iter: 607 loss: 0.248403266 test loss: 0.352091819\n",
"Iter: 608 loss: 0.248350427 test loss: 0.349674672\n",
"Iter: 609 loss: 0.24818337 test loss: 0.340173453\n",
"Iter: 610 loss: 0.248464972 test loss: 0.297125\n",
"Iter: 611 loss: 0.248059526 test loss: 0.324544311\n",
"Iter: 612 loss: 0.264347136 test loss: 0.325499892\n",
"Iter: 613 loss: 0.248002201 test loss: 0.323862612\n",
"Iter: 614 loss: 0.247905821 test loss: 0.321160525\n",
"Iter: 615 loss: 0.248641685 test loss: 0.313878596\n",
"Iter: 616 loss: 0.247898623 test loss: 0.32027778\n",
"Iter: 617 loss: 0.247864932 test loss: 0.320960402\n",
"Iter: 618 loss: 0.24777928 test loss: 0.323839813\n",
"Iter: 619 loss: 0.248578787 test loss: 0.34164995\n",
"Iter: 620 loss: 0.247766092 test loss: 0.325538218\n",
"Iter: 621 loss: 0.247549325 test loss: 0.321047068\n",
"Iter: 622 loss: 0.247985661 test loss: 0.302414268\n",
"Iter: 623 loss: 0.247466311 test loss: 0.315621346\n",
"Iter: 624 loss: 0.247182697 test loss: 0.316256195\n",
"Iter: 625 loss: 0.247952119 test loss: 0.322821289\n",
"Iter: 626 loss: 0.247085243 test loss: 0.317254066\n",
"Iter: 627 loss: 0.24672094 test loss: 0.324142128\n",
"Iter: 628 loss: 0.247891307 test loss: 0.357705295\n",
"Iter: 629 loss: 0.246612236 test loss: 0.330786288\n",
"Iter: 630 loss: 0.246358663 test loss: 0.321796209\n",
"Iter: 631 loss: 0.249384552 test loss: 0.291516304\n",
"Iter: 632 loss: 0.24635084 test loss: 0.320315629\n",
"Iter: 633 loss: 0.24626632 test loss: 0.326986372\n",
"Iter: 634 loss: 0.246462509 test loss: 0.355260909\n",
"Iter: 635 loss: 0.246235013 test loss: 0.334397167\n",
"Iter: 636 loss: 0.246100023 test loss: 0.333959311\n",
"Iter: 637 loss: 0.246204734 test loss: 0.332575977\n",
"Iter: 638 loss: 0.246012941 test loss: 0.333355218\n",
"Iter: 639 loss: 0.24578464 test loss: 0.32611376\n",
"Iter: 640 loss: 0.246837333 test loss: 0.302418113\n",
"Iter: 641 loss: 0.245746911 test loss: 0.321602404\n",
"Iter: 642 loss: 0.245599434 test loss: 0.313342154\n",
"Iter: 643 loss: 0.246898249 test loss: 0.288064599\n",
"Iter: 644 loss: 0.24559243 test loss: 0.311200321\n",
"Iter: 645 loss: 0.24559164 test loss: 0.313055784\n",
"Iter: 646 loss: 0.245550543 test loss: 0.31209296\n",
"Iter: 647 loss: 0.24551253 test loss: 0.310997069\n",
"Iter: 648 loss: 0.245448053 test loss: 0.307031184\n",
"Iter: 649 loss: 0.245448083 test loss: 0.307123929\n",
"Iter: 650 loss: 0.245573685 test loss: 0.307948142\n",
"Iter: 651 loss: 0.245385259 test loss: 0.307175815\n",
"Iter: 652 loss: 0.245335698 test loss: 0.306795955\n",
"Iter: 653 loss: 0.245407835 test loss: 0.305814594\n",
"Iter: 654 loss: 0.245312139 test loss: 0.30636543\n",
"Iter: 655 loss: 0.245224029 test loss: 0.304703653\n",
"Iter: 656 loss: 0.245001853 test loss: 0.298410207\n",
"Iter: 657 loss: 0.247020453 test loss: 0.274371684\n",
"Iter: 658 loss: 0.244969532 test loss: 0.295039445\n",
"Iter: 659 loss: 0.244766563 test loss: 0.29575941\n",
"Iter: 660 loss: 0.244942278 test loss: 0.298458427\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Iter: 661 loss: 0.244650468 test loss: 0.296867579\n",
"Iter: 662 loss: 0.24498637 test loss: 0.265942812\n",
"Iter: 663 loss: 0.244529873 test loss: 0.285937101\n",
"Iter: 664 loss: 0.244489953 test loss: 0.282799572\n",
"Iter: 665 loss: 0.244598359 test loss: 0.270800173\n",
"Iter: 666 loss: 0.244477406 test loss: 0.279743522\n",
"Iter: 667 loss: 0.244359538 test loss: 0.281208575\n",
"Iter: 668 loss: 0.24553749 test loss: 0.290195912\n",
"Iter: 669 loss: 0.244354799 test loss: 0.281566113\n",
"Iter: 670 loss: 0.244166672 test loss: 0.281065136\n",
"Iter: 671 loss: 0.244058087 test loss: 0.280418843\n",
"Iter: 672 loss: 0.243979558 test loss: 0.280404061\n",
"Iter: 673 loss: 0.245552748 test loss: 0.266955823\n",
"Iter: 674 loss: 0.24389933 test loss: 0.277415574\n",
"Iter: 675 loss: 0.24384819 test loss: 0.273958206\n",
"Iter: 676 loss: 0.244600907 test loss: 0.262434751\n",
"Iter: 677 loss: 0.243847907 test loss: 0.273909241\n",
"Iter: 678 loss: 0.243815377 test loss: 0.273878843\n",
"Iter: 679 loss: 0.243824854 test loss: 0.274070084\n",
"Iter: 680 loss: 0.243791953 test loss: 0.273902\n",
"Iter: 681 loss: 0.243670762 test loss: 0.275152147\n",
"Iter: 682 loss: 0.243595064 test loss: 0.280503333\n",
"Iter: 683 loss: 0.243548289 test loss: 0.278428644\n",
"Iter: 684 loss: 0.243334055 test loss: 0.286530286\n",
"Iter: 685 loss: 0.243924707 test loss: 0.322435617\n",
"Iter: 686 loss: 0.243268609 test loss: 0.294974536\n",
"Iter: 687 loss: 0.243151784 test loss: 0.291579455\n",
"Iter: 688 loss: 0.24287042 test loss: 0.278805673\n",
"Iter: 689 loss: 0.245828032 test loss: 0.23257716\n",
"Iter: 690 loss: 0.242838115 test loss: 0.273241132\n",
"Iter: 691 loss: 0.242453143 test loss: 0.272512108\n",
"Iter: 692 loss: 0.242207468 test loss: 0.275172651\n",
"Iter: 693 loss: 0.242058516 test loss: 0.273148477\n",
"Iter: 694 loss: 0.242266014 test loss: 0.282521933\n",
"Iter: 695 loss: 0.242017701 test loss: 0.275708228\n",
"Iter: 696 loss: 0.241999015 test loss: 0.272437513\n",
"Iter: 697 loss: 0.24199903 test loss: 0.272534728\n",
"Iter: 698 loss: 0.241947517 test loss: 0.271111816\n",
"Iter: 699 loss: 0.241953641 test loss: 0.265706211\n",
"Iter: 700 loss: 0.241908193 test loss: 0.268447399\n",
"Iter: 701 loss: 0.241841629 test loss: 0.266748548\n",
"Iter: 702 loss: 0.241846308 test loss: 0.26067546\n",
"Iter: 703 loss: 0.241790131 test loss: 0.263636053\n",
"Iter: 704 loss: 0.241707027 test loss: 0.260745317\n",
"Iter: 705 loss: 0.242455572 test loss: 0.25208506\n",
"Iter: 706 loss: 0.241703555 test loss: 0.260067254\n",
"Iter: 707 loss: 0.241619438 test loss: 0.259756684\n",
"Iter: 708 loss: 0.241812021 test loss: 0.259301096\n",
"Iter: 709 loss: 0.241587818 test loss: 0.259507656\n",
"Iter: 710 loss: 0.2414276 test loss: 0.256048322\n",
"Iter: 711 loss: 0.241376296 test loss: 0.244006142\n",
"Iter: 712 loss: 0.24128367 test loss: 0.248965517\n",
"Iter: 713 loss: 0.241138414 test loss: 0.247609794\n",
"Iter: 714 loss: 0.24131164 test loss: 0.246757284\n",
"Iter: 715 loss: 0.24106048 test loss: 0.246470198\n",
"Iter: 716 loss: 0.240947589 test loss: 0.249948502\n",
"Iter: 717 loss: 0.240822688 test loss: 0.264972925\n",
"Iter: 718 loss: 0.240803361 test loss: 0.260795623\n",
"Iter: 719 loss: 0.240677416 test loss: 0.257261\n",
"Iter: 720 loss: 0.240665108 test loss: 0.24574469\n",
"Iter: 721 loss: 0.240566924 test loss: 0.25115788\n",
"Iter: 722 loss: 0.240416765 test loss: 0.246460661\n",
"Iter: 723 loss: 0.240415171 test loss: 0.246770635\n",
"Iter: 724 loss: 0.24034743 test loss: 0.246588618\n",
"Iter: 725 loss: 0.240703285 test loss: 0.248873025\n",
"Iter: 726 loss: 0.240337357 test loss: 0.246622086\n",
"Iter: 727 loss: 0.240285501 test loss: 0.244871631\n",
"Iter: 728 loss: 0.240225896 test loss: 0.238384992\n",
"Iter: 729 loss: 0.240219131 test loss: 0.239886388\n",
"Iter: 730 loss: 0.240084901 test loss: 0.239954472\n",
"Iter: 731 loss: 0.239881426 test loss: 0.240955442\n",
"Iter: 732 loss: 0.239877388 test loss: 0.240764335\n",
"Iter: 733 loss: 0.23922281 test loss: 0.244657695\n",
"Iter: 734 loss: 0.24000746 test loss: 0.26838854\n",
"Iter: 735 loss: 0.238880992 test loss: 0.251741052\n",
"Iter: 736 loss: 0.240084127 test loss: 0.27651158\n",
"Iter: 737 loss: 0.238729432 test loss: 0.25702107\n",
"Iter: 738 loss: 0.238304898 test loss: 0.258420736\n",
"Iter: 739 loss: 0.242032319 test loss: 0.284529895\n",
"Iter: 740 loss: 0.238287792 test loss: 0.259147793\n",
"Iter: 741 loss: 0.238058731 test loss: 0.25684157\n",
"Iter: 742 loss: 0.239774555 test loss: 0.26121214\n",
"Iter: 743 loss: 0.238039657 test loss: 0.256374687\n",
"Iter: 744 loss: 0.23786588 test loss: 0.252257407\n",
"Iter: 745 loss: 0.238396898 test loss: 0.241431519\n",
"Iter: 746 loss: 0.237819389 test loss: 0.248925209\n",
"Iter: 747 loss: 0.237706631 test loss: 0.249193192\n",
"Iter: 748 loss: 0.238123462 test loss: 0.251433194\n",
"Iter: 749 loss: 0.237677693 test loss: 0.249493793\n",
"Iter: 750 loss: 0.237572983 test loss: 0.251719713\n",
"Iter: 751 loss: 0.23769711 test loss: 0.261158109\n",
"Iter: 752 loss: 0.237519547 test loss: 0.255108476\n",
"Iter: 753 loss: 0.237389505 test loss: 0.255945981\n",
"Iter: 754 loss: 0.238581091 test loss: 0.262748957\n",
"Iter: 755 loss: 0.237379983 test loss: 0.256266475\n",
"Iter: 756 loss: 0.237331569 test loss: 0.259609\n",
"Iter: 757 loss: 0.237540588 test loss: 0.275724918\n",
"Iter: 758 loss: 0.237321377 test loss: 0.262069166\n",
"Iter: 759 loss: 0.237275273 test loss: 0.263427615\n",
"Iter: 760 loss: 0.237233579 test loss: 0.269700199\n",
"Iter: 761 loss: 0.237223804 test loss: 0.267766327\n",
"Iter: 762 loss: 0.23742716 test loss: 0.288664132\n",
"Iter: 763 loss: 0.237172857 test loss: 0.274223566\n",
"Iter: 764 loss: 0.23708123 test loss: 0.275883257\n",
"Iter: 765 loss: 0.237296805 test loss: 0.283493459\n",
"Iter: 766 loss: 0.23704645 test loss: 0.277742326\n",
"Iter: 767 loss: 0.236978561 test loss: 0.276205748\n",
"Iter: 768 loss: 0.236877844 test loss: 0.270466387\n",
"Iter: 769 loss: 0.236873224 test loss: 0.271494061\n",
"Iter: 770 loss: 0.237451643 test loss: 0.27570197\n",
"Iter: 771 loss: 0.236722395 test loss: 0.272203058\n",
"Iter: 772 loss: 0.236336544 test loss: 0.27176702\n",
"Iter: 773 loss: 0.235723883 test loss: 0.269040853\n",
"Iter: 774 loss: 0.235712767 test loss: 0.269449\n",
"Iter: 775 loss: 0.235772058 test loss: 0.28598\n",
"Iter: 776 loss: 0.235346749 test loss: 0.276154101\n",
"Iter: 777 loss: 0.234951794 test loss: 0.265900284\n",
"Iter: 778 loss: 0.234940156 test loss: 0.267262608\n",
"Iter: 779 loss: 0.256529629 test loss: 0.285589635\n",
"Iter: 780 loss: 0.234762073 test loss: 0.268096894\n",
"Iter: 781 loss: 0.234500751 test loss: 0.270148724\n",
"Iter: 782 loss: 0.234105721 test loss: 0.279014915\n",
"Iter: 783 loss: 0.23409766 test loss: 0.277859032\n",
"Iter: 784 loss: 0.233116403 test loss: 0.279787093\n",
"Iter: 785 loss: 0.247239634 test loss: 0.310525566\n",
"Iter: 786 loss: 0.233110383 test loss: 0.280016929\n",
"Iter: 787 loss: 0.232686475 test loss: 0.283295631\n",
"Iter: 788 loss: 0.236681327 test loss: 0.289031386\n",
"Iter: 789 loss: 0.23268117 test loss: 0.283752233\n",
"Iter: 790 loss: 0.232335404 test loss: 0.286610603\n",
"Iter: 791 loss: 0.233659834 test loss: 0.294935793\n",
"Iter: 792 loss: 0.232235789 test loss: 0.288584262\n",
"Iter: 793 loss: 0.231850624 test loss: 0.2858724\n",
"Iter: 794 loss: 0.231504753 test loss: 0.281290263\n",
"Iter: 795 loss: 0.231410965 test loss: 0.28177014\n",
"Iter: 796 loss: 0.259475291 test loss: 0.481473863\n",
"Iter: 797 loss: 0.230637848 test loss: 0.286289185\n",
"Iter: 798 loss: 0.229703218 test loss: 0.276027828\n",
"Iter: 799 loss: 0.241193101 test loss: 0.256152272\n",
"Iter: 800 loss: 0.229698941 test loss: 0.275279462\n",
"Iter: 801 loss: 0.22949332 test loss: 0.263190806\n",
"Iter: 802 loss: 0.229344159 test loss: 0.267506391\n",
"Iter: 803 loss: 0.229248017 test loss: 0.266282529\n",
"Iter: 804 loss: 0.229240745 test loss: 0.266522706\n",
"Iter: 805 loss: 0.22921963 test loss: 0.266275287\n",
"Iter: 806 loss: 0.229249209 test loss: 0.265344948\n",
"Iter: 807 loss: 0.2292092 test loss: 0.265950501\n",
"Iter: 808 loss: 0.229023799 test loss: 0.266931564\n",
"Iter: 809 loss: 0.228542671 test loss: 0.271576434\n",
"Iter: 810 loss: 0.232723564 test loss: 0.307831347\n",
"Iter: 811 loss: 0.228457198 test loss: 0.274781197\n",
"Iter: 812 loss: 0.228457 test loss: 0.284759641\n",
"Iter: 813 loss: 0.228313327 test loss: 0.279984772\n",
"Iter: 814 loss: 0.227778 test loss: 0.284495085\n",
"Iter: 815 loss: 0.227699384 test loss: 0.283247888\n",
"Iter: 816 loss: 0.227554411 test loss: 0.284707397\n",
"Iter: 817 loss: 0.227294624 test loss: 0.28405568\n",
"Iter: 818 loss: 0.227024779 test loss: 0.279262751\n",
"Iter: 819 loss: 0.228146031 test loss: 0.265069067\n",
"Iter: 820 loss: 0.226967692 test loss: 0.276142865\n",
"Iter: 821 loss: 0.226583943 test loss: 0.27637729\n",
"Iter: 822 loss: 0.226073518 test loss: 0.277858794\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Iter: 823 loss: 0.226043567 test loss: 0.277494937\n",
"Iter: 824 loss: 0.225294039 test loss: 0.285337538\n",
"Iter: 825 loss: 0.2344497 test loss: 0.331543505\n",
"Iter: 826 loss: 0.225266427 test loss: 0.286883563\n",
"Iter: 827 loss: 0.224934131 test loss: 0.287805885\n",
"Iter: 828 loss: 0.226904422 test loss: 0.28967762\n",
"Iter: 829 loss: 0.224884212 test loss: 0.288226366\n",
"Iter: 830 loss: 0.22469236 test loss: 0.289436728\n",
"Iter: 831 loss: 0.22549811 test loss: 0.296377748\n",
"Iter: 832 loss: 0.22464563 test loss: 0.29044047\n",
"Iter: 833 loss: 0.224446714 test loss: 0.291412741\n",
"Iter: 834 loss: 0.224576131 test loss: 0.297183365\n",
"Iter: 835 loss: 0.224319175 test loss: 0.29338184\n",
"Iter: 836 loss: 0.230328038 test loss: 0.403449893\n",
"Iter: 837 loss: 0.224149615 test loss: 0.301636696\n",
"Iter: 838 loss: 0.223710597 test loss: 0.298977256\n",
"Iter: 839 loss: 0.223391667 test loss: 0.289182514\n",
"Iter: 840 loss: 0.223231196 test loss: 0.29269731\n",
"Iter: 841 loss: 0.222842842 test loss: 0.28949675\n",
"Iter: 842 loss: 0.222713649 test loss: 0.280839473\n",
"Iter: 843 loss: 0.222446471 test loss: 0.284129143\n",
"Iter: 844 loss: 0.221824393 test loss: 0.269243449\n",
"Iter: 845 loss: 0.226202041 test loss: 0.222688407\n",
"Iter: 846 loss: 0.221763983 test loss: 0.263528407\n",
"Iter: 847 loss: 0.220575958 test loss: 0.261116505\n",
"Iter: 848 loss: 0.227101281 test loss: 0.240764365\n",
"Iter: 849 loss: 0.220243365 test loss: 0.259294569\n",
"Iter: 850 loss: 0.218579292 test loss: 0.264636815\n",
"Iter: 851 loss: 0.227888659 test loss: 0.320975333\n",
"Iter: 852 loss: 0.21819368 test loss: 0.268888801\n",
"Iter: 853 loss: 0.216580242 test loss: 0.267689\n",
"Iter: 854 loss: 0.248555824 test loss: 0.29738766\n",
"Iter: 855 loss: 0.216579378 test loss: 0.267703861\n",
"Iter: 856 loss: 0.216417864 test loss: 0.267862827\n",
"Iter: 857 loss: 0.216391027 test loss: 0.267791331\n",
"Iter: 858 loss: 0.216243967 test loss: 0.271261275\n",
"Iter: 859 loss: 0.217608973 test loss: 0.288834691\n",
"Iter: 860 loss: 0.216237783 test loss: 0.272180259\n",
"Iter: 861 loss: 0.216089606 test loss: 0.272038251\n",
"Iter: 862 loss: 0.216089517 test loss: 0.272039235\n",
"Iter: 863 loss: 0.215696931 test loss: 0.266506135\n",
"Iter: 864 loss: 0.216470197 test loss: 0.242954627\n",
"Iter: 865 loss: 0.215513334 test loss: 0.260267735\n",
"Iter: 866 loss: 0.216684371 test loss: 0.254683912\n",
"Iter: 867 loss: 0.215356112 test loss: 0.255584657\n",
"Iter: 868 loss: 0.214996666 test loss: 0.260840386\n",
"Iter: 869 loss: 0.214951187 test loss: 0.259447545\n",
"Iter: 870 loss: 0.214194536 test loss: 0.257447571\n",
"Iter: 871 loss: 0.221346885 test loss: 0.261965692\n",
"Iter: 872 loss: 0.214130908 test loss: 0.256977618\n",
"Iter: 873 loss: 0.21375984 test loss: 0.259076864\n",
"Iter: 874 loss: 0.217112571 test loss: 0.266347587\n",
"Iter: 875 loss: 0.213742658 test loss: 0.259608448\n",
"Iter: 876 loss: 0.213422194 test loss: 0.255308032\n",
"Iter: 877 loss: 0.21379362 test loss: 0.244689941\n",
"Iter: 878 loss: 0.213261455 test loss: 0.250306547\n",
"Iter: 879 loss: 0.213027865 test loss: 0.246186435\n",
"Iter: 880 loss: 0.21517171 test loss: 0.234259486\n",
"Iter: 881 loss: 0.213011771 test loss: 0.245081335\n",
"Iter: 882 loss: 0.212829873 test loss: 0.243472099\n",
"Iter: 883 loss: 0.212872192 test loss: 0.239544034\n",
"Iter: 884 loss: 0.212698147 test loss: 0.241130382\n",
"Iter: 885 loss: 0.212365448 test loss: 0.237910911\n",
"Iter: 886 loss: 0.212258458 test loss: 0.238658935\n",
"Iter: 887 loss: 0.21128428 test loss: 0.235543251\n",
"Iter: 888 loss: 0.21190764 test loss: 0.235579655\n",
"Iter: 889 loss: 0.210444823 test loss: 0.233568937\n",
"Iter: 890 loss: 0.223488107 test loss: 0.256937742\n",
"Iter: 891 loss: 0.20993489 test loss: 0.233751521\n",
"Iter: 892 loss: 0.209372744 test loss: 0.242947564\n",
"Iter: 893 loss: 0.208888277 test loss: 0.237178057\n",
"Iter: 894 loss: 0.20846355 test loss: 0.241736054\n",
"Iter: 895 loss: 0.21268715 test loss: 0.270150363\n",
"Iter: 896 loss: 0.208446205 test loss: 0.242882356\n",
"Iter: 897 loss: 0.208311319 test loss: 0.23626852\n",
"Iter: 898 loss: 0.208285987 test loss: 0.237978324\n",
"Iter: 899 loss: 0.209221691 test loss: 0.245108411\n",
"Iter: 900 loss: 0.208177924 test loss: 0.239517614\n",
"Iter: 901 loss: 0.208116889 test loss: 0.240801424\n",
"Iter: 902 loss: 0.208177209 test loss: 0.24665162\n",
"Iter: 903 loss: 0.208080828 test loss: 0.242883682\n",
"Iter: 904 loss: 0.208042294 test loss: 0.235112295\n",
"Iter: 905 loss: 0.207959428 test loss: 0.238615632\n",
"Iter: 906 loss: 0.208048582 test loss: 0.247939929\n",
"Iter: 907 loss: 0.207878053 test loss: 0.241917551\n",
"Iter: 908 loss: 0.207757249 test loss: 0.241947472\n",
"Iter: 909 loss: 0.207825407 test loss: 0.242475897\n",
"Iter: 910 loss: 0.207678661 test loss: 0.24208957\n",
"Iter: 911 loss: 0.207540751 test loss: 0.243304029\n",
"Iter: 912 loss: 0.207218528 test loss: 0.248436183\n",
"Iter: 913 loss: 0.211261079 test loss: 0.280619591\n",
"Iter: 914 loss: 0.20719181 test loss: 0.250346541\n",
"Iter: 915 loss: 0.207005695 test loss: 0.253017515\n",
"Iter: 916 loss: 0.207372352 test loss: 0.265203685\n",
"Iter: 917 loss: 0.206935897 test loss: 0.256454498\n",
"Iter: 918 loss: 0.206821531 test loss: 0.256592721\n",
"Iter: 919 loss: 0.206689671 test loss: 0.25750488\n",
"Iter: 920 loss: 0.206672773 test loss: 0.257214606\n",
"Iter: 921 loss: 0.206589714 test loss: 0.259825259\n",
"Iter: 922 loss: 0.20678091 test loss: 0.271107674\n",
"Iter: 923 loss: 0.206560954 test loss: 0.262750506\n",
"Iter: 924 loss: 0.206528321 test loss: 0.263949245\n",
"Iter: 925 loss: 0.206476048 test loss: 0.269229919\n",
"Iter: 926 loss: 0.206475392 test loss: 0.268694133\n",
"Iter: 927 loss: 0.206394047 test loss: 0.268133193\n",
"Iter: 928 loss: 0.206201524 test loss: 0.266080469\n",
"Iter: 929 loss: 0.208383173 test loss: 0.26017195\n",
"Iter: 930 loss: 0.206182629 test loss: 0.265299827\n",
"Iter: 931 loss: 0.20642741 test loss: 0.26902777\n",
"Iter: 932 loss: 0.206102908 test loss: 0.266342372\n",
"Iter: 933 loss: 0.205794394 test loss: 0.263548076\n",
"Iter: 934 loss: 0.205637664 test loss: 0.253807038\n",
"Iter: 935 loss: 0.205490917 test loss: 0.257539481\n",
"Iter: 936 loss: 0.204963297 test loss: 0.259361982\n",
"Iter: 937 loss: 0.204879656 test loss: 0.269975513\n",
"Iter: 938 loss: 0.204523593 test loss: 0.264367133\n",
"Iter: 939 loss: 0.204095036 test loss: 0.266270757\n",
"Iter: 940 loss: 0.20629397 test loss: 0.275734097\n",
"Iter: 941 loss: 0.203995511 test loss: 0.267476887\n",
"Iter: 942 loss: 0.203661025 test loss: 0.261562794\n",
"Iter: 943 loss: 0.205745563 test loss: 0.241191179\n",
"Iter: 944 loss: 0.203616098 test loss: 0.258788347\n",
"Iter: 945 loss: 0.203484148 test loss: 0.265147239\n",
"Iter: 946 loss: 0.203440383 test loss: 0.26296556\n",
"Iter: 947 loss: 0.203275323 test loss: 0.262190282\n",
"Iter: 948 loss: 0.20292744 test loss: 0.259646207\n",
"Iter: 949 loss: 0.208558708 test loss: 0.262192905\n",
"Iter: 950 loss: 0.202916503 test loss: 0.259240031\n",
"Iter: 951 loss: 0.202531219 test loss: 0.26039958\n",
"Iter: 952 loss: 0.204747856 test loss: 0.267384231\n",
"Iter: 953 loss: 0.202453211 test loss: 0.261095405\n",
"Iter: 954 loss: 0.202337921 test loss: 0.263057828\n",
"Iter: 955 loss: 0.202323899 test loss: 0.262522519\n",
"Iter: 956 loss: 0.20226334 test loss: 0.264608711\n",
"Iter: 957 loss: 0.203146964 test loss: 0.274307042\n",
"Iter: 958 loss: 0.202263415 test loss: 0.264653265\n",
"Iter: 959 loss: 0.202187032 test loss: 0.265314698\n",
"Iter: 960 loss: 0.202025294 test loss: 0.268111616\n",
"Iter: 961 loss: 0.204804048 test loss: 0.28504923\n",
"Iter: 962 loss: 0.202021241 test loss: 0.26866129\n",
"Iter: 963 loss: 0.201850757 test loss: 0.268879235\n",
"Iter: 964 loss: 0.201579392 test loss: 0.269885838\n",
"Iter: 965 loss: 0.201576889 test loss: 0.269798\n",
"Iter: 966 loss: 0.201379359 test loss: 0.268614\n",
"Iter: 967 loss: 0.201403826 test loss: 0.264450669\n",
"Iter: 968 loss: 0.201231152 test loss: 0.266495764\n",
"Iter: 969 loss: 0.201047868 test loss: 0.274130195\n",
"Iter: 970 loss: 0.201021641 test loss: 0.271936566\n",
"Iter: 971 loss: 0.2007875 test loss: 0.272512972\n",
"Iter: 972 loss: 0.200871035 test loss: 0.275997072\n",
"Iter: 973 loss: 0.20062539 test loss: 0.273853242\n",
"Iter: 974 loss: 0.200563699 test loss: 0.274185061\n",
"Iter: 975 loss: 0.200500667 test loss: 0.273957253\n",
"Iter: 976 loss: 0.200241476 test loss: 0.266217113\n",
"Iter: 977 loss: 0.201874077 test loss: 0.241625056\n",
"Iter: 978 loss: 0.20021005 test loss: 0.262787253\n",
"Iter: 979 loss: 0.200115189 test loss: 0.26090768\n",
"Iter: 980 loss: 0.200038567 test loss: 0.253711402\n",
"Iter: 981 loss: 0.200010151 test loss: 0.256107777\n",
"Iter: 982 loss: 0.199928582 test loss: 0.254136413\n",
"Iter: 983 loss: 0.200387299 test loss: 0.247563288\n",
"Iter: 984 loss: 0.199917331 test loss: 0.253118306\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Iter: 985 loss: 0.199847877 test loss: 0.252717465\n",
"Iter: 986 loss: 0.199667245 test loss: 0.251393169\n",
"Iter: 987 loss: 0.201262459 test loss: 0.251091599\n",
"Iter: 988 loss: 0.199635446 test loss: 0.250813633\n",
"Iter: 989 loss: 0.199352831 test loss: 0.255140454\n",
"Iter: 990 loss: 0.200970769 test loss: 0.274036825\n",
"Iter: 991 loss: 0.199308738 test loss: 0.257479787\n",
"Iter: 992 loss: 0.278157771 test loss: 0.311696768\n",
"Iter: 993 loss: 0.199043304 test loss: 0.256893963\n",
"Iter: 994 loss: 0.198745623 test loss: 0.261571616\n",
"Iter: 995 loss: 0.198721424 test loss: 0.260456949\n",
"Iter: 996 loss: 0.198522404 test loss: 0.260845542\n",
"Iter: 997 loss: 0.19850795 test loss: 0.262851745\n",
"Iter: 998 loss: 0.198358372 test loss: 0.261769176\n",
"Iter: 999 loss: 0.197956905 test loss: 0.261366338\n",
"Iter: 1000 loss: 0.198530346 test loss: 0.262739033\n",
"Iter: 1001 loss: 0.197758242 test loss: 0.261287689\n",
"Iter: 1002 loss: 0.196896762 test loss: 0.258020252\n",
"Iter: 1003 loss: 0.200982496 test loss: 0.255706221\n",
"Iter: 1004 loss: 0.196661338 test loss: 0.256359905\n",
"Iter: 1005 loss: 0.196100473 test loss: 0.263361931\n",
"Iter: 1006 loss: 0.198252067 test loss: 0.29868874\n",
"Iter: 1007 loss: 0.195980668 test loss: 0.269232482\n",
"Iter: 1008 loss: 0.195871323 test loss: 0.269950151\n",
"Iter: 1009 loss: 0.196775839 test loss: 0.275549948\n",
"Iter: 1010 loss: 0.195864305 test loss: 0.270239413\n",
"Iter: 1011 loss: 0.198493198 test loss: 0.308800876\n",
"Iter: 1012 loss: 0.195794478 test loss: 0.274628073\n",
"Iter: 1013 loss: 0.195720538 test loss: 0.277497917\n",
"Iter: 1014 loss: 0.196060359 test loss: 0.28931883\n",
"Iter: 1015 loss: 0.195708975 test loss: 0.279395878\n",
"Iter: 1016 loss: 0.195655718 test loss: 0.278042167\n",
"Iter: 1017 loss: 0.195764989 test loss: 0.273774713\n",
"Iter: 1018 loss: 0.195634052 test loss: 0.276637346\n",
"Iter: 1019 loss: 0.195587575 test loss: 0.277554721\n",
"Iter: 1020 loss: 0.195535347 test loss: 0.281634\n",
"Iter: 1021 loss: 0.19552882 test loss: 0.280567706\n",
"Iter: 1022 loss: 0.195436597 test loss: 0.28076908\n",
"Iter: 1023 loss: 0.195342571 test loss: 0.281870306\n",
"Iter: 1024 loss: 0.195323512 test loss: 0.281495363\n",
"Iter: 1025 loss: 0.19524017 test loss: 0.279729962\n",
"Iter: 1026 loss: 0.195602328 test loss: 0.2734752\n",
"Iter: 1027 loss: 0.195222571 test loss: 0.278559506\n",
"Iter: 1028 loss: 0.195200682 test loss: 0.278598815\n",
"Iter: 1029 loss: 0.195122 test loss: 0.278774709\n",
"Iter: 1030 loss: 0.194980189 test loss: 0.280053586\n",
"Iter: 1031 loss: 0.19497399 test loss: 0.279791623\n",
"Iter: 1032 loss: 0.194794104 test loss: 0.277462095\n",
"Iter: 1033 loss: 0.19524619 test loss: 0.26926437\n",
"Iter: 1034 loss: 0.194729313 test loss: 0.275190711\n",
"Iter: 1035 loss: 0.194648176 test loss: 0.275369912\n",
"Iter: 1036 loss: 0.194611311 test loss: 0.275789618\n",
"Iter: 1037 loss: 0.194569379 test loss: 0.275654078\n",
"Iter: 1038 loss: 0.19399479 test loss: 0.267161876\n",
"Iter: 1039 loss: 0.193920434 test loss: 0.269348115\n",
"Iter: 1040 loss: 0.194303721 test loss: 0.281236678\n",
"Iter: 1041 loss: 0.193567634 test loss: 0.273739606\n",
"Iter: 1042 loss: 0.193566754 test loss: 0.283248216\n",
"Iter: 1043 loss: 0.193520159 test loss: 0.278305769\n",
"Iter: 1044 loss: 0.193485782 test loss: 0.277800411\n",
"Iter: 1045 loss: 0.193766505 test loss: 0.276348323\n",
"Iter: 1046 loss: 0.193483487 test loss: 0.277647018\n",
"Iter: 1047 loss: 0.193465486 test loss: 0.269054651\n",
"Iter: 1048 loss: 0.193329588 test loss: 0.272383809\n",
"Iter: 1049 loss: 0.193202674 test loss: 0.274171114\n",
"Iter: 1050 loss: 0.193130508 test loss: 0.281612843\n",
"Iter: 1051 loss: 0.193075731 test loss: 0.278612137\n",
"Iter: 1052 loss: 0.192892119 test loss: 0.278077275\n",
"Iter: 1053 loss: 0.192455962 test loss: 0.276082814\n",
"Iter: 1054 loss: 0.213335991 test loss: 0.26837489\n",
"Iter: 1055 loss: 0.192449674 test loss: 0.275976688\n",
"Iter: 1056 loss: 0.191907898 test loss: 0.270612329\n",
"Iter: 1057 loss: 0.194892555 test loss: 0.265165538\n",
"Iter: 1058 loss: 0.191830486 test loss: 0.268303126\n",
"Iter: 1059 loss: 0.191701055 test loss: 0.2675412\n",
"Iter: 1060 loss: 0.191700384 test loss: 0.267576545\n",
"Iter: 1061 loss: 0.192149848 test loss: 0.263147801\n",
"Iter: 1062 loss: 0.191658244 test loss: 0.266487271\n",
"Iter: 1063 loss: 0.191540971 test loss: 0.268282503\n",
"Iter: 1064 loss: 0.191435486 test loss: 0.276214957\n",
"Iter: 1065 loss: 0.191408664 test loss: 0.273683876\n",
"Iter: 1066 loss: 0.191060856 test loss: 0.272569239\n",
"Iter: 1067 loss: 0.190897852 test loss: 0.268030912\n",
"Iter: 1068 loss: 0.19072406 test loss: 0.269976854\n",
"Iter: 1069 loss: 0.190603137 test loss: 0.270160168\n",
"Iter: 1070 loss: 0.191590309 test loss: 0.273354292\n",
"Iter: 1071 loss: 0.190593779 test loss: 0.270272195\n",
"Iter: 1072 loss: 0.190564334 test loss: 0.269934416\n",
"Iter: 1073 loss: 0.190665558 test loss: 0.268426657\n",
"Iter: 1074 loss: 0.190556034 test loss: 0.26963827\n",
"Iter: 1075 loss: 0.190508291 test loss: 0.27442798\n",
"Iter: 1076 loss: 0.190503284 test loss: 0.273239762\n",
"Iter: 1077 loss: 0.190617457 test loss: 0.280201346\n",
"Iter: 1078 loss: 0.19045563 test loss: 0.27534911\n",
"Iter: 1079 loss: 0.190374523 test loss: 0.272498459\n",
"Iter: 1080 loss: 0.190371707 test loss: 0.27292\n",
"Iter: 1081 loss: 0.190187797 test loss: 0.271359116\n",
"Iter: 1082 loss: 0.189901128 test loss: 0.265511185\n",
"Iter: 1083 loss: 0.189897761 test loss: 0.266054541\n",
"Iter: 1084 loss: 0.189563185 test loss: 0.266464204\n",
"Iter: 1085 loss: 0.190427721 test loss: 0.272036076\n",
"Iter: 1086 loss: 0.189448893 test loss: 0.267278522\n",
"Iter: 1087 loss: 0.189313024 test loss: 0.270781368\n",
"Iter: 1088 loss: 0.189936042 test loss: 0.286751628\n",
"Iter: 1089 loss: 0.189289689 test loss: 0.273249924\n",
"Iter: 1090 loss: 0.189231187 test loss: 0.2727274\n",
"Iter: 1091 loss: 0.189334571 test loss: 0.271115631\n",
"Iter: 1092 loss: 0.189205974 test loss: 0.272147328\n",
"Iter: 1093 loss: 0.188967884 test loss: 0.270119637\n",
"Iter: 1094 loss: 0.188429669 test loss: 0.262367874\n",
"Iter: 1095 loss: 0.196661711 test loss: 0.23967272\n",
"Iter: 1096 loss: 0.18839784 test loss: 0.260455549\n",
"Iter: 1097 loss: 0.188057482 test loss: 0.253594697\n",
"Iter: 1098 loss: 0.193254754 test loss: 0.242149889\n",
"Iter: 1099 loss: 0.188057095 test loss: 0.253303289\n",
"Iter: 1100 loss: 0.216121241 test loss: 0.250888377\n",
"Iter: 1101 loss: 0.187964693 test loss: 0.252218\n",
"Iter: 1102 loss: 0.187895611 test loss: 0.255205512\n",
"Iter: 1103 loss: 0.18845658 test loss: 0.269273698\n",
"Iter: 1104 loss: 0.187891036 test loss: 0.25624311\n",
"Iter: 1105 loss: 0.187803298 test loss: 0.25562039\n",
"Iter: 1106 loss: 0.187768981 test loss: 0.253940314\n",
"Iter: 1107 loss: 0.187721014 test loss: 0.254509807\n",
"Iter: 1108 loss: 0.187598869 test loss: 0.251948416\n",
"Iter: 1109 loss: 0.187647656 test loss: 0.242608175\n",
"Iter: 1110 loss: 0.187511727 test loss: 0.24767904\n",
"Iter: 1111 loss: 0.187406138 test loss: 0.245059013\n",
"Iter: 1112 loss: 0.187397182 test loss: 0.245597422\n",
"Iter: 1113 loss: 0.187332928 test loss: 0.245271429\n",
"Iter: 1114 loss: 0.187638134 test loss: 0.244552389\n",
"Iter: 1115 loss: 0.187321782 test loss: 0.245092168\n",
"Iter: 1116 loss: 0.187279597 test loss: 0.243965179\n",
"Iter: 1117 loss: 0.187418908 test loss: 0.240057752\n",
"Iter: 1118 loss: 0.187267169 test loss: 0.243038535\n",
"Iter: 1119 loss: 0.187251478 test loss: 0.242830247\n",
"Iter: 1120 loss: 0.187290043 test loss: 0.242093951\n",
"Iter: 1121 loss: 0.187246054 test loss: 0.242621616\n",
"Iter: 1122 loss: 0.187178075 test loss: 0.243008196\n",
"Iter: 1123 loss: 0.187018633 test loss: 0.244628802\n",
"Iter: 1124 loss: 0.188854322 test loss: 0.254437864\n",
"Iter: 1125 loss: 0.187005281 test loss: 0.245309845\n",
"Iter: 1126 loss: 0.186869502 test loss: 0.243727624\n",
"Iter: 1127 loss: 0.186928347 test loss: 0.238000557\n",
"Iter: 1128 loss: 0.186774895 test loss: 0.241121262\n",
"Iter: 1129 loss: 0.186671451 test loss: 0.242619514\n",
"Iter: 1130 loss: 0.187941015 test loss: 0.252051204\n",
"Iter: 1131 loss: 0.186670154 test loss: 0.242816895\n",
"Iter: 1132 loss: 0.187580377 test loss: 0.244732931\n",
"Iter: 1133 loss: 0.186658695 test loss: 0.242959633\n",
"Iter: 1134 loss: 0.186654 test loss: 0.243059292\n",
"Iter: 1135 loss: 0.186667278 test loss: 0.243553564\n",
"Iter: 1136 loss: 0.186652973 test loss: 0.243158266\n",
"Iter: 1137 loss: 0.186517641 test loss: 0.244620398\n",
"Iter: 1138 loss: 0.186500624 test loss: 0.251190275\n",
"Iter: 1139 loss: 0.186409339 test loss: 0.247988537\n",
"Iter: 1140 loss: 0.186219379 test loss: 0.246860042\n",
"Iter: 1141 loss: 0.186281905 test loss: 0.243562639\n",
"Iter: 1142 loss: 0.18608579 test loss: 0.245121941\n",
"Iter: 1143 loss: 0.186040476 test loss: 0.245825171\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Iter: 1144 loss: 0.186023757 test loss: 0.245544046\n",
"Iter: 1145 loss: 0.186003 test loss: 0.246262684\n",
"Iter: 1146 loss: 0.186181456 test loss: 0.249352813\n",
"Iter: 1147 loss: 0.186002269 test loss: 0.246469915\n",
"Iter: 1148 loss: 0.185988531 test loss: 0.246213183\n",
"Iter: 1149 loss: 0.185982481 test loss: 0.245288983\n",
"Iter: 1150 loss: 0.185975388 test loss: 0.245666429\n",
"Iter: 1151 loss: 0.185947284 test loss: 0.245986909\n",
"Iter: 1152 loss: 0.185884297 test loss: 0.247290328\n",
"Iter: 1153 loss: 0.186797604 test loss: 0.254228622\n",
"Iter: 1154 loss: 0.185881525 test loss: 0.247641489\n",
"Iter: 1155 loss: 0.185817689 test loss: 0.248307481\n",
"Iter: 1156 loss: 0.185880288 test loss: 0.251474738\n",
"Iter: 1157 loss: 0.185780376 test loss: 0.249407381\n",
"Iter: 1158 loss: 0.185754 test loss: 0.250289291\n",
"Iter: 1159 loss: 0.18573463 test loss: 0.254101\n",
"Iter: 1160 loss: 0.18572551 test loss: 0.25264594\n",
"Iter: 1161 loss: 0.185681447 test loss: 0.252298743\n",
"Iter: 1162 loss: 0.185534909 test loss: 0.250928402\n",
"Iter: 1163 loss: 0.185514942 test loss: 0.244730696\n",
"Iter: 1164 loss: 0.185373455 test loss: 0.247616023\n",
"Iter: 1165 loss: 0.185754925 test loss: 0.250266343\n",
"Iter: 1166 loss: 0.185284719 test loss: 0.248220831\n",
"Iter: 1167 loss: 0.185115859 test loss: 0.246145889\n",
"Iter: 1168 loss: 0.186018512 test loss: 0.240875915\n",
"Iter: 1169 loss: 0.185090035 test loss: 0.245096087\n",
"Iter: 1170 loss: 0.185047954 test loss: 0.245144144\n",
"Iter: 1171 loss: 0.185004979 test loss: 0.245364785\n",
"Iter: 1172 loss: 0.184997305 test loss: 0.245296821\n",
"Iter: 1173 loss: 0.184883595 test loss: 0.245538071\n",
"Iter: 1174 loss: 0.18452014 test loss: 0.246622726\n",
"Iter: 1175 loss: 0.184720904 test loss: 0.254994303\n",
"Iter: 1176 loss: 0.184210598 test loss: 0.249808416\n",
"Iter: 1177 loss: 0.185799137 test loss: 0.301487416\n",
"Iter: 1178 loss: 0.183715716 test loss: 0.263352245\n",
"Iter: 1179 loss: 0.183500543 test loss: 0.265983373\n",
"Iter: 1180 loss: 0.185126 test loss: 0.277555883\n",
"Iter: 1181 loss: 0.183481961 test loss: 0.26698029\n",
"Iter: 1182 loss: 0.183300391 test loss: 0.26669535\n",
"Iter: 1183 loss: 0.183990926 test loss: 0.274245858\n",
"Iter: 1184 loss: 0.183251783 test loss: 0.267014951\n",
"Iter: 1185 loss: 0.183035225 test loss: 0.268331\n",
"Iter: 1186 loss: 0.1831965 test loss: 0.274405271\n",
"Iter: 1187 loss: 0.182904929 test loss: 0.27067\n",
"Iter: 1188 loss: 0.182728067 test loss: 0.270603746\n",
"Iter: 1189 loss: 0.18306534 test loss: 0.270908982\n",
"Iter: 1190 loss: 0.18265146 test loss: 0.270587\n",
"Iter: 1191 loss: 0.248181224 test loss: 0.26347518\n",
"Iter: 1192 loss: 0.182541981 test loss: 0.268417299\n",
"Iter: 1193 loss: 0.182419747 test loss: 0.270299315\n",
"Iter: 1194 loss: 0.182881787 test loss: 0.279962152\n",
"Iter: 1195 loss: 0.182391107 test loss: 0.271940023\n",
"Iter: 1196 loss: 0.182261527 test loss: 0.272753924\n",
"Iter: 1197 loss: 0.184255078 test loss: 0.278747588\n",
"Iter: 1198 loss: 0.182261512 test loss: 0.272778869\n",
"Iter: 1199 loss: 0.182201579 test loss: 0.275239915\n",
"Iter: 1200 loss: 0.182943031 test loss: 0.286967248\n",
"Iter: 1201 loss: 0.182200924 test loss: 0.275515318\n",
"Iter: 1202 loss: 0.182176262 test loss: 0.275573522\n",
"Iter: 1203 loss: 0.18210943 test loss: 0.275812477\n",
"Iter: 1204 loss: 0.182574645 test loss: 0.277153164\n",
"Iter: 1205 loss: 0.182094678 test loss: 0.275996089\n",
"Iter: 1206 loss: 0.182046175 test loss: 0.274309039\n",
"Iter: 1207 loss: 0.182285979 test loss: 0.268496543\n",
"Iter: 1208 loss: 0.182038084 test loss: 0.27331993\n",
"Iter: 1209 loss: 0.182011619 test loss: 0.274883598\n",
"Iter: 1210 loss: 0.182052508 test loss: 0.281456918\n",
"Iter: 1211 loss: 0.181999847 test loss: 0.276964068\n",
"Iter: 1212 loss: 0.18196626 test loss: 0.278558791\n",
"Iter: 1213 loss: 0.181924149 test loss: 0.285207897\n",
"Iter: 1214 loss: 0.181920424 test loss: 0.283720791\n",
"Iter: 1215 loss: 0.181902304 test loss: 0.284579456\n",
"Iter: 1216 loss: 0.18206057 test loss: 0.288205206\n",
"Iter: 1217 loss: 0.181900963 test loss: 0.284837484\n",
"Iter: 1218 loss: 0.181896299 test loss: 0.285161197\n",
"Iter: 1219 loss: 0.181893349 test loss: 0.28503558\n",
"Iter: 1220 loss: 0.181885883 test loss: 0.285384446\n",
"Iter: 1221 loss: 0.181872398 test loss: 0.286814362\n",
"Iter: 1222 loss: 0.182168543 test loss: 0.294825256\n",
"Iter: 1223 loss: 0.181872085 test loss: 0.286915898\n",
"Iter: 1224 loss: 0.181833073 test loss: 0.287916541\n",
"Iter: 1225 loss: 0.18175064 test loss: 0.292197734\n",
"Iter: 1226 loss: 0.182954326 test loss: 0.320643067\n",
"Iter: 1227 loss: 0.181747153 test loss: 0.293322384\n",
"Iter: 1228 loss: 0.181690991 test loss: 0.293177158\n",
"Iter: 1229 loss: 0.18171604 test loss: 0.29323408\n",
"Iter: 1230 loss: 0.181653157 test loss: 0.293075383\n",
"Iter: 1231 loss: 0.181599 test loss: 0.292328596\n",
"Iter: 1232 loss: 0.181448922 test loss: 0.289526314\n",
"Iter: 1233 loss: 0.182122827 test loss: 0.278873563\n",
"Iter: 1234 loss: 0.181399137 test loss: 0.28690359\n",
"Iter: 1235 loss: 0.181345448 test loss: 0.29501304\n",
"Iter: 1236 loss: 0.181310967 test loss: 0.291801959\n",
"Iter: 1237 loss: 0.181163564 test loss: 0.293743938\n",
"Iter: 1238 loss: 0.18116051 test loss: 0.293487489\n",
"Iter: 1239 loss: 0.181137487 test loss: 0.291693598\n",
"Iter: 1240 loss: 0.181213915 test loss: 0.285064757\n",
"Iter: 1241 loss: 0.181131259 test loss: 0.290188432\n",
"Iter: 1242 loss: 0.181108847 test loss: 0.290014893\n",
"Iter: 1243 loss: 0.181044087 test loss: 0.289348781\n",
"Iter: 1244 loss: 0.181359023 test loss: 0.286915928\n",
"Iter: 1245 loss: 0.181020454 test loss: 0.288736284\n",
"Iter: 1246 loss: 0.183992505 test loss: 0.271347553\n",
"Iter: 1247 loss: 0.180788815 test loss: 0.284227937\n",
"Iter: 1248 loss: 0.18858476 test loss: 0.341157764\n",
"Iter: 1249 loss: 0.180687174 test loss: 0.288488269\n",
"Iter: 1250 loss: 0.180617601 test loss: 0.285984486\n",
"Iter: 1251 loss: 0.180591136 test loss: 0.286735862\n",
"Iter: 1252 loss: 0.180549383 test loss: 0.289121717\n",
"Iter: 1253 loss: 0.180681571 test loss: 0.299775302\n",
"Iter: 1254 loss: 0.180537686 test loss: 0.291362017\n",
"Iter: 1255 loss: 0.180415675 test loss: 0.294933647\n",
"Iter: 1256 loss: 0.180415034 test loss: 0.294841379\n",
"Iter: 1257 loss: 0.180314079 test loss: 0.296055526\n",
"Iter: 1258 loss: 0.180170819 test loss: 0.30089733\n",
"Iter: 1259 loss: 0.180164739 test loss: 0.300067216\n",
"Iter: 1260 loss: 0.179831356 test loss: 0.306274801\n",
"Iter: 1261 loss: 0.179687053 test loss: 0.334595859\n",
"Iter: 1262 loss: 0.179511622 test loss: 0.321644902\n",
"Iter: 1263 loss: 0.179735377 test loss: 0.315829217\n",
"Iter: 1264 loss: 0.17940031 test loss: 0.319177479\n",
"Iter: 1265 loss: 0.179609388 test loss: 0.325616479\n",
"Iter: 1266 loss: 0.179343149 test loss: 0.321139127\n",
"Iter: 1267 loss: 0.179636449 test loss: 0.34786436\n",
"Iter: 1268 loss: 0.179310799 test loss: 0.326338649\n",
"Iter: 1269 loss: 0.179287627 test loss: 0.328117967\n",
"Iter: 1270 loss: 0.179419309 test loss: 0.335983723\n",
"Iter: 1271 loss: 0.179285243 test loss: 0.329076\n",
"Iter: 1272 loss: 0.179244906 test loss: 0.330394119\n",
"Iter: 1273 loss: 0.179125 test loss: 0.335849792\n",
"Iter: 1274 loss: 0.179470986 test loss: 0.368314415\n",
"Iter: 1275 loss: 0.179063693 test loss: 0.343829334\n",
"Iter: 1276 loss: 0.179013059 test loss: 0.340432346\n",
"Iter: 1277 loss: 0.179010451 test loss: 0.341015458\n",
"Iter: 1278 loss: 0.178992391 test loss: 0.337331325\n",
"Iter: 1279 loss: 0.178991914 test loss: 0.337785363\n",
"Iter: 1280 loss: 0.178978652 test loss: 0.335844457\n",
"Iter: 1281 loss: 0.178996444 test loss: 0.328200102\n",
"Iter: 1282 loss: 0.178971052 test loss: 0.333222836\n",
"Iter: 1283 loss: 0.17895627 test loss: 0.332859397\n",
"Iter: 1284 loss: 0.178939849 test loss: 0.331473231\n",
"Iter: 1285 loss: 0.178937212 test loss: 0.331823677\n",
"Iter: 1286 loss: 0.178919181 test loss: 0.330974817\n",
"Iter: 1287 loss: 0.178875938 test loss: 0.327653855\n",
"Iter: 1288 loss: 0.179357544 test loss: 0.312741637\n",
"Iter: 1289 loss: 0.178871542 test loss: 0.326218188\n",
"Iter: 1290 loss: 0.178797856 test loss: 0.327325672\n",
"Iter: 1291 loss: 0.178731173 test loss: 0.333058745\n",
"Iter: 1292 loss: 0.178712294 test loss: 0.330935717\n",
"Iter: 1293 loss: 0.178496286 test loss: 0.3174164\n",
"Iter: 1294 loss: 0.178462029 test loss: 0.32076332\n",
"Iter: 1295 loss: 0.179870054 test loss: 0.316070586\n",
"Iter: 1296 loss: 0.178398922 test loss: 0.319598\n",
"Iter: 1297 loss: 0.178344786 test loss: 0.322995752\n",
"Iter: 1298 loss: 0.178477719 test loss: 0.337521732\n",
"Iter: 1299 loss: 0.17832689 test loss: 0.326671362\n",
"Iter: 1300 loss: 0.178303614 test loss: 0.328800768\n",
"Iter: 1301 loss: 0.178468123 test loss: 0.337765962\n",
"Iter: 1302 loss: 0.17830123 test loss: 0.329679906\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Iter: 1303 loss: 0.178274661 test loss: 0.329415739\n",
"Iter: 1304 loss: 0.178242385 test loss: 0.328489423\n",
"Iter: 1305 loss: 0.178238764 test loss: 0.328689694\n",
"Iter: 1306 loss: 0.178255752 test loss: 0.319993913\n",
"Iter: 1307 loss: 0.178199798 test loss: 0.324632913\n",
"Iter: 1308 loss: 0.178164214 test loss: 0.32207197\n",
"Iter: 1309 loss: 0.178185284 test loss: 0.312339664\n",
"Iter: 1310 loss: 0.178141162 test loss: 0.317858249\n",
"Iter: 1311 loss: 0.177971169 test loss: 0.319880337\n",
"Iter: 1312 loss: 0.179065123 test loss: 0.328969806\n",
"Iter: 1313 loss: 0.177954301 test loss: 0.320822597\n",
"Iter: 1314 loss: 0.177834868 test loss: 0.312189341\n",
"Iter: 1315 loss: 0.177823976 test loss: 0.314074844\n",
"Iter: 1316 loss: 0.177757904 test loss: 0.310954869\n",
"Iter: 1317 loss: 0.177756861 test loss: 0.311276495\n",
"Iter: 1318 loss: 0.177734047 test loss: 0.311630934\n",
"Iter: 1319 loss: 0.177732244 test loss: 0.313400954\n",
"Iter: 1320 loss: 0.17771478 test loss: 0.312460303\n",
"Iter: 1321 loss: 0.177675456 test loss: 0.313261449\n",
"Iter: 1322 loss: 0.17757903 test loss: 0.316552192\n",
"Iter: 1323 loss: 0.178438 test loss: 0.334846437\n",
"Iter: 1324 loss: 0.17756635 test loss: 0.318481594\n",
"Iter: 1325 loss: 0.177484095 test loss: 0.316292882\n",
"Iter: 1326 loss: 0.177579522 test loss: 0.308500797\n",
"Iter: 1327 loss: 0.177440986 test loss: 0.313310385\n",
"Iter: 1328 loss: 0.177391842 test loss: 0.314211458\n",
"Iter: 1329 loss: 0.177368969 test loss: 0.317975789\n",
"Iter: 1330 loss: 0.177344576 test loss: 0.316360712\n",
"Iter: 1331 loss: 0.177666709 test loss: 0.329564095\n",
"Iter: 1332 loss: 0.177304924 test loss: 0.31933558\n",
"Iter: 1333 loss: 0.177264988 test loss: 0.32344228\n",
"Iter: 1334 loss: 0.177551597 test loss: 0.341547132\n",
"Iter: 1335 loss: 0.177261502 test loss: 0.325124294\n",
"Iter: 1336 loss: 0.177236333 test loss: 0.324852198\n",
"Iter: 1337 loss: 0.177174896 test loss: 0.323818833\n",
"Iter: 1338 loss: 0.177742556 test loss: 0.319881618\n",
"Iter: 1339 loss: 0.177166402 test loss: 0.323290408\n",
"Iter: 1340 loss: 0.177105591 test loss: 0.324666053\n",
"Iter: 1341 loss: 0.177002087 test loss: 0.33045879\n",
"Iter: 1342 loss: 0.177001938 test loss: 0.330186874\n",
"Iter: 1343 loss: 0.176942483 test loss: 0.331407547\n",
"Iter: 1344 loss: 0.17710501 test loss: 0.336727709\n",
"Iter: 1345 loss: 0.17692329 test loss: 0.3326509\n",
"Iter: 1346 loss: 0.176911473 test loss: 0.332189769\n",
"Iter: 1347 loss: 0.176894888 test loss: 0.330403119\n",
"Iter: 1348 loss: 0.176894248 test loss: 0.330676138\n",
"Iter: 1349 loss: 0.176904961 test loss: 0.334687978\n",
"Iter: 1350 loss: 0.176866278 test loss: 0.332454711\n",
"Iter: 1351 loss: 0.176799327 test loss: 0.333501667\n",
"Iter: 1352 loss: 0.176681891 test loss: 0.33790803\n",
"Iter: 1353 loss: 0.17668204 test loss: 0.337842405\n",
"Iter: 1354 loss: 0.177811086 test loss: 0.380192965\n",
"Iter: 1355 loss: 0.176537201 test loss: 0.347480655\n",
"Iter: 1356 loss: 0.176455989 test loss: 0.345653653\n",
"Iter: 1357 loss: 0.177277267 test loss: 0.34014073\n",
"Iter: 1358 loss: 0.176452816 test loss: 0.34525463\n",
"Iter: 1359 loss: 0.176452 test loss: 0.35339278\n",
"Iter: 1360 loss: 0.176313579 test loss: 0.349233836\n",
"Iter: 1361 loss: 0.176241621 test loss: 0.354706407\n",
"Iter: 1362 loss: 0.176649809 test loss: 0.379708827\n",
"Iter: 1363 loss: 0.176229566 test loss: 0.357809275\n",
"Iter: 1364 loss: 0.176207647 test loss: 0.357099593\n",
"Iter: 1365 loss: 0.176250249 test loss: 0.354508519\n",
"Iter: 1366 loss: 0.176198915 test loss: 0.356284\n",
"Iter: 1367 loss: 0.176163971 test loss: 0.35633719\n",
"Iter: 1368 loss: 0.176082283 test loss: 0.3566131\n",
"Iter: 1369 loss: 0.17705074 test loss: 0.359612852\n",
"Iter: 1370 loss: 0.176074848 test loss: 0.35675177\n",
"Iter: 1371 loss: 0.175898865 test loss: 0.354135692\n",
"Iter: 1372 loss: 0.176627219 test loss: 0.343899161\n",
"Iter: 1373 loss: 0.175863475 test loss: 0.352285981\n",
"Iter: 1374 loss: 0.175880045 test loss: 0.329174\n",
"Iter: 1375 loss: 0.175575271 test loss: 0.340275079\n",
"Iter: 1376 loss: 0.175439596 test loss: 0.339378148\n",
"Iter: 1377 loss: 0.175710917 test loss: 0.335884303\n",
"Iter: 1378 loss: 0.175382063 test loss: 0.338357985\n",
"Iter: 1379 loss: 0.175260201 test loss: 0.339656949\n",
"Iter: 1380 loss: 0.176114514 test loss: 0.347865701\n",
"Iter: 1381 loss: 0.175252602 test loss: 0.340233952\n",
"Iter: 1382 loss: 0.175190464 test loss: 0.343743205\n",
"Iter: 1383 loss: 0.175105482 test loss: 0.358510584\n",
"Iter: 1384 loss: 0.175101072 test loss: 0.355544239\n",
"Iter: 1385 loss: 0.174994022 test loss: 0.354853362\n",
"Iter: 1386 loss: 0.175329134 test loss: 0.353289336\n",
"Iter: 1387 loss: 0.174963 test loss: 0.354332328\n",
"Iter: 1388 loss: 0.174879238 test loss: 0.35545966\n",
"Iter: 1389 loss: 0.174681261 test loss: 0.360132754\n",
"Iter: 1390 loss: 0.176917 test loss: 0.387536\n",
"Iter: 1391 loss: 0.174663186 test loss: 0.362153\n",
"Iter: 1392 loss: 0.174328342 test loss: 0.362872511\n",
"Iter: 1393 loss: 0.174849302 test loss: 0.367807597\n",
"Iter: 1394 loss: 0.174173191 test loss: 0.364130139\n",
"Iter: 1395 loss: 0.173698738 test loss: 0.365724981\n",
"Iter: 1396 loss: 0.179042771 test loss: 0.379320353\n",
"Iter: 1397 loss: 0.173680618 test loss: 0.366111\n",
"Iter: 1398 loss: 0.173567235 test loss: 0.35911566\n",
"Iter: 1399 loss: 0.173990846 test loss: 0.336443514\n",
"Iter: 1400 loss: 0.173543692 test loss: 0.353928179\n",
"Iter: 1401 loss: 0.173416167 test loss: 0.354317218\n",
"Iter: 1402 loss: 0.173275456 test loss: 0.356932223\n",
"Iter: 1403 loss: 0.173254579 test loss: 0.356051832\n",
"Iter: 1404 loss: 0.173144892 test loss: 0.354020327\n",
"Iter: 1405 loss: 0.1735989 test loss: 0.348427743\n",
"Iter: 1406 loss: 0.173121437 test loss: 0.352714211\n",
"Iter: 1407 loss: 0.173068032 test loss: 0.364204913\n",
"Iter: 1408 loss: 0.173008338 test loss: 0.359303743\n",
"Iter: 1409 loss: 0.172930032 test loss: 0.364113748\n",
"Iter: 1410 loss: 0.173048988 test loss: 0.384577364\n",
"Iter: 1411 loss: 0.172892824 test loss: 0.370627791\n",
"Iter: 1412 loss: 0.173262164 test loss: 0.364341795\n",
"Iter: 1413 loss: 0.172825858 test loss: 0.368539929\n",
"Iter: 1414 loss: 0.172772273 test loss: 0.365719676\n",
"Iter: 1415 loss: 0.172895297 test loss: 0.354984045\n",
"Iter: 1416 loss: 0.172751918 test loss: 0.362718642\n",
"Iter: 1417 loss: 0.17291142 test loss: 0.361534\n",
"Iter: 1418 loss: 0.172721371 test loss: 0.362280428\n",
"Iter: 1419 loss: 0.172701553 test loss: 0.361715764\n",
"Iter: 1420 loss: 0.172678992 test loss: 0.35952875\n",
"Iter: 1421 loss: 0.172676161 test loss: 0.360064059\n"
]
}
],
"source": [
"##### TRAIN MODEL WITH L-BFGS optimizer ##########\n",
"\n",
"## Define the type of loss function we want to use\n",
"loss_fun = tf.keras.losses.MeanSquaredError()\n",
"\n",
"## Create a function that agregates our cnn model, etc to pass it to the optimizer\n",
"func = function_factory(model_cnn, loss_fun, x_train_scaled_rowcol, y_train, x_test_scaled_rowcol, y_test)\n",
"\n",
"## Convert initial model parameters to a 1D tf.Tensor\n",
"init_params = tf.dynamic_stitch(func.idx, model_cnn.trainable_variables)\n",
"\n",
"## Train the model with L-BFGS solver\n",
"results = tfp.optimizer.lbfgs_minimize(value_and_gradients_function=func, \\\n",
" initial_position=init_params, max_iterations=500)\n",
"\n",
"## After training, the final optimized parameters are still in results.position\n",
"## so we have to manually put them back to the model\n",
"func.assign_new_model_parameters(results.position)"
]
},
{
"cell_type": "code",
"execution_count": 94,
"metadata": {
"ExecuteTime": {
"end_time": "2020-07-16T11:49:52.558401Z",
"start_time": "2020-07-16T11:49:51.655191Z"
},
"hidden": true
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"RMSE on train: 0.41554323\n",
"RMSE on test: 0.60005313\n"
]
},
{
"data": {
"image/png": "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\n",
"text/plain": [
""
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
}
],
"source": [
"# do some prediction\n",
"%matplotlib inline\n",
"y_train_pred3=model_cnn.predict(x_train_scaled_rowcol)\n",
"y_test_pred3=model_cnn.predict(x_test_scaled_rowcol)\n",
"print('RMSE on train:',np.sqrt(mean_squared_error(y_train,y_train_pred3)))\n",
"print('RMSE on test:',np.sqrt(mean_squared_error(y_test,y_test_pred3)))\n",
"\n",
"plt.semilogy(func.history[4:])\n",
"plt.legend(['loss (x_train)','val_loss (test)'])\n",
"plt.show()"
]
},
{
"cell_type": "markdown",
"metadata": {
"ExecuteTime": {
"end_time": "2020-07-16T11:46:12.482684Z",
"start_time": "2020-07-16T11:46:12.461705Z"
},
"hidden": true
},
"source": [
"From the figure above we can see that around iteration 700 the val_loss (metric of the test set) begins to rise a bit while the loss (metric for the train set) keeps going down. This is a sign that the model is over-fitting the data for optimizations beyond this point. By looking at the output list of the training, we can see that the minimum of val_loss happens at iteration 723 where we have:\n",
"\n",
"loss: 0.240415171 test loss: 0.246770635\n",
"\n",
"If we compute the sqrt() of these two mse we get: \n",
"\n",
"RMSE on train: 0.49032\n",
"RMSE on test: 0.49676"
]
},
{
"cell_type": "markdown",
"metadata": {
"ExecuteTime": {
"end_time": "2020-06-13T15:41:11.815118Z",
"start_time": "2020-06-13T15:41:11.697027Z"
},
"hidden": true
},
"source": [
"So, to sum up, with the L-BFGS optimizer using full batch size (the whole dataset at a time) we get results comparable with those obtained with the Adam optimizer, but with faster computation times."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Computing the Regression coefficients of a model\n",
"\n",
"From a Physics/Chemistry point of view, many researchers need to find which spectral features (absorption bands) are being more used to predict whatever they need to predict. This is helpful because absorption bands allow to identify chemical compounds and in last instance to help understand what physical or even biological processes are in play. So, there is a need to explore the interpretability of NN models. Cui and Fearn 2018, suggest looking at the \"*regression coefficients of the NN*\" (see section 2.7 of the paper for details) . Here below we try to implement this type of \"feature importance\" visualization for the CNN and the PLS models."
]
},
{
"cell_type": "code",
"execution_count": 61,
"metadata": {
"ExecuteTime": {
"end_time": "2020-07-20T10:21:49.772171Z",
"start_time": "2020-07-20T10:21:49.750201Z"
}
},
"outputs": [],
"source": [
"## Compute predictions using the PLS model on the dataset with preprocessing = detrend + SNV \n",
"\n",
"## PLS MODEL\n",
"pls_model = PLSRegression(n_components=8)\n",
"## Fit PLS model to train data and predict test set\n",
"pls_model.fit(x_train_prep, y_train)\n",
"pls_pred = pls_model.predict(x_test_prep)"
]
},
{
"cell_type": "code",
"execution_count": 62,
"metadata": {
"ExecuteTime": {
"end_time": "2020-07-20T10:21:52.326347Z",
"start_time": "2020-07-20T10:21:52.270493Z"
},
"code_folding": []
},
"outputs": [],
"source": [
"## Compute predictions using the CNN model on the dataset with preprocessing = standardize rows and columns \n",
"\n",
"## CNN MODEL\n",
"## Load the precomputed model best weights\n",
"model_cnn.load_weights('Cui_cnn2.h5')\n",
"## make prediction\n",
"cnn_pred=model_cnn.predict(x_test_scaled_rowcol)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Probably implementing a separate function to compute the regression coefs. ($w_i$) would be more elegant, but for the sake of simplicity I'll implement everything sequentially..."
]
},
{
"cell_type": "code",
"execution_count": 69,
"metadata": {
"ExecuteTime": {
"end_time": "2020-07-20T10:25:49.048195Z",
"start_time": "2020-07-20T10:25:46.779735Z"
}
},
"outputs": [],
"source": [
"## Define a subset of 100 samples to compute CNN and PLS regression coefficients\n",
"x_subset_cnn=x_test_scaled_rowcol[:100,:]\n",
"x_subset_pls=x_test_prep[:100,:]\n",
"\n",
"y_subset_cnn=cnn_pred[:100,:].copy()\n",
"y_subset_pls=pls_pred[:100,:].copy()\n",
"\n",
"## Define epsilon\n",
"epsilon=1E-5\n",
"\n",
"## Create matrices for CNN and PLS regression coefs\n",
"w_cnn = np.zeros((x_subset_cnn.shape[0],x_subset_cnn.shape[1]))\n",
"w_pls = np.zeros((x_subset_pls.shape[0],x_subset_pls.shape[1]))\n",
"\n",
"## Compute regression coeffs according to eq. (7) of the paper\n",
"for i in np.arange(x_subset_cnn.shape[1]):\n",
" ## create array with wavelenghts (aka features)\n",
" x_pert_cnn=x_subset_cnn.copy()\n",
" x_pert_pls=x_subset_pls.copy()\n",
" ## apply epsilon pertubation to feature i only\n",
" x_pert_cnn[:,i]=x_subset_cnn[:,i]+epsilon\n",
" x_pert_pls[:,i]=x_subset_pls[:,i]+epsilon\n",
" ## compute new model prediction where the input is the locally perturbed spectra\n",
" y_pred_pert_cnn=model_cnn.predict(x_pert_cnn)\n",
" y_pred_pert_pls=pls_model.predict(x_pert_pls)\n",
" ## compute the regression coefficients\n",
" w_cnn[:,i]= (y_pred_pert_cnn[:,0] - y_subset_cnn[:,0])/epsilon\n",
" w_pls[:,i]= (y_pred_pert_pls[:,0] - y_subset_pls[:,0])/epsilon\n",
" "
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Visualize the regression coefficients for PLS and the CNN. Note: The PLS regress. coef. are the same for every sample so we just need to plot the 1st one. For the CNN, the regress. coef. is sample dependent (we plot 100 samples here)."
]
},
{
"cell_type": "code",
"execution_count": 74,
"metadata": {
"ExecuteTime": {
"end_time": "2020-07-20T10:32:10.944580Z",
"start_time": "2020-07-20T10:32:10.506959Z"
}
},
"outputs": [
{
"data": {
"image/png": "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\n",
"text/plain": [
""
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
}
],
"source": [
"plt.figure(figsize=(12,6))\n",
"plt.subplot(211)\n",
"plt.plot(np.arange(100),w_pls[0])\n",
"# plt.plot(np.arange(100),w_pls.T)\n",
"plt.title('PLS regression coefficients')\n",
"plt.ylabel(r'$w_i$', fontsize=14)\n",
"plt.axhline(0, lw='1',ls='dashed',c='k')\n",
"plt.subplot(212)\n",
"plt.plot(np.arange(100),w_cnn.T)\n",
"plt.title('CNN regression coefficients')\n",
"plt.ylabel(r'$w_i$', fontsize=14)\n",
"plt.axhline(0, lw='1',ls='dashed',c='k')\n",
"plt.xlabel('Index of wavelength')\n",
"plt.tight_layout()\n",
"plt.show()"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Our PLS reg. coef. are not as noisy as they appear in the paper. On the other hand the CNN reg. coef. are noisier than they appear in the paper!! At this point I cannot tell if this difference is due to the different implementation of the models or if it is due to a bug in computation of $w_i$. The scale for the CNN $w_i$ is also different from that of the paper's figure 14..."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Visualize the activations/outputs of the intermediate layers of the CNN "
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"In order to visualize the outputs of the convolution layer, we use this piece of code that we already used in other notebook. It was adapted from https://stackoverflow.com/questions/54195973/how-to-most-effectively-visualize-1d-convolutions"
]
},
{
"cell_type": "code",
"execution_count": 80,
"metadata": {
"ExecuteTime": {
"end_time": "2020-07-20T11:26:19.020720Z",
"start_time": "2020-07-20T11:26:18.957736Z"
}
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"(108, 100, 1)\n",
"Model: \"model_4\"\n",
"_________________________________________________________________\n",
"Layer (type) Output Shape Param # \n",
"=================================================================\n",
"reshape_1_input (InputLayer) [(None, 100)] 0 \n",
"_________________________________________________________________\n",
"reshape_1 (Reshape) (None, 100, 1) 0 \n",
"_________________________________________________________________\n",
"conv1d_1 (Conv1D) (None, 100, 1) 6 \n",
"=================================================================\n",
"Total params: 6\n",
"Trainable params: 6\n",
"Non-trainable params: 0\n",
"_________________________________________________________________\n",
"None\n"
]
}
],
"source": [
"from tensorflow.keras import models\n",
"\n",
"# get the first 3 layers from the trained model (the input layer doesnt count!)\n",
"layers_outputs = [layer.output for layer in model_cnn.layers[:2]]\n",
"\n",
"# define a sub-model based on the selected layers\n",
"activation_model = models.Model(inputs=model_cnn.input, outputs=layers_outputs)\n",
"\n",
"# compute the activations as the output of that sub-model. Here we feed x_test2_emsc data to that submodel\n",
"activations = activation_model.predict(x_test_scaled_rowcol)\n",
"\n",
"# The last element of the \"activations\" var correspond to the outputs of \n",
"# conv1d layer and has dimensions (Nsamples, SampleSize-FilterSize+1, Nfilters)\n",
"\n",
"# activations for the first conv1d layer\n",
"conv1d_activations = activations[1]\n",
"\n",
"print(conv1d_activations.shape)\n",
"print(activation_model.summary())"
]
},
{
"cell_type": "markdown",
"metadata": {
"ExecuteTime": {
"end_time": "2020-07-20T11:29:39.834791Z",
"start_time": "2020-07-20T11:29:39.826804Z"
}
},
"source": [
"In order to check what kind of pre-processing is being done in the convolutional layers, we can plot the activations/output of those layer (red dashed lines and blue shading) overlayed on a mean-spectrum of the input data. This works because in our Conv1d layers we defined padding='same' that ensures that the dimension of the data after the convolution layer is the same as the input data. \n",
"\n",
"There are multiple test samples and therefore multiple activations... We can plot the mean activation to have an idea of the transformation the conv. layer is doing or we can plot all activations in the set if we want"
]
},
{
"cell_type": "code",
"execution_count": 156,
"metadata": {
"ExecuteTime": {
"end_time": "2020-07-20T12:10:43.212328Z",
"start_time": "2020-07-20T12:10:42.446305Z"
}
},
"outputs": [
{
"data": {
"image/png": "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\n",
"text/plain": [
""
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
},
{
"data": {
"image/png": "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\n",
"text/plain": [
""
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
},
{
"data": {
"image/png": "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\n",
"text/plain": [
""
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
}
],
"source": [
"## Plot the mean output (over all samples). Note that here we only have 1 filter in the conv1d layer\n",
"plt.figure(figsize=(12,3))\n",
"ax=plt.subplot()\n",
"plt.title('Mean of activations in Conv1d layer')\n",
"## get lims for imshow\n",
"extent=[min(x_scale), max(x_scale),-2,2]\n",
"## plot just activation for sample 1\n",
"#ax.imshow(conv1d_activations[1, : , 0][np.newaxis,:], cmap=\"RdBu\", aspect=\"auto\", extent=extent) \n",
"## plot the mean activation (over all samples) as a bluish background\n",
"ax.imshow(np.abs(np.mean(conv1d_activations[:, : , 0],axis=0)[np.newaxis,:]), cmap=\"Blues\", aspect=\"auto\", extent=extent, label='dd') \n",
"## plot the mean of the activations\n",
"plt.plot(x_scale, np.mean(conv1d_activations[: , :, 0],axis=0), 'r--', label='mean of activations')\n",
"## plot the mean of the test spectra\n",
"plt.plot(x_scale, np.mean(x_test_scaled_rowcol, axis=0),'k',label='mean input spectrum')\n",
"plt.xlabel(r'$\\lambda$ (nm)')\n",
"plt.legend()\n",
"plt.show()\n",
"\n",
"plt.figure(figsize=(12,3))\n",
"ax1=plt.subplot()\n",
"plt.title('Individual activations (red) and test spectra (black)')\n",
"# ax.plot(x_scale, np.mean(conv1d_activations[: , :, 0],axis=0), 'r--', label='mean of activations')\n",
"g1=ax1.plot(x_scale, conv1d_activations[: , :, 0].T, 'r-', lw=1, alpha=0.2, label='activations')\n",
"# ax.plot(x_scale, np.mean(x_test_scaled_rowcol, axis=0),'k',label='input spectra')\n",
"g2=ax1.plot(x_scale, x_test_scaled_rowcol.T,'k',alpha=0.1, lw=1,label='mean input spectrum')\n",
"plt.xlim(min(x_scale), max(x_scale))\n",
"plt.xlabel(r'$\\lambda$ (nm)')\n",
"plt.show()\n",
"\n",
"## Compute first derivative of test spectra\n",
"x_test_scaled_rowcol_1stderiv= savgol_filter(x_test_scaled_rowcol,3,2,deriv=1)\n",
"\n",
"plt.figure(figsize=(12,3))\n",
"ax=plt.subplot()\n",
"plt.title('Mean of activations (red) and 2 x mean of 1st derivative')\n",
"## get lims for imshow\n",
"extent=[min(x_scale), max(x_scale),-2,2]\n",
"## plot the mean of the activations\n",
"plt.plot(x_scale, np.mean(conv1d_activations[: , :, 0],axis=0), 'r--', label='mean of activations')\n",
"## plot the mean of the 1st derivative of spectra\n",
"plt.plot(x_scale, 2*np.mean(x_test_scaled_rowcol_1stderiv, axis=0),'k',label='2 x mean of 1st derivative of spectrum')\n",
"plt.xlabel(r'$\\lambda$ (nm)')\n",
"plt.ylim(-0.5,1)\n",
"plt.legend()\n",
"plt.show()"
]
},
{
"cell_type": "markdown",
"metadata": {
"ExecuteTime": {
"end_time": "2020-07-20T11:48:11.663083Z",
"start_time": "2020-07-20T11:48:11.645089Z"
}
},
"source": [
"By looking at the panel of the middle, those with a trained eye can see that the activations curves resemble the behaviour of a first derivative of the spectra. To test that hypothesis I implemented a comparison (last panel) between the mean of the activation and a scaled version of the mean of 1st derivatives (computed using the Sav. Golay filter). We can see that the activations in the convolutional layer behave as some kind of \"non-zero centered wavelength-weighted\" 1st derivative of the spectra. Food for thought!!"
]
}
],
"metadata": {
"hide_input": false,
"kernelspec": {
"display_name": "Python 3",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.6.12"
},
"toc": {
"base_numbering": 1,
"nav_menu": {},
"number_sections": true,
"sideBar": true,
"skip_h1_title": true,
"title_cell": "Table of Contents",
"title_sidebar": "Contents",
"toc_cell": false,
"toc_position": {
"height": "588px",
"left": "28px",
"top": "110px",
"width": "265px"
},
"toc_section_display": true,
"toc_window_display": false
}
},
"nbformat": 4,
"nbformat_minor": 2
}