nlp-lab/Project/advanced_approach
2018-07-27 14:50:47 +02:00
..
doc2VecModel.p renamed a lot of stuff 2018-07-27 12:11:38 +02:00
Evaluation_sentiment_dataset.ipynb renamed a lot of stuff 2018-07-27 12:11:38 +02:00
Evaluation.ipynb renamed a lot of stuff 2018-07-27 12:11:38 +02:00
README.md added README.md for advanced approach 2018-07-27 14:50:47 +02:00
simple_twitter_learning_old.ipynb renamed a lot of stuff 2018-07-27 12:11:38 +02:00
simple_twitter_learning.ipynb renamed a lot of stuff 2018-07-27 12:11:38 +02:00
test.csv renamed a lot of stuff 2018-07-27 12:11:38 +02:00
twitter_learning.py renamed a lot of stuff 2018-07-27 12:11:38 +02:00

Advanced Approach


Folder Overview

Filename short_description
twitter_learning.py module containing the main classes for the learning process
Learner.ipynb notebook containing a user interface to control the learn process
Evaluation_sentiment_dataset.ipynb notebook creating an evaluation on the sentiment dataset

twitter_learning.py

We wanted the possibility to load and save classifier and also to continue the train process of a loaded classifier on other datasets. Since this functionality doesn't fit well in a single linear processed notebook, we wrote this module containing three classes for managing train data, classifier and train process:

sample_data_manager

class for preprocessing twitter data provided as json files. Creates samples with sentiment labels and provides train and validation samples

pipeline_manager

creates, saves or loads a sklean-pipeline with vectorizer and keras classifier. When a new classifier is created, the vectorizer is fit by all samples that a given sample_data_manager has stored at that moment.

trainer

expects a pipeline_manager and a sample_data_manager. It controls the train process and feeds the pipeline with data from the sample_data_manager. Since we want the possibility to continue training on loaded classifiers, this class modifies the pipeline during runtime to prevent a reset of the vectorizers while fitting new data.


Learner.ipynb

In short: the user interface for methods provided in twitter_learning.py This file provides all controls for the train process and feeds the classifier with data. To use it just run all cells and jump to the user interface Part. In order to train on a dataset, just use it in the following way:

load datasets

  • setup the root folder for the json files containing our twitter samples
  • then you can set the range of files that will be loaded. Also you can setup more preprocessing steps, like top-emoji usage filtering or only load samples containing a specific set of emojis (given as string)

create/save/load classifier

<img src="data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAABUcAAAIjCAYAAAAgHao4AAAABHNCSVQICAgIfAhkiAAAABl0RVh0U29mdHdhcmUAZ25vbWUtc2NyZWVuc2hvdO8Dvz4AACAASURBVHic7N15WJT13sfxz7CJCKi4b4iYCrigmLa4dARxya180nJLLTE7LbaqZadOx5N2TC3PSXtSK+2UZZrllpYCetRKcyUEBRUR0MQFZZGdef7wOI+IjKgDwzDv13VxXXLf9/x+37mH+Q5+uBeD0Wg0CgAAAAAAAADsjIO1CwAAAAAAAAAAayAcBQAAAAAAAGCXCEcBAAAAAAAA2CWnq/+IiIiwZh2wYROOdLV2CbBhS9rstnYJAOxMcHCwtUsAAAAAUEk4XftNp06drFUHbNkRaxcAW0bfwe3Yv38/Pzu4Lfv377d2CQAAAAAqEaebbwKYt6TNbo7CwW25csQ6ARcAAAAAALAOrjkKAAAAAAAAwC4RjgIAAAAAAACwS4SjAAAAAAAAAOwS1xytAJfS01VUVHTT7Wp6esrBgbwagHXl5eXJxcXF2mUAAAAAAFDuCEfL2boNG3UyKUmS5OXlpY6dOskgg2l9ckqynJ2c5OHpqePHjmrwgAcJJQBYRXZ2ttas36isy5fl5VVbgx/sJ0dHR2uXhXJ09uxZxcTE6MKFC3JwcFDjxo3Vtm1bubm5Wbs0AAAAAKgQt3WY4qVLl/TOO+9YuhatXLlSu3fvtvi21nQ1GJUkz5o11aKFr3xatDB9xcfHKzklRS19W+rMmVRt2PijFasFUFbl1QdLEx0drTlz5ujw4cPlMn5ySoq+XrVaWdmXJYN0IS1NX65YqbSLF8tlvsri4sWLmjlzprXLsLhVq1aZ/YwsKirSpk2btHLlStWoUUN333232rVrpwsXLmjhwoU6dOhQBVYLAAAAANZT5c/hfu+995Sammozc5w6fdoi4+DWHDx4UK+99prV509MTJSHh4fFxx8xYoQWLFhwS48pr1ruxO2+106ePKk333xTly5dKoeqysf1Nf/888/q16+f/Pz8tHr1aq1YscJic/22d59+3BKpIqOx2PL8ggJ9t3aDYg8fsdhcqBzWrl2rzMxMPfXUU7r77rvVrFkztWzZUn379tXo0aP1ww8/KCYmxtplAgAAAEC547R6VEmNGjXSli1b1LZt2zJtv2bNGg0ePLicq6q889u6zMxMVa9evdRTwOvUqaNHHnnEpk4Vvr7m7Oxs1a5dW5L0wAMPyMvL65bGu9E+Kioq0roNm3TuwgVdvdqHsTBfxsJCGZxcZHBwkFFG/bzrN6Wc/kO9ez1gmSdnp272c1pRjh49qv3796tWrVr68MMPb7hNYWGh1q1bJ19fX7m6ulZwhQAAAABQcSwSjp48eVIbN27UuXPn5OXlpd69e6tVq1aSpNjYWG3ZskVpaWlyd3dXjx491KVLF0lXThONjIxUVlaW/Pz8VFBQUOoc5rYtbY5ly5YpKytLS5cu1SOPPCJfX1+z9WzdulV79uxRTk6Omjdvrocfflju7u6SrlyXbcOGDUpJSZGnp6eCg4PVtm3bG85Rmoz0dJ1ISCi2zMenhZydnXT8+PHb2/lV1JkzZ1S7du0Ku/7qr7/+qjfeeKNC5jI3f9I1l2FAcde/1+rXr6+PP/5YXbt2VXh4uP785z/r/PnzN3x/G41Gfffdd+rQoYMyMzP10UcfadCgQdq/f7+Sk5Pl5+enQYMG3XDeEydOaNOmTTp37pwaNGiggQMHqlGjRiW2u93eUtq6a2v+8ssvlZaWpi+++EJDhgxRSkqK8vPz1a9fP0ml96fMzMwS+6h+/fqmmtes/0EXLl40BaPVnSU3l0KlXCySs0ORHK/eIM4gJSYlKTxym0JsMCCNi4vT3r17Vb9+fcXHx6ugoED33HOP6fW5VmxsrMLDw02vY/fu3dWlSxetX79eTk5Opn1+9fTzV155Ra6urjp69Kh++uknZWRkqE2bNsrKylJgYKB8fHy0aNEidenSRREREXr66aeVk5Nj+pmqXbu26TPz3Llz+uSTTzR16lRTPQsXLlS/fv1Uv35908/tgQMHlJycrDZt2mjgwIEyGAyKjo7W1q1blZWVpTZt2ig/P7/U/bF3716NGTNG4eHhOnXqlCTJ1dVVOTk5pm2GDh2qhIQEHT58WB07drTUSwEAAAAAlc4dn1afmZmpL774Qp07d9ZLL72knj17asWKFUpLS1NBQYFWrVqlbt266ZVXXlG/fv20YcMG5ebm6vTp01q9erWCg4P1/PPPq27duoqNjb3hHOa2NTfH2LFjVaNGDY0bN06+vr5mt01ISNAvv/yikSNH6vnnn1dBQYG2b99umuPzzz9XmzZt9NJLL6lv3776/vvvlZKSUmKOG3J0Uu1O/aW7HtCBTM9iX5fqdtC5mgGKTrOfm55s3bpVnTt3lru7u+677z4dOHBA0pVAtFmzZnr33XfVuHFjxcfHKzY2VsHBwfLw8JCfn5+++eYb0zirV6+Wv7+/XF1d5ePjo4ULF0qSevfurdTUVIWEhCg8PFySzI6TnJyspk2byuG/QdCMGTPUpEkTubu7q1+/fjpz5sxN5xw4cKD++te/mrZLTU2Vs7OzEv4bht/K/NfasWOHunbtKg8PD3Xq1EkbN268aS2StGLFCvn5+cnLy0vjx49Xdnb2Lb8e1zM33/X7rDzc6L2WmZmpc+fO6eWXX5aXl1ep7+/rZWdn68yZMxoxYoSeffZZRUVF6cSJEyW2S09P1/Lly9WzZ0+98MIL8vb21qpVq0psd7u9xdy6a40aNUq1a9fW6NGj5efnV2Lu0vrT9fvo2mBUkgLbtzPdHq52daP6ti5UkYOLXKq5ytGp+N/OHB0c1b5d2Y7EroyOHj0qDw8PTZo0SWPGjNH27dtL/FGqoKBA3377re6//369/PLL6tu3r3744Qfl5OTI39+/2GdUdHS0WrduLVdXV126dEnffPONevXqpeeee06enp6Ki4szbXv1NXjppZfk5uamL7/8Up07d9aLL76onj176ptvvlFaWtpNn0NOTo7OnDmjxx57TM8884x+//13JSYm6vTp0/ruu+9M89etW9fstWn/+OMP+fj4qEePHnJzc5Obm5sefvhhtWvXTo6OjmrevLlat26t1q1b6zSXegEAAABQxd1xOBobG6uGDRvq7rvvVvXq1dW2bVu1bNlS0dHRMhgMmjRpkoKCguTm5qZatWpJuvIfvOjoaLVt29Z0V9zu3buX+I/7Vea2NTfH9cxtm5eXJ4PBoNzcXHl4eOjxxx9X7969JUnx8fGm4Kh69epq3bq1OnTooGPHjpVpH7k4O8u3+mX5OJ4v9auFR+Gt7XgblZKSosGDB+u1117TsWPHdP/992vEiBGm9WfOnFFsbKxSUlLk6+urPn36aNCgQTp58qTmzp2rJ598Urt371ZOTo5Gjx6tV155RSkpKZozZ46ee+45paena8uWLapfv77Cw8MVEhKi7OzsUseRip/SHhERoXnz5mnt2rU6fPiwcnNzNWvWLEkyO+ejjz6q1atXm57H1aCsRYsWtzT/tc6cOaMHH3xQYWFhOnnypF5//XU98sgjOn78uNlaDhw4oLFjx+pvf/ub4uPj5e/vrzVr1tzW63GVuflutM8qSmFhoUJDQ+Xh4XFLvaCgoMB01GD16tXVuHFjXb58ucR20dHR8vHxUUBAgNzd3RUcHKwuXbqosLD4+/V2e4u5dWV1s/507T66nm8LHz3y8BA1qeWkni2M2hzvqLQcBzk4Fv9oqOHmppGPPqL69ereUm2VSbVq1Uyvec2aNRUUFKSDBw8W28ZgMGjixImm17FmzZqSrryOLVq0UF5eniksjI6OVvv27SVJUVFRatWqlfz9/eXm5qZevXqZjgyWir8GVz8zO3fuXOIz82au/7lt1KiRLl++rEOHDikgIKBMn6dXn6eTk5MCAgI0ZcoUTZ06VX5+fho2bJj+8pe/6IknnlD16tXl6upa4mcdAAAAAKqaOz6tPj09vcS177y8vHTp0iU5OjrqxIkTWr9+vXJyclSnTh0ZDFeOU8rIyCjxuDp16txwDnPbmpvjeua2bdOmjUJCQrRx40ZdunRJ/v7+6tatm+rVq6e0tDSlpqZq7ty5xcbr0KFDmfaRi5OjOga0NrtNbk6ODu7eWabxbNnXX3+tXr166ZFHHpF05YhDb29v5eXlSZLy8/M1e/ZsNWjQQN99950aNmyoF198UZI0YMAAjRo1Slu2bFHHjh21Z88eBQQESJJatGghg8GgtLQ0eXp6Fptz06ZNpY7TtWtXbdmyRV999ZUk6fLlyzIYDEpPT1enTp20adMmUzjg4OBQ6pxDhgzRxIkTFR8fr1atWumbb77RuHHjbnn+a33//ffq1KmTwsLCJEnDhg3T8uXLtWLFCr388sul1vL111/rf/7nfzR8+HBJ0pQpU7R06dLbej2uMvfcb7TPKoqDg4MphLqVXuDg4FCm649evHixWO9xdnbWvffeW2K72+0t5taV1c3607X76EY8PTzUv3N9fRGeqJwig+RwzT4rMiq/IF++Af5ycXYuc02V0dUA/aqaNWsqMTGx2DaOjo5KTEw0HS167evo4OCgNm3aKDY2Vk5OTkpPTzddPubSpUum68FKV8LHqwH51cdefQ3S09OLbStJtWvXLvVmYcZrbpJ1/c/t1dpK+xw2ty8uXbokZ2dn5efnm0Lgq2OmpKSoSZMmOn/+/C1f2xYAAAAAbM0dh6Oenp46efJksWVpaWlq1KiRjhw5ooiICD3xxBOqV6+ejEajZsyYYXrc9acRlnZaobltzc1xPXPbXrp0SS1btlSXLl2Unp6u7du366uvvtLzzz8vd3d3NW3aVOPHjzeNlZGRYfWbatiixMTEYpcfcHNz0+TJk03fOzg4qEGDBpKunHJ86NAhNWvWrNgYo0aNkouLi7Zu3apnnnlGly5dUqtWrW54WvrNxklPT5eLi4vphiMDBgzQzJkzNXnyZJ08eVIPP/ywpkyZIn9/f7Nzenp6qn///lq9erUef/xx7du3T+vXr7/l+a+VnJxc4lINLVu2VFJSktlaUlJSTKHNVa1b3zicv9nrcZW5+W60zz777LMbzmdp14Zdt9ILSgtNr+fh4WG6JqN05QZGv/zyizp37lxsu9vtLebWlZW5/lRUVHTT52owFsil6JL6dmumb7ackEu1ajI4GGQsMiovN1dGY5EuXy79sgy2Ij09XYWFhaa+ff78+RKh8ZEjRxQZGanx48erXr16Kioq0jvvvGNa7+/vr/DwcDk4OMjf319O/730gIeHh86ePWvazmg0Kj093fT9ta+Bp6dniWsLX7x40XQd22vDUKPRqIyMjBuOc61b+TyVpHbt2mnPnj1q1aqV1q5dq969e8vX11fZ2dn6+eefdfbsWY0aNUoHDhww/eEEAAAAAKqqOz6t3s/PT6dOndKBAweUl5en2NhYxcfHq127dsrKypKrq6s8PDyUlZWliIgIFRQUKDMzU+3bt9ehQ4cUGxur3Nxc7dq1q9h/Lq9lbltzc1x19Sg4c9vGxcVp8eLFOnv2rJydneXq6irn/x4p1apVK6Wmpmr//v3Ky8tTYmKiPvzww2L1Xn+kHW6sUaNGSk5ONn1fUFCg9957z3TU1LUBZ8OGDdW1a1clJSWZvnbv3q1XX31V69ev19tvv62FCxdq3759Wr58eanhqLlxNm7cqP79+5u2TUpKUmhoqKKiohQdHS13d3cNGTJEkm4652OPPabVq1dr1apVxW64cyvzX6tJkyYlroOZkJCgJk2amK2ladOmJS75kHDdjcDK+npcZW6+G+2z8lTae60sveBWtW3bVvHx8Tpy5Iiys7P1n//8R7t27VK1atXKPLe53mJuXVmVpT/djMGtlmp7uMpoLFJuTrbycvOUm5Mto7FIrq6uCuoUeEs1VUaXL19WeHi4cnJylJCQoD179pQ4+j8rK0vVqlWTu7u7MjMzFRkZqYKCAmVlZUm68seJS5cu6bfffjOdUi9dCRvj4uIUFxen3Nxc7dixQ1lZWTcMM9u0aWP6zMzNzTV9ZgYEBMjV1VXZ2dmKiYlRUVGRtm/fbvbGSle1b99eMTExxT4jz507V+r2QUFBOn78uPLz8/XII4/o999/16JFi7RixQq5u7tr5MiRCg8Pl4+PT4mjXAEAAACgqrnjcNTDw0OjRo3Sr7/+qtmzZysyMlLDhw9XnTp11KFDBzVo0EDz5s3Tp59+Kg8PDz3wwANau3atGjRooKFDh2rLli2aN2+eUlJSFBIScsNTXc1ta24O6Uq48fnnn+vYsWNmtw0KClLr1q31ySefaO7cuUpMTNTDDz8s6cq13UaPHq09e/Zo9uzZWrNmjfr376/mzZuXmAPmDRs2TBs3btS6deuUlpamWbNm6cMPP7zh9RD79eun6OhoLV26VJmZmdq+fbsCAgIUExOj1NRUeXh4qFGjRkpNTdVbb72l3Nxcpaammh5/9Ygrc+OsX79eAwcOND1mw4YNuv/++xUbG6saNWrI09NTNWrUkKSbzjlw4EDFxMToww8/1NixY8v0PK6f/1pDhgzRnj17tGzZMmVmZmr16tXauHGjhg0bZraWkSNHatWqVfruu++UkZGhf/3rX4qJibmj18PcfDfaZ+XF3HvtZr3gdnh5eWn48OEKDw/X3LlzFR8fr8cee6xE6HW7vcXcurK6WX+6GaPBScb0VLk4VldAgL8kqaiwQNKVI4mHDBogzxu8P21No0aNVFhYqPnz52vNmjXq1atXiSOs27dvr4YNG+qDDz7Q0qVL5eHhoZ49e5p+hpycnExHTbdo0cL0uDp16mjo0KHatGmT5s+fr7y8PLVq1eqGR4R7enpq5MiR2rVrl+bMmaPIyEgNGzZMdevWlbu7u3r37q1169Zp/vz5kq78jLi4uJh9bg0aNNDDDz+s8PBwvf/++0pJSVFwcHCpl45wdHTUo48+qvDwcB06dEh9+vTRs88+q7CwMPn6+mrlypXKzs5WSEjILe1jAAAAALBFBuN/z+GLiIhQp06drF1PlfPxkk/l4OCgof/ziKpVK/0/uKmpqdr4ww9ydHTSpLAnKrDCO7d//34FBweXefsffvhB06ZNMwXWCxYsUFBQkM6cOaOmTZsWO1Jq165deuGFFxQVFaUmTZrojTfe0OOPP67s7GyNHj1amzdvVuPGjfXss8/qjz/+0IYNG7R//35NnjxZS5cu1apVqxQaGnrDcUaOHKmhQ4cWC89yc3P11FNPad26dcrNzVXnzp31z3/+U4GBgTedU5JGjBihn3/+WSdOnCgWoJV1/sTERLVr184U7G7btk0vv/yyDh8+LF9fX82cOVMDBw68aS0rV67UX/7yF505c0YDBw5Uu3bt5OPjo0cffbTMr8e1tZib79dffy2xz7Zt21amnwX6TuXgcnyFimq2UkGdIEXHxGjHzl/VuHEj9Q7+k9yqV7d2eTe0f//+Mv/sxMXF6T//+Y8mTJhwx/OuW7dOLi4u6tu3r2nZpUuXdPHixWKB9IIFC/TII4+YLhNSHq6egl/Wy0Rcr6CgQPv27VN0dLTpCOv69eurc+fOJYLjquRWP7MAAAAAVG2Eo+Xs4O/RiomNVVGR8abbOjg4KMCvjQI7tL/ptpUJ/9HE7aLvVA6OF2NkdG2gItcrN7rLysoyHTFdWVV0OJqfn6+0tDR99tlnGjt2rBo2bGha98cff2jx4sUaMWKEmjVrpgMHDui3337TM888c9vBJcoPn1kAAAAArnXHN2SCeYHt2ymwfTtrlwEApSqsFVDs+8oejFpDYmKivv32W913333FglHpynWFBw8erM2bNysnJ0d169bV8OHDCUYBAAAAwAYQjgIAqrTWrVurdevWdzTGXXfdpalTp5a6PjAwUIGBtn/jKgAAAACwN3d8QyYAAAAAAAAAsEWEowAAAAAAAADsEuEoAAAAAAAAALtU7G71wO3izr+4HfQdAAAAAABgTcVuyETABaAi0XMAAAAAAIA1cVo9AAAAAAAAALtEOAoAAAAAAADALhGOAgAAAAAAALBLhKMAAAAAAAAA7BLhKAAAAAAAAAC7RDgKAAAAAAAAwC4RjgIAAAAAAACwS4SjAAAAAAAAAOwS4SgAAAAAAAAAu0Q4CgAAAAAAAMAuEY4CAAAAAAAAsEuEowAAAAAAAADsEuEoAAAAAAAAALtEOAoAAAAAAADALhGOAgAAAAAAALBLhKMAAAAAAAAA7BLhKAAAAAAAAAC75GTpAX0/yrT0kLAjx592t3YJAAAAAAAAsBMcOQoAAAAAAADALln8yNElbXYrODjY0sPCDkREREjiZwcAAAAAAAAVgyNHAQAAAAAAANglwlEAAAAAAAAAdolwFAAAAAAAAIBdIhwFAAAAAAAAYJcIRwEAAAAAAADYJcJRAAAAAAAAAHbJquHotm3b1L59e2uWAAAAAAAAAMBOceQoAAAAAAAAALtk8+HomTNnlJeXZ+0yAAAAAAAAANiYShOO5uTkqHfv3powYYKMRqMkKTY2VsHBwfLw8JCfn5+++eYbSVcC0WbNmundd99V48aNFR8fr9WrV8vf31+urq7y8fHRwoULTWPPmDFDTZo0kbu7u/r166czZ85Y5TkCAAAAAAAAqDwqRTial5enhx9+WE2aNNGiRYtkMBiUnZ2tPn36aNCgQTp58qTmzp2rJ598Urt375Z0JSCNjY1VSkqKWrZsqdGjR+uVV15RSkqK5syZo+eee07p6emKiIjQvHnztHbtWh0+fFi5ubmaNWuWlZ8xAAAAAAAAAGtzsnYBeXl5GjZsmA4cOKCkpCQ5OFzJazdt2qSGDRvqxRdflCQNGDBAo0aN0pYtW/Tkk08qPz9fs2fPVoMGDZSXl6c9e/YoICBAktSiRQsZDAalpaXp8uXLMhgMSk9PV6dOnbRp0yYVFhZa7fkCAAAAAAAAqBysfuRoXFyc6tWrJ4PBoM8//9y0PCEhQYcOHVKzZs1MXxs2bFB6erokycHBQQ0aNJAkubi4aOvWrerVq5eCgoI0e/ZsU8g6YMAAzZw5U5MnT5aXl5cmTZqkxMTEin+iAAAAAAAAACoVqx852rJlSy1evFhff/21nn/+eT300EPy8vJSw4YN1bVrV23dutW07enTp+Xi4qKCggJT+ClJ69ev19tvv62tW7fK399fhYWFqlGjhiQpKSlJoaGhmjRpkpKTk/WPf/xDQ4YMUVxcXEU/VQAAAAAAAACViNWPHK1evboMBoNGjBihDh066LXXXpMk9evXT9HR0Vq6dKkyMzO1fft2BQQEKCYmpsQYqamp8vDwUKNGjZSamqq33npLubm5Sk1N1YYNG3T//fcrNjZWNWrUkKenpyk4BQAAAAAAAGC/rB6OXmvBggX6/PPPtWvXLnl5eWnDhg36+OOP1aBBAz355JOaP3++evToUeJxI0aMUGBgoLy9vdWzZ081bNhQ06dP18SJE/XEE0+of//+6t69u5o1a6YdO3Zo6dKlFf/kAAAAAAAAAFQqBqPRaJSkiIgIBQcH3/GAlhoH9oefHQAAAAAAAFSkSnXkKAAAAAAAAABUFMJRAAAAAAAAAHaJcBQAAAAAAACAXSIcBQAAAAAAAGCXCEcBAAAAAAAA2CUnaxcAXCsiIsLaJQAAAAAAAFRqwcHB1i6hyiAcRaXBGxsAAAAAAMA8DiyzLE6rBwAAAAAAAGCXCEcBAAAAAAAA2CXCUQAAAAAAAAB2iXAUAAAAAAAAgF0iHAUAAAAAAABglwhHAQAAAAAAANglwlEAAAAAAAAAdolwFAAAAAAAAIBdIhwFAAAAAAAAYJcIRwEAAAAAAADYJcJRAAAAAAAAAHaJcBQAAAAAAACAXSIcBQAAAAAAAGCXCEcBAAAAAAAA2CXCUQAAAAAAAAB2iXAUAAAAAAAAgF0iHAUAAAAAAABglwhHAQAAAAAAANglwlEAAAAAAAAAdolwFAAAAAAAAIBdIhwFAAAAAAAAYJcIRwEAAAAAAADYJSdrFwBYSk5OjpKSkpSWlqbMzExJkru7u2rXrq1mzZrJ1dXVyhUCAAAAAACgMiEcRZUQHx+vU6dOydvbWwEBAXJ3d5ckZWZm6uzZs9q1a5caN26sVq1aWblSAAAAAAAAVBacVg+bt3v3buXn56tbt25q0aKFKRiVrhw52qJFC3Xr1k35+fnavXu3FSsFAAAAAABAZVLlwtFdu3apf//+1i7jhlJSUtS+fXtrl1GlxMfHy93dXQEBAXJyKv1AaCcnJwUEBMjV1VXx8fEVWCEAAAAAAAAqqyoXjsJ+5OTk6NSpU2rdunWZHxMQEKBTp04pJyenHCtDecvPz9dDDz2kb775xtqlAAAAAAAAG0Y4egPnzp1TXl6etcvATSQlJcnb29vsEaPXc3Jykre3t5KSksqxMpS3efPmKSoqytplAAAAAAAAG1dpb8h07tw5DRgwQO+8845WrVqlAwcOqHfv3poxY4YMBkOZxsjNzdWECRPUpEkTzZo1SwaDQUePHtVbb72lqKgoNWjQQC+++KIGDBigc+fOaciQIRozZozmzJmjH374QQkJCZozZ46Sk5NVt25dPfXUUxo9erQk6V//+peWL1+ujIwM3X333ZozZ47q1q17S8/xxx9/LHV8a4mIiLDq/LfCzc1N7dq1u+XH1atXT9HR0QSkkoKDg61dwi3buXOntm3bpo4dO1q7FAAAAAAAYOMq9ZGj6enpOnz4sP73f/9XP/74o9asWVPmG+rk5eVp0qRJatCggWbOnCmDwaCcnByNHTtWISEh2rFjh6ZPn65p06bp4MGDkq4EskePHtUvv/yi5s2b68UXX1RYWJh+/vlnvf766/rrX/+qjIwM/fzzz/rkk0+0aNEibd68WXl5efroo49u6bnl5uaWOj7K5vLly8VuvlRW7u7uunz5cjlUhPJ24cIFvf7663r//ffl6upq7XIAAAAAAICNq7RHjkpXAsRRo0ZJkmrWrKl27dopLS3tpo/Ly8vTBFHMAwAAIABJREFUs88+q5iYGO3cuVMODlcy4G3btqlevXp64oknJEm9evXSkCFDtHPnTg0fPlz5+fmaNm2a6tatq7y8PK1Zs0atWrWSJDVt2lQGg0Hp6enKycmRwWBQZmam2rZtq6VLl6qwsPCWnpvBYCh1fA8Pj1saC7AXU6ZM0YQJE9SmTRtrlwIAAAAAAKqASh2OOjo6qnbt2qbvy3o6fUJCgrp06SKDwaDvvvtOw4YNkyQlJycrLi5O3bp1K7b9kCFDJEkODg6mU+NdXFy0a9cuvfXWW8rIyJCPj48pZO3Vq5deeeUVzZgxQ6dOnVKfPn00ceJE3XXXXWV+bubGR9m4ubkpMzPzlo8ezczMlJubG0eP2pgvv/xSRUVFGjFihAoLC2U0GlVYWKiioiLeOwAAAAAA4LZU6nC0rGHo9Zo3b65Zs2Zp3bp1+tvf/qbQ0FDVqlVL9erVU2BgoL766ivTtqmpqXJ2dlZhYWGxgCUiIkL//Oc/tXz5ct11110qLCw0Xd/y9OnT6t69u0aNGqXTp0/r448/1lNPPaXw8PAy12hufGuypWtQxsfH6+zZs7ccjp49e1Z16tTRvffeW06VoTzs3btXW7duVevWrU3Ldu3apXXr1mn58uVWrAwAAAAAANiqKnm4laurqwwGgwYPHiw/Pz+99957kqSePXsqLi5Oq1atUlZWln777Tf16dNHR48eLTHG+fPnVaNGDdWvX1/nz5/XBx98oNzcXJ0/f16RkZEaNmyYjh49Kjc3N7m7u6t69eq3VKO58SXpyJEjioyMNG2fnp6u9evXF7smaWRkpOLi4m5nF1UJzZo108mTJ1VQUFDmxxQUFCgpKUnNmjUrx8pQHubNm6fjx4+bvu699169++67BKMAAAAAAOC2Vclw9Fpvv/22Vq9erQMHDqhWrVr65JNP9NVXX6lr166aNm2a3nzzTXXp0qXE4wYNGiR/f391795djz76qOrVq6dnnnlGr7/+uoYNG6YHHnhAw4cPV7du3bRnzx5TAFtW5saXpA0bNmjmzJmm7VNSUvT888/r9OnTpmUzZszQxo0bb3PP2D5XV1c1btxY8fHxZX5MTEyMGjVqxM18AAAAAAAAIIPRaDRKV07ztsQp1ZYa50b+W+ptn26Pqmn37t3y8PBQq1at5OR04ytFFBQUKCYmRjk5OeratWsFVwgAAAAAAGAZ5Zm92aNKfc3R6xkMBsXExGjJkiWlbtO3b1/17du3Aqv6f5W5tqqsa9euio+P186dO+Xt7a169eqZrkOamZmps2fPKikpSY0aNVKHDh2sXC0AAAAAAAAqC5s6chQwJycnR0lJSUpLS1NmZqYkyd3dXbVr11azZs04lR4AAAAAANg8sjfLsqkjRwFzXF1d1apVK2uXAQAAAAAAABtR5W/IBAAAAAAAAAA3QjgKAAAAAAAAwC4RjgIAAAAAAACwS4SjAAAAAAAAAOwS4SgAAAAAAAAAu0Q4CgAAAAAAAMAuEY4CAAAAAAAAsEuEowAAAAAAAADsEuEoAAAAAAAAALtEOAoAAAAAAADALjlZuwBYR0REhLVLAKqM4OBga5dgV+hfgOXQvyoW/QuwHPpXxaJ/AZZD/6p8CEftGG9I4M7xi6J10L+AO0f/sg76F3Dn6F/WQf8C7hz9q3LitHoAAAAAAAAAdolwFAAAAAAAAIBdIhwFAAAAAAAAYJcIRwEAAAAAAADYJcJRAHZt2efLtezz5dYuAwBuGf0LgK2ifwGwVfSvqom71QOwWx8v/lTr1m+UJOXk5uipsCesXBEAlA39C4Cton8BsFX0r6qLcBSAXVq0+DOtW79R89+fLUma/OIUORgcFDZhnHULA4CboH8BsFX0LwC2iv5VtXFavY1LSUnRmDFjFBgYqKFDhyomJsai47/00kvq2rWrgoODTV9nz56VJOXm5mratGkKDg7WkCFDFBkZaXrc7a4DKkpScrLmvz9bLX1bqKVvC81/f7ZOJiVZuyy79Nlnn2n+/PkWH3f16tV66KGH1LNnT4WEhOj77783rSuP3gZUFPqX9e3du1dDhw5Vu3btFBISoi+//NKi4588eVLTp09Xhw4dNHny5GLr+P0Ltoz+ZX3h4eHq37+/2rZtqwcffNDivYBeg6qK/lXFGf8rPDzcaAmWGgdl079/f+OsWbOM586dM3766afGu+++25ibm3vTx5X1dZo4cWKp27700kvG2bNnGwsKCoxxcXHGLl26GI8dO3ZH6wBbQ8+7fVFRUca5c+caAwMDjR988EGZH1fWfb5mzRrj2bNnjUaj0RgbG2u85557jNHR0UajsXx6G2Br6F+3JyMjw9ilSxfjokWLjOnp6caff/7Z2LFjR+PmzZtv+tiy7vOjR48af/rpJ+Onn35qfP7554ut4/cvgP51u1JTU41t27Y1btiwwXj+/HnjRx99ZAwICDCeO3fupo8t6z6n1wDmkb1VThw5asOioqKUnJysl156SXXq1NH48ePl4eFh0b/OZWVlydPTs8Tyc+fOKTw8XJMnT5ajo6NatWqlhx56SN9+++1trwNgX6KiopSWlqY6deqUy/iDBw9W3bp1JUl+fn7q0qWL9u3bJ8nyvQ2A/YiOjpajo6PCwsLk4eGh++67TwMGDNCOHTssNkfLli0VGhoqJ6fiV8Di9y8Ad2L//v3y9vbWgw8+KC8vL4WFhcnZ2VnHjx+3yPj0GgC2inDUhiUkJKhly5ZycXExLfP391dCQoLF5sjMzNTy5cs1dOhQPfTQQ/rhhx8kSTExMWrTpk2xuTt27KhDhw7d9joA9mXUqFGaMWOGAgMDy32uoqIixcTEqGXLlpIs39sA2I/27dtr9erVpu/z8vK0b98+dezYsdzn5vcvAHciNDRU69evlySdPn1aGzZskIODg/z8/CwyPr0GgK3ihkw27OLFi3J3dy+2zMPDQ+fPn7fYHMOHD1eNGjX05ptv6sSJE5o4caJatGih8+fPy8PDo9i2np6eSktLu+11AFBePv/8czVq1Ej333+/JMv3NgD2o0aNGqpRo4YkKT4+XtOnT1ezZs3Ur1+/cp+b378A3AmDwSCDwaCTJ09q4MCBcnNz0+uvv16iP9wueg0AW0U4asNq1aqlrKysYssyMjLk7e1tsTlGjhxp+nfHjh3Vr18/RUREqF27dsrIyCgxt5eXl7y8vG5rHQCUh507d+rf//63Vq5caVpm6d4GwL5kZ2dr7ty5Wrt2rSZNmqTx48fLYDCU+7y3+zsW/QvAtby9vRUVFSWj0ahHH31UTZs21b333nvH49JrANgqTqu3YT4+Pjp+/Ljy8/NNy+Lj4+Xj41Nuc2ZmZsrT01P+/v6Ki4tTXl6eaV1UVJQCAgJuex0AWNqRI0c0ZcoUffTRR2Z/Mb/T3gbAfhQVFSksLEwpKSnavHmznnjiiQoJRiXx+xeAO7JgwQJNnTrV9L3BYFDDhg0VFRVlkfHpNQBsFeGoDQsMDFTjxo21ZMkSFRQUaM2aNUpLS1NISIjF5lixYoUuX74s6coH244dOxQSEqL69evrT3/6kxYuXKiioiKdOHFCa9as0fDhw297HQBY0pkzZxQWFqa///3vat26dbF1lu5tAOzHtm3bdOTIEU2dOlWZmZlKSUlRSkqKLl68WO5z8/sXgDvRvXt3bdy4Ubt27VJOTo62b9+uyMhI3XPPPRYZn14DwFYRjtq4RYsWaceOHerSpYuWLl2qZcuWqVq1ahYZu6CgQMeOHdOAAQMUEhKiGTNmaOHChWrcuLEkadasWTp16pRCQ0P1wgsv6N1331Xz5s3vaB1QUf7y1gwdO/7/Ny87djxBf3lrhhUrgqVNnz5dZ8+e1cyZMxUaGqrQ0FB99NFH5dbbgIpC/7KuqKgonT9/XsHBwerRo4fp64MPPrDYHGFhYQoNDdW//vUvRUZGKjQ0VGFhYZL4/Qu2jf5lXYGBgXrjjTc0bdo0BQUFadasWZo9e7ZFb45Jr0FVRf+q2gxGo9EoSREREQoODr7jAS01DsoXrxPs3aLFn2nt+h80//3ZkqTJL07RkEEDFDZh3C2Nw3up4rHPYe/oX7aLfQ57R/+yXexz2LvK1r94T1oWN2QCYJcmho2XUUZNfnGKJGnQwP63/MEGANZA/wJgq+hfAGwV/atqIxwFYLeeCntCrtVcJUljHx95k60BoPKgfwGwVfQvALaK/lV1EY4CsGt8qAGwVfQvALaK/gXAVtG/qiZuyAQAAAAAAADALhGOAgAAAAAAALBLhKMAAAAAAAAA7BLhKAAAAAAAAAC7RDgKAAAAAAAAwC5xt3o7FhERYe0SAOC20L8A2Cr6FwBbRf8CUFURjtqp4OBga5cAALeF/gXAVtG/ANgq+heAqozT6gEAAAAAAADYJcJRAAAAAAAAAHaJcBQAAAAAAACAXSIcBQAAAAAAAGCXuCGTneJOg4DlcIH6ikX/AiyH/lWx6F+A5dC/Khb9C7Ac+lflQzhqx3hDAneOXxStg/4F3Dn6l3XQv4A7R/+yDvoXcOfoX5UTp9UDAAAAAAAAsEuEowAAAAAAAADsEuEoAAAAAAAAALtEOAoAAAAAAADALhGOAgAAAAAAALBLhKMAAAAAAAAA7BLhKMpNWlqafH19rV0GANwy+hcAW0X/AgAAuDWEozYuJSVFY8aMUWBgoIYOHaqYmBhrlwQAt+Szzz7T/PnzrV0GAJTZ3r17NXToULVr104hISH68ssvrV0SAJRJeHi4+vfvr7Zt2+rBBx9UZGRkuc53zz33KDg4WLm5ucWWT5kyRUuWLCnXuQGgrAhHbVxYWJjatm2riIgIDRo0SGPHjlVeXp61ywJsyokTJyyyDW7N77//rnnz5hGMAneA/lXxMjMz9fTTT6t///765Zdf9Pe//11z5szRli1brF0aYFPoXxXv7Nmzmjx5sp577jlt375dgwcP1rPPPqvz58+X67z5+fn68MMPy3UOoCLRv6oewlEbFhUVpeTkZL300kuqU6eOxo8fLw8Pj3L7619aWpqGDh2q+fPnq1u3bvrss88kSUeOHNHw4cP1wAMPaMSIEUpMTCzx2N9//109e/Y0fV9YWChfX19lZGSUS61AWV28eFGdOnXS+PHjS91m3Lhx6tSpkw4cOFCBlVV9UVFRSktLU506dcp9LvoXqiL6l3VER0fL0dFRYWFh8vDw0H333acBAwZox44d5TIf/QtVEf3LOvbv3y9vb289+OCD8vLyUlhYmJydnXX8+PFynXf69OlaunSp4uLiSt0mISFBjz/+uHr27KnQ0FB9/fXXkq70wF69emnKlCnq0aOHgoKC9PHHH5seV5ZeCFgS/atqIhy1YQkJCWrZsqVcXFxMy/z9/ZWQkFBucx44cEDu7u6KjIzUuHHjlJ+frwkTJujJJ5/Utm3bFBoaqunTp5fb/ICl1apVS0OGDNHSpUtv+AE3btw4LVu2TM2bN5ePj0/FF1iFjRo1SjNmzFBgYGCFzEf/QlVD/7KO9u3ba/Xq1abv8/LytG/fPnXs2LHc5qR/oaqhf1lHaGio1q9fL0k6ffq0NmzYIAcHB/n5+ZXrvL6+vpowYYLeeOMNGY3GEuvz8/P15JNPqm/fvvrPf/6jxYsXa968efr1118lSYmJiRo1apQiIyP12Wefac6cObp06RK9EFZB/6qaCEdt2MWLF+Xu7l5smYeHR7mfFvHkk0/KxcVFBoNBe/fuVVFRkfr06SNJ6tOnj3777TcVFhaWaw2AJX3wwQcKDAws8QF39YMtMDBQW7duVa1ataxYJSyB/oWqhv5V8WrUqKFGjRpJkuLj4zV69Gg1a9ZM/fr1K9d56V+oauhfFc9gMMjBwUEnT55U3759NXPmTL3++uvy8PAo97mffvppXbhwwXRE6LX27dsng8GgUaNGSZJ8fHz02GOPmYJcSQoMDJSTk5MCAwNVo0YNnT59ml4Iq6F/VT1O1i4At69WrVrKysoqtiwjI0Pe3t4VVsOZM2d08eJFhYSEmJY1adJE2dnZFVYDcKdq1aqlrVu36k9/+pOWLl1qWn71L358sFVN9C9UBfQv68jOztbcuXO1du1aTZo0SePHj5fBYKiw+elfqAroX9bj7e2tqKgoGY1GPfroo2ratKnuvffecp3TxcVF77zzjiZNmqTevXsXW/fHH3+U+D+st7e3YmNjbzhWtWrVZDQazfbC6w8iAiyJ/lX1EI7aMB8fHx0/flz5+flydnaWdOUIhgEDBlRYDfXq1VOrVq30/fffl1iXlpZm+rfBYOAXdlRqN/qAq1mzpr7//ns+2Koo+heqCvpXxSoqKjJdb3Tz5s2qWbNmhddA/0JVQf+qWAsWLNDJkyf1j3/8Q9KVHtGwYUNFRUWVezgqXblzfZ8+fTRjxgy5urqaltevX1/JycnFtk1JSTEdpV8ac70QKG/0r6qF0+ptWGBgoBo3bqwlS5aooKBAa9asUVpaWrG/nJW3Tp06KTU1VT/99JNp2cmTJyVd+Yues7Ozzp49q0aNGiktLU0xMTHKy8szXT8GqEyufsAFBgaqZs2a2rp1a7leQw7WRf9CVUL/qjjbtm3TkSNHNHXqVGVmZiolJUUpKSm6ePFihdVA/0JVQv+qON27d9fGjRu1a9cu5eTkaPv27YqMjNQ999xTYTW89tpr+uWXX7Rr1y7Tss6dOysnJ0erVq2SdCUYXbFihYYOHWp2LHO9EKgI9K+qg3DUxi1atEg7duxQly5dtHTpUi1btkzVqlWrsPmrV6+uxYsXa8mSJerVq5d69eqlBQsWSJLc3Nz05z//WQ899JAyMjL06quvavTo0QoJCdH69evVtGnTCqsTKKurH3B8sFV99C9UNfSvihEVFaXz588rODhYPXr0MH198MEHFVYD/QtVDf2rYgQGBuqNN97QtGnTFBQUpFmzZmn27NkVdnNM6cprPX36dCUlJZmWubi46JNPPtG3336rXr16acKECZo2bdpNfxbM9UKgotC/qgaD8b+3i4uIiFBwcPAdD2ipcVC+eJ0Ay+C9VPHY54Bl8F6qeOxzwDJ4L1U89jlgGWRvlRNHjgIAAAAAAACwS4SjAAAAAAAAAOwS4SgAAAAAAAAAu0Q4CgAAAAAAAMAuEY4CAAAAAAAAsEuEowAAAAAAAADsEuEoAAAAAAAAALtEOAoAAAAAAADALjlZuwBYT0REhLVLAIDbQv8CYKvoXwBsFf0LQFVFOGqngoODrV0CANwW+hcAW0X/AmCr6F8AqjJOqwcAAAAAAABglwhHAQAAAAAAANglwlEAAAAAAAAAdolwFAAAAAAAAIBdIhwFAAAAAAAAYJcIRwEAAAAAAADYJcJRAAAAAAAAAHaJcBQAAAAAAACAXSIcBQAAAAAAAGCXCEcBAAAAAAAA2CXCUQAAAAAAAAB2iXAUAAAAAAAAgF0iHAUAAAAAAABglwhHAQAAAAAAANglwlEAAAAAAAAAdsmpPAaNiIgoj2EBAAAAAAAAwGIsHo4GBwdbekgAAAAAAAAAsDhOqwcAAAAAAABglwhHAQAAAAAAANglwlEAAAAAAAAAdolwFAAAAAAAAIBdIhwFAAAAAAAAYJcIRwEAAAAAAADYJcJRAAAAAAAAAHaJcBQAAAAAAACAXSIcBQAAAAAAAGCXCEcBAAAAAAAA2CXCUQAAAAAAAAB2iXAUAAAAAAAAgF0qt3B09+7dGjNmjO6++2517dpVTz31lA4fPlxe09kcc/snMTFRvr6+mjRpUrnN/8cff8jX11cTJkwwLVu7dq2CgoLk7+8vSRo4cKB8fX1VWFhYbnUAAAAAAAAA1lIu4eiGDRs0cuRI/frrr/Lx8VGtWrW0efNmDR8+XMeOHbP4fHfffbe6detm8XHLS0XvnxtxcnJSmzZt1LhxY9OyTz/9VBcvXtQLL7wgSfL29labNm0qpB4AAAAAAACgolk8HM3NzdWsWbNUVFSkjz/+WKtWrdJPP/2k559/XpmZmVq8eLGlp7QplWX/1K1bVxs3btTf/vY307KsrCw5OjrqqaeekiQtXLhQGzdulKOjY4XUBAAAAAAAAFQki4ejhw4d0qlTp9S1a1cFBwebloeFhentt99Wjx49FBsbK19fX7366qt67733NGbMGEnStm3bNHjwYLVv316DBg1SeHi46fHLli1Tr1695O/vrx49emjBggWSpP79++vChQs6ffq07rnnnpuOc62jR4+qZ8+eevvttzV16lR16tRJISEh2rZtm2mb0sYaMmSI7rrrLmVkZCg/P1/+/v5q2bKl0tPTlZGRobvuukuDBg26rf1zI6U9f0latWqVevfuLX9/fwUHB2vJkiU3XXf9qfv9+/fXsWPHVFhYKF9fX2VkZJQ4rb60fVHa6zl27Fh169ZNycnJN3xOAAAAAAAAgDVZPBxNTEyUJPn5+RVbXqNGDY0ZM0YDBgwwLdu2bZu++uortWjRQnFxcZowYYIuX76sCRMmKC8vTxMnTtTvv/+u3bt36+2335aLi4vGjRsnNzc3zZ07V7t27dL7778vT09P1alTR5988onZca6Xn5+v5ORk/fvf/9bx48cVFBSkhIQEvfrqq5JkdqyePXuqqKhI+/fv1+HDh5Wbmyuj0aioqCj9/vvvKioqUs+ePe9o/1xl7vnHxMRo6tSpcnR01NNPP62aNWtq5syZ+vHHH82uu97777+vJk2ayNHRUatXr5abm1ux9WXZr9e+npJ0/vx5nT59mmuWAgAAAAAAoFJysvSARqNRksp0Krazs7N27twpZ2dnvfPOOyosLNTo0aPVuXNn1atXT2+88Ya++OILTZgwQe+99546d+4so9Go8+fPKz4+XomJiRo+fLicnJzk4uKidu3amR3nH//4xw3raNeunVauXClJ6tu3r+Lj43XhwgWtXLmy1LGGDRumBQsWaO/evapXr57q1KkjZ2dnHTx4UAaDQZL0pz/96Y72z1W1a9cu9fnXrl1bRqNRNWrUUJs2bTRgwADFx8erRYsWSkxMLHXd9fz8/OTq6iqDwaCOHTuWWG9uX4wbN67E6ylJ69evL/NzBAAAAAAAACqaxcPR5s2bS7pyyvq18vLy9OWXX6pWrVqmoybvueceU5B29dTra6+BKUknTpyQl5eXDh48qLfffltFRUWqV69eqfObG6c0rVu3Nv27fv36io+PV1FRkdmxOnbsKE9PT+3bt0/169dXUFCQHB0ddeDAAUmSh4eHOnXqVGKusuyfoKCgYuvMPf9u3bpp9OjRWrlypZ5++mk5Ojqqe/fu+utf/2p23a0qy3699vUEAAAAAAAAKjuLn1bftm1bNWjQQDt37tSePXtMyzdt2qQZM2YUu/6nk9P/Z7NXA78vvvhCx48f1/Hjx3X48GF98cUXWrRokb744gu98sorOnjwoEJCQkrMe/WITHPj3CpzYzk5Oen+++/XgQMHtG/fPnXq1EmBgYE6ePCgDh48qG7duhV7frezf64y9/xTUlLUu3dvhYeH66OPPlKPHj20bds2ffDBB2bXWXJfXHWj5wsAAAAAAABUVhYPR11dXfXaa69JksaMGaMnnnhC48eP18svvywXFxeFhYXd8HFDhgyRwWDQ3//+dy1evFjPPvus/Pz8tGzZMlPwGR4erjfffFNff/21JJmuZenk5KQLFy4oIiLC7Di36mZjPfDAA8rKylJiYqI6d+6sjh076ty5czp37pweeOABi+0fc89/9+7dGjdunF544QXTafaS5O7ubnadpffFjXBDJgAAAAAAAFRmFg9HJWnw4MH6/PPPFRQUpP379+vQoUPq3r27vvrqKwUGBt7wMZ07d9Y///lPGY1Gzfs/9u48rqpq///4G0FUwMxyNi1QSxGi9DrklCIahIFXv1qmlpZm1r2ZpmnlLaduj8rMhptWNjiUfktvV0NNTZRyHn7mWCGR5hygCDiAyPr94fV8ATmbc+AweV7Px4NHnb3PWeuz1l577XM+7mHGDO3fv1/jx4/X448/rqFDh6pVq1bauXOn/vjjD7388suqX7++7WFAjz76qG688Ua9/PLLluU4q7Cyrj5wycvLS8HBwQoODrbdS7SghzEVtX+s2j9gwAA988wzOnXqlN5++2398MMPio6O1tixYy3XubovCsIDmQAAAAAAAFCeeZj/npYYGxur0NDQso4HAAAAAAAAAEpFiZw5CgAAAAAAAADlHclRAAAAAAAAAG6J5CgAAAAAAAAAt0RyFAAAAAAAAIBbIjkKAAAAAAAAwC2RHAUAAAAAAADglkiOAgAAAAAAAHBLJEcBAAAAAAAAuCWSowAAAAAAAADcEslRAAAAAAAAAG6J5CgAAAAAAAAAt0RyFAAAAAAAAIBbIjkKAAAAAAAAwC2RHAUAAAAAAADglkiOAgAAAAAAAHBLJEcBAAAAAAAAuCWSowAAAAAAAADcEsnRcmLr1q2KiIgo6zAccuzYMQUHB5dafTExMerQoYPWrFlTanUWJjk5WbfffrtD7y3t/rqelLdtv2bNGj344INlHUaBJk2apICAAK1atSrPcsZf+Zd7G1ltr/I8/opr1KhRmj9/fomVX1L7QVHjLun2FpUr+6ks2jhs2DC98847ttfOHkPK63YpC67sC6u5qzhjztEYBw8erICAgAL/hg0bJknq0aNHgetnzJgh6dqxlfv9TZo0Uffu3fX666/r0qVLBb6noDIrEvYNAMD1rsImR9u1a6f4+PiyDgMlYMeOHQoICNCxY8ckSZ988olefPFF9ejRQ2PGjNFTTz1VJnG445jL3eb8/VEacm972JeTk6OVK1eqYcOGWrFiRVmHUyqKuj+WxTguj3FU9P7D9aMkj61lcQzZuXOn+vTpo6CgIHXv3l1ffPFFqdVdEvvn9bDPf/DBB9q5c6d27typ3r17q2/fvrbXb7/9tu19kyZNsi2/+jdy5Ei75U6dOlVpBmPQAAAgAElEQVQ7d+7Uli1bNGnSJMXGxmrUqFF53uNomddDPwMAUJF5lVXFycnJuuGGG+Tt7V1WIVx3rpc+9ff319tvv62aNWtKks6ePatbbrlFkvT3v/9djRs3dqq8ovZL/jgqgpIcA2XRH7m3fW7Xy1h3lW3btunChQuaPHmyxo0bp8zMTFWpUqWswyp3kpOT1bBhQ5eOY1fML2fOnHFJLK5QWHsq4rwI92XvGFJSMjIyNHLkSA0fPlxz587Vvn379NRTT6lu3boKCwsr8fpLYv+8Hvb56tWr2/6/SpUq8vLyKrA9vr6+TrXTz8/P9v7atWvr888/V9euXbVv3z4FBQU5Veb10M8AAFRkJXLm6JYtWxQVFaWgoCD17dtXBw4ckHTlR1fHjh01e/ZstW/fXocOHVJCQoIGDhyo4OBghYWFafny5bZyVq1apR49eqhFixbq3LmzFixYIOnK5TEpKSkaNGiQNm3aJEmW5eRnr9z8kpOT1a5dO33//fd68skn1b59e02cOFHGGMs6o6Oj9dVXX0mSDh48qICAAC1cuFCS9Ntvv6lZs2ZKTU21G19mZqYGDx6sCRMmFFpXQX1q1b733ntP99xzj4KCgjRkyBAlJyfbjcPednSmP63qs7fOGKNx48bJx8dHw4cP15EjRzR8+HB9//33+uabb/Taa6/ZynCmX3JzdBtdjaOgMSdJy5cvV1hYmEJCQvKMDSv2+usf//iHpk2bZnvf4cOH1aJFC6WnpxerrdKVMxJ69+6t4OBg9erVS+vXr5ckrVu3TiNGjNCMGTMUFRWlHj162M5yyd/m3NvFqszC9htHx2D+bV9QO+3FcObMGbVr105fffWVhgwZos6dO2vFihV65513NGDAAN1zzz2Wl1lajemcnBy9+eabatu2rbp06aI5c+YU2s/5L8dLSUnR7bffriNHjlhuW0etWLFC3bt3V2hoqDw8PGz1FsSq/+3F72wf5VacttubgwraH63GY+5x88cff9jG8erVqxUcHGz7u+OOO/Twww/b6i+t+cUZVuPP0XGUv/+cOY7kngeKeoyUrlzuHBYWprvvvlvPP/+8Ll68aLfNRZ1rrFiNdauxbRW3q9prtY/ai82ZOc9q3s/PmbnJqo3OfJ9zdG4pTP5jiLMxW40Re3PTvn375OnpqeHDh6t69eq65557FBkZqQ0bNpTIXJhf/uO0M9/57PWFM2UW9diZm6v2o7JQv359tWvXThs3bnT6s0XddomJifrLX/6SZ1lERIRtf7IqpyL3NQAALmf+a+3atcYVTpw4YYKDg82KFStMUlKSmTZtmgkLCzPGGJOUlGRuv/1289xzz5k///zTXLhwwXTo0MF88sknJjU11cTGxpqgoCDz008/mYsXL5oWLVqY//3f/zWnT582K1asME2aNDFpaWnGGGPatm1rfv31V2OMsSwnv8LKzS0pKck0b97cvPfee8YYY1JTU01QUJDZsmWLZZ1vvfWWefbZZ40xxnzxxRfmzjvvNKNGjbK97tev3zV1bdmyxYSHh5vMzEwzZMgQ89xzz5nLly8X2r78fWrVvo0bN5qQkBCzZ88ec/z4cTNgwAAzZcoUp7fj0aNHTVBQUKH9aVWf1bqkpCTTrFkzWyzdunUzu3btMsYY89Zbb5mpU6c63S/5ObKN8seRe8wlJSUZf39/M3bsWHP+/Hnz008/maZNm5rt27dfU5ej/bVhwwbTqVMn2+f+9a9/maeeeqrYbU1KSjLBwcFm4cKFJjU11SxfvtwEBgaaw4cPm9jYWNO8eXMzd+5cY4wxx44dMx06dDAbN24ssM1X+8OqTKv9xpkxmH/b52+nVQynT582TZs2NZ9//rkxxpjFixcbf39/8/XXXxtjjPn0009Nr169CqzTahutXr3aNGnSxLz++uvmzJkzJi4uzoSEhJhVq1ZZxvPvf//bRERE2OqYP3++eeihhwrdto7Izs42bdq0Md9//70xxpi//e1vtrFsTN7xV9h+Zy9+Z/oov6K23WoOMubasWk1HvOPm9z79VXJycmmU6dOZsWKFYXG5ur5Jfc2yv3/+VmNP2fHUf7+c/Q4kn8eKMoxcv/+/aZ58+YmJibGnD592syePdv4+/ubefPmXRNnUeea/HL3q1WZVm23ittV7bXaR61ic2bOs5r3c/eTM2PKqo3OfJ9zZm4pyOOPP25mzpxpe537GOJMzFZjxGpuysjIMMePH7fVkZmZaSIiIsw333xTYnNhbrn3T2eOt4X1hSNlFvXY6eiYc2Y/sjJhwgQzceLEa5aHhYUZf3//PH8dOnSwrc8/tsLCwsx//vOfa8p54YUXbOUXVmZuRd12v/32m2ndunWeZeHh4Wbjxo2W5ZRGXwMAUJG4/LL6mJgYtW/f3vZwoTFjxqhBgwbKysqSJF26dEkTJkxQrVq1tGrVKtWuXVuPPfaYJKlbt26Kjo7Wxo0b1aJFCy1dulTNmjWTJN1yyy3y8PBQWlpanstjJCkuLs5uOSEhIXne6+Hh4XC50pWzOAcOHChJqlGjhoKCgnTmzBnLOu+99149/fTTkqTt27frkUce0eLFiyVdufy1a9euBfZdVlaW/va3v+nAgQPauHGjKlWqVGj7+vfvn6dPs7Ky7Lbv4sWL8vDwUEZGhlq2bKnPP/9cly9fLtJ2dKQ/repzJhZ7nOmX/Iq6jfIbO3asqlWrppCQEAUHBxd6eaxVf7Vr107nzp3T/v371bJlS8XExNjuXVWctq5evVqBgYF66KGHJEn333+/li1bpuXLl6t58+by8/PToEGDJEkNGjTQgw8+qCVLlqhDhw5222FVZr9+/ezuN97e3sXa7rnb+eWXX9qN4aGHHtLly5f117/+VZLUunVrValSRQ888IDt9Weffeb0NpKkmjVrasyYMfLy8lKXLl308MMPKyYmRsnJyXbjGTRokF588UUdOnRIt912m5YvX66+fftKcm7+KsjWrVuVlZWlzp07S5J69uypF198scBL6632O6ttmv/+aM7Mo2FhYUVqu7e3t905KP8l34WNx9zjpqCzb7Kzs/X0008rKirKVl95mF8KYm/8GWOKNY4cPY7kH1NFOUZmZGQoPDxckZGRkqQRI0ZoyZIlBcZV1LnGilWZjz/+uN22x8TE2I3bVe212ket9jsfHx+n5jx78/6YMWNs73FmbrLqG2fmC2e/oxWHVcxWY6Ry5cp25yZfX1/5+vpKunLW+EsvvaRGjRopPDxcly5dKvG5MDdnv/M5MkaLOj4l+3NXy5YtbeW7aj8qqkmTJtn2GUm27+HO8PDwkIeHR7HKdMV35MLKKeu+BgCgvHF5cvTYsWN57glZrVo1DR061Pa6UqVKth+TR48eVXx8vDp27JinjOjoaHl7e2vr1q165ZVXlJ6erttuu83uFwqrcvJzplxJ8vT0zHP/n6tfeKzqvOuuu5SZmanExETt2LFD48eP18KFC3Xo0CFt27ZNw4cPL7Cu33//XW3atJGHh4e++eYb9evXz6H25e5Tq/Z169ZNY8eO1dSpU3X8+HH17NlTTzzxhJo2bXpNLIVtR0f606o+Z2Kxx5l+ya+o2yg3T09P1a1b1/a6cuXKhX7Gqr+8vLwUFham1atXq2rVqjp+/Li6detW7LaePHnymvu0Nm7cWCdOnFDz5s1Vt27dPPtAgwYNtHXrVst2WJUp2d9virvdc7ezsBgk6YYbbrDVX6VKFVtiJ/cPl/wKmyMaNmwoL6//mzoDAgK0c+dOy3iqV6+url276rvvvlOfPn20f/9+2yWFzsxfBVm+fLnS0tLUvHnzPMvXr1+v++67L88yq/53pD8d7aPcitr2CxcuODQHSYWPBav9Q5KmTZsmb29vPffcc7ZlZT2/2GNv/BV3HDl6HMmvKMfIkydP6rbbbsuz3N/fv8DyizrXWLEq06rtVnG7qr1W+6gj28XROc+Red+ZMWXVRmfGk7Pf0YrDKmarMeLp6Wk5N124cEFvvfWWli1bpieffFJDhw6Vh4eHqlatWuJzYW7OHG8dHaPFGZ/25q7cXLUfFZWz9xwtyNGjR/P843JRyizud6WcnJxCyynrvgYAoLxxeXK0Tp062rdvn+11dna2Pv30Uw0YMEBS3n8xrV27tkJCQmz3YpOkP//8U5UrV1ZsbKzeffddffnll2ratKkuX75su7l5flbl5OdMuZL9H1pWdXp6eqpTp05asmSJvLy8VK9ePbVt21Zff/21cnJyFBgYWGCZt956q1577TV9++23mjJlinr06KEbb7zRsq7Lly/n6VOr9p04cUKdOnXSwIEDdeLECX344YcaMWKE1q5de00shW3H4tbnTCz2ONMv+TmyjVJSUizrd+RHeH6Fjb/w8HC9+eab8vT0VHh4uO2skOK0tW7dutqxY0eeZUePHrWdrXH8+HFdunTJtr8kJiaqTp06lu0orEx7fVPc7Z67nYXFUFSFbaNjx44pOzvb9iPv999/1y233FJoPL169dKcOXNUrVo19ezZ03Z2kTPzV37Z2dlatWqV/vWvf9nOKpKkJ598UitXrrwmOWrV/870p7PzaFHavmjRIrtzUP4zyAqL3Wr/WLx4sWJjY7Vs2TKHj0+lMb/YY2/8FWccSXL4OJJfUY6Rc+bM0eHDh/O8/+p9F/Mr6lxjxapMq7bXr1/fbtyuaq/VPursfmfFkXnfmTFl1TfOxO3KNhbGKmarMVKpUiW7c5Ovr6/tfqNr1qxRjRo18pRR0nNhbs4cb636wtEyi3rszM1V+1FZOXXqlLZu3apx48YVqxxnvytlZ2fb/j8nJ0enTp0qtJyK3tcAALiay/85PiIiQnFxcVq7dq3Onj2rWbNmaf78+bYvgLl16dJF8fHxWrx4sc6dO6ft27erZ8+eSkhIUEpKinx9fVWnTh2lpKRo5syZyszMzPOD8ty5c4WWk58j5TqisDrvvfdezZs3T23atJEktW/fXvPmzdO9995r98dc1apV5eHhoaioKDVv3lxvvvmmS9u3bt069evXTwkJCfLx8ZGfn5+qVatWYCyObsei1udMLEXdBoUpyja6OuaKqrDx16lTJx07dkwLFixQVFSUS9rao0cP7d27V0uWLNG5c+e0atUqrV+/Xvfff78kKTU1VdOnT1d6ero2b96sL7/8Ur1797Zsc2Fl2uOK7V7cGApT2DZKTU3Vm2++qbS0NG3atElffPGF+vbtW2g8oaGhSkhI0Lx589SnTx9bfYVt29jYWLvbefPmzbpw4cI1l2mHh4dr7dq1yszMzLPcqv+d6U9n59GitN2ROejq2CzqWNi9e7deffVVzZ49WzfeeGOedWUxvzjC3vgrSrz25jNXHCet4omKitLKlSu1atUqZWRkaO7cuXbjLIn93KpMq7Zbxe2q9lrto676/iIVPu8X1qb8rNrozPe5wt5rNR86yypmqzFiNTfFxcXp119/1fjx45WRkaFjx47p2LFjtodwltRcWBBnjreOjtHijE97c1duxdmPYmNjdfDgwcI2u0udP39e6enpOn36tDZv3qyhQ4cqNDS02Al9Z7ZdjRo1lJaWpu+++07Z2dmaPXu27dhvVY6r5iwAAK4bV28+6qoHMhlz5Wb/ERERJjAw0PTp08fs3bvXGHPtg3aMMWbXrl2mT58+JjAw0ISGhpolS5YYY67cKHzkyJEmODjYdO/e3cydO9dMnz7dREZGGmOMmTx5srnzzjvNjz/+aFlOfoWVm1tB8Q4YMMCsXLmy0DpPnTpl/P39zVdffWWMMeaXX34x/v7+tod95Hf1gUxXJSQkmObNm9seJGCvrvwxWrUvMzPTjB071tx9992mZcuW5sEHHzQHDhwoMB5j7G/H/DfPL0p9VuscfSCTM/1SkMK2Uf4yco+5gsrv37+/Wb169TX1ONpfVz3zzDOmffv2tgdyuaKtW7ZsMVFRUaZly5YmIiLCtr/HxsaaBx54wEyZMsW0bt3adO7c2Xz22WcOtdlemVb7jbNjMP8DmfKXay+G06dPG39/f9v7Dh06ZEJCQmyv9+zZYzp37lxgnVbbaPPmzWbq1Klm0qRJpnXr1qZbt27mf//3fwuN56pnnnnGdOrUyeTk5ORZbjWXdO7c2cyaNavAWMePH2+efvrpa5anpaWZO+6445qHXRTW/4XF70gf2VOUttubg4y59hjg6HjM/XrChAmmWbNmJiQkxPaX+6FopTW/OPpApsLGn6PHwfz958xxJP8DmYp6jFy+fLnp3r27CQkJMaNHjzazZs0y3377bYGxFmWuyS9/v9ors7CxbRW3K9prtY9axebMnGc17+fvJ2fGlL02OvN9rrD3Ws2Hxjj3QKbCtovVfGhvbpo5c+Y1D+Dx9/c3r7zyiu2zrp4Lc8u9Tzh7vLXXF46WWdRjpzNjzmp7de7c2bz//vt223eV1QOZFi9ebPdzBT2Q6er2DQgIMKGhoea1114zWVlZDpeZW3G23axZs0yrVq1Mp06dzPvvv2/+8Y9/mF27dhVajqvmaAAArgcexhgjXfkX19DQ0LLO1ZZ7/+2uYp35A1iZOHGiqlWrppdeeqnE61q3bp3ef//9Er/Rfnncb3JyckrsXnaAuyrrfb2s6y/Prs55pTXvA6Xp0KFD2rx58zW3fwIAAHAEmQEn5X8KJeAqFy9eVHx8vJYvX57ncrvrQXncb0iMAq5X1vt6WddfnjHn4Xr21VdfXXOLGQAAAEe5/IFMAIpm+/btGjVqlIYNG6YWLVqUdTgAAAAVwvPPP1/WIQAAgAqMy+oBAAAAAAAAuCWusQIAAAAAAADglkiOAgAAAAAAAHBLJEcBAAAAAAAAuCWSowAAAAAAAADcEslRAAAAAAAAAG6J5CgAAAAAAAAAt0RyFAAAAAAAAIBbIjkKAAAAAAAAwC2RHAUAAAAAAADglkiOAgAAAAAAAHBLXmUdAAAAAMqn2NjYsg4BKLbQ0NCyDsGtMG/gesHcAbgPkqMA4CYCAgKUmJhY5P8CcE/8OERFRqKubDBvoKJj7gDcC5fVA4CbuJrgLMp/AwICSjVWAAAAAABKA2eOVkBZly7ryImzyjifJWOM5Xs9PDzk5+OtRvVryLuyZylFCKA8Ks7cwZmjAAAAAIDrEcnRCuiPE6mq7ltFAY1qysPDw/K9xhj9mXJOR06cVZPGN5VShADKo+LMHVxaDwAAAAC4HpEcrYDOnb+kJo1uKjS5IV05+6vOzb46mZxRCpEBKM+KM3eQGAUAAABQ3l28eFFHjhzRmTNnlJFx5beMn5+fatasqUaNGqlq1aplHCHKI5KjFZAxxqHkxlUeHh6FXkIL4PpXnLmDM0cBAGVh7rwvJUmPPvJwGUcCoCJh7nBPBw8e1PHjx9W4cWMFBgbKz89PkpSRkaGkpCRt3bpVDRo0ULNmzco4UpQ3JEcBAIUiMQoAKG0ffvypvo1ZKUm6mHlRI4Y/VsYRAagImDvc07Zt2+Tn56eOHTvKyytvqsvPz09+fn5q1KiR4uPjtW3bNrVt27aMIkV5xNPqUe6MGjVK8+fPL+swAORSlKfVsy9XDGyn4qMPAdf76OPP9G3MSr3z9ht65+039G3MSn085/OyDqtC7+8VOXbAUaU9d7BflQ8HDx6Un5+fAgMDr0mM5ubl5aXAwEBVrVpVBw8eLMUIUd6RHMV1aefOnerTp4+CgoLUvXt3ffHFF2UdElChFffMUfbJ/9OuXTvFx8c7/bkdO3YoICBAx44dK4Go4CqlOdbL05j4448/9NJLL+nOO+/UqFGj8qzLzMzUhAkTFBoaqujoaK1bt86hdXBvR44e1Ttvv6EmAf5qEuCvd95+Q38cOVLWYeVRFse28rTfF9ehQ4f08MMPq127durevbtWrVpV1iHhOlCWc0dZft+9nuYGZ128eFHHjx/X7bff7vBnAgMDdfz4cV28eLEEI0NJe/bZZxUQEJDn7+effy5SWVxWj+tORkaGRo4cqeHDh2vu3Lnat2+fnnrqKdWtW1dhYWFlHR5QIRXnnqPutE8mJyfrhhtukLe3t8s/6+/vr7fffls1a9YsbpgoIaU91svTmLh06ZK6du2qpk2b6qeffsqz7sUXX1S9evW0Zs0aJSYmauDAgVq0aJECAgIs18G9TZ38jzyvmwT4X7OsLJXVsa087ffFNWHCBLVp00bz5s3Tzz//rP79+2vXrl08LAXFUlZzR1l/372e5gZnHTlyRI0bN7Y8YzQ/Ly8vNW7cWEeOHOH+oxXYoUOH9MYbb6hNmza2ZfXr1y9SWSRHkUdiYqL69++vHTt22JZFREToH//4hzp06KD33ntPX375pdLT0/WXv/xF06dPV61atSRJCQkJeuWVV7Rnzx7VrVtXo0ePVmRkZKF1xsTEaObMmUpJSVGPHj3y/OvNjh07NG3aNP3222+69dZbNXbsWHXt2lWStGXLFv3zn/9UYmKi7rjjDk2dOlWBgYHat2+fPD09NXz4cEnSPffco8jISG3YsEGLFi1ScHCw7ayWlJQU3XPPPVq7dq0aNWpk2QZ79QHuwJHEqL19uTj7pFU8pTlXrVu3TosWLdIdd9yh9evX68KFCxoyZIgGDhyo5ORkRUdHa/DgwZo+fbpWrFihSpUqFVjH4MGDlZKSokGDBmnmzJm6/fbbr/ns77//runTp+vo0aOqVauWRowYoUGDBskYo3Hjxik+Pl7JycmKjIzUq6++qsWLF+unn35SWFiYpk6dKg8PD8s2Ws25cExpjnVJdsdz7jGxevVqPffcc7bPZGVlqXXr1vryyysPpCjqMdpRTZo0UZMmTa65tDA5OVlr167Vtm3b5OnpqWbNmql3795asmSJhg4danfduHHjXBYbUBylub9HR0dr4MCB6t+/vw4ePKj77rtPr776qgYMGKDffvtN4eHh+u677xw+FhTlO7YjMbjqUtQ///xT4eHh8vLyUnBwsCpXrqzU1FTVq1fPJeUDJaE054QzZ84oPDxczz33nFasWKHffvtNL7zwgg4ePKgtW7bo0KFDmjJlinr06OHU90Qr5fE3b2xsrOV6Hx8fBQUFOV1u7dq1tW/fPh0pZ1cllKXQ0NCyDsEphw4dUps2bXTrrbcWuywuq4fDNm3apE8++UQfffSR1qxZo6ysLM2aNUvSlVPZH330UXXv3l0bNmzQSy+9pAkTJmj37t2WZR44cEDjxo3T6NGjFRsbqyZNmmjNmjWSrvygeuyxx/TQQw9pw4YNeuqpp/T000/rjz/+0MmTJ/XEE09o5MiRWr9+vVq1amU72AQHB+vf//63rY6srCz9v//3/3TXXXcpMjIyzyU7K1euVOvWrdWoUSPLNljVB7iDws7istqXi7pPFlVJzFWS9MMPP6hWrVpatmyZ5s6dqw8++ECbNm2SdGW+SkhI0ObNm9W4cWO7dcyfP18333yzFixYoA4dOlzz2VtvvVWjR4/W8OHDtWnTJr344ouaNGmS0tPTr4knLS1Nv/zyi2bPnq1Vq1Zp6dKl2rZtm2UbrbYTHFPaY91qPOfWs2dP7d27V3v37tX69etVp04dDR48WFLxxn1xHThwQHfccUeeM6Lvuusu7d+/33IdUB6U9v5+7733avPmzZKk7du3q3r16tq6daskaevWrbr77rtVo0aNPJ+xdywo6ndsR2JwlWHDhmnixInat2+f5s6dq7Zt25IYRblWFt93T58+rQsXLujzzz/Xs88+q7/97W9q2LChFi5cqCeeeELvvvtugZ+zNzdYqai/ec+fP297Kr0z/Pz8dP78+RKICKXh9OnTSktL08svv2y7lcVXX31V5PJIjsJhFy9elIeHhzIyMlS3bl19/vnnGjt2rCQpLi5OtWvX1mOPPaYaNWqoW7duio6O1saNGy3LjImJUXh4uCIjI1WzZk2NGDFCTZs2lSStXr1agYGBeuihh1SjRg3df//96ty5s5YvX66YmBi1b99eERERqlWrlsaMGaOHH35YWVlZ8vX1tZ1KffDgQQ0aNEiNGjVSeHi4wsLC9Pvvv+vQoUOSpOXLl6tv376FtsGqPsAdFHbmqNW+XNR9sqhKYq6SrnyBGjRokCSpQYMGevDBB7VkyRJJVy4pnjBhgmrXru10Hbk/6+HhoaVLl6p///6qWbOmbrnlFnl4eCgtLe2az2VmZmrgwIGSpBo1aigoKEhnzpwpdC6zt53gmNIe61bjuSDZ2dl6+umnFRUVpYiICEnFG/fFlZKSourVq+dZdsMNN+jMmTOW64DyoLT39/yJyUceecSWmNy2bZvt6qnc7B0LivoduygxFFW7du109uxZjRs3Tm+88QZnjKPcK4vvu5cvX9Zf//pXSVLr1q1VpUoVPfDAA7bXZ8+eLfBz9uaGwtrHb15UFMnJybrzzjs1ZMgQbdu2TS+//LKmTJmiDRs2FKk8kqMVkIeHh4wxDr/fGFPo6fNWcnJyJEndunXT2LFjNXXqVLVq1UoTJ0603fD56NGjio+PV8eOHW1/69atU0ZGhmXZJ0+e1G233ZZnmb+/v21d48aN86xr3LixTpw4oWPHjuVZV61aNQ0dOtR29smFCxc0bdo0DRw4UOHh4Zo9e7aqVq2q6tWrq2vXrvruu+/0559/av/+/bYfj1ZtKKw+oCIoztxR2JmjVvuyVLR90lklOVdJUt26dVWp0v8dNhs0aKCTJ09KkipVqmS7bN/ZOnJ/1tvbW1u3btXDDz+sBx54QB999FGeOnPz9PTMc1+pq9vKqv7CthMKV9pj3Wo8F2TatGny9vbOc4l9ccZ9cd10003XnPmcnp6um266yXIdUB6U9v5+1113KTMzU4mJidqxY4cGDhyoS5cu6dChQ9q2bZvuvffeaz5j71hQ1O/YRYmhKC5duqRHHnlEkyZN0sqVKzVjxgwNGTJEP/74o0vKB0pCWX3fveGGGyRd2b+rVKmiKlWq2F7bY29usFJRf1JWSYMAACAASURBVPP6+PgU6TtNRkaGfHx8SiAilIbbb79d//nPf9StWzf5+Pjo3nvvVWRkZJGviuOeoxWQn4+3/kw5pzo3+xY6yRlj9GfKOfn5OD6hZWdn2/4/JydHp06dkiSdOHFCnTp10sCBA3XixAl9+OGHGjFihNauXavatWsrJCRECxcutH32zz//VOXKlS3rql+/vg4fPpxn2dV7ftStWzfP/QSlKz/wWrZsqUqVKmnfvn15Yv700081YMAA+fr6avjw4apevbrWrFlzzeVHvXr10pw5c1StWjX17NlTvr6+kmTZhkWLFtmtL/9ZL0B5VZy5o7AzR6325ZycnCLtk4UpzblKko4fP65Lly7Z3puYmKg6depIUp4EprN15P5sbGys3n33XX355Zdq2rSpLl++bPceSva2oVX9c+bMsbud4JjSHutW4zm/xYsXKzY2VsuWLSvWmHSlFi1aKD4+XllZWbYfV3v27FFgYKDlOqA8KO393dPTU506ddKSJUvk5eWlevXqqW3btvr666+Vk5OjwMBApaSk5PmMvWNBUb9jOxKDKyQmJio7O1tdunSRJN133306ePCgYmJi1LlzZ5fUAbhaWXzfLaqinBxVp06dcvmbt7D7YB48eFBJSUlOX1qflJSkm2++We3bty9OeCgju3fv1vHjx/P8Q4OXl5eqVatWpPI4c7QCalS/hjLOZ2nPr6f0088nLP/2/HpKGeez1Kh+jcIL1pVT7tPS0vTdd98pOztbs2fPVmZmpqQrDyTp16+fEhIS5OPjIz8/P9vA69Kli+Lj47V48WKdO3dO27dvV8+ePZWQkGBZX1RUlFauXKlVq1YpIyNDc+fOtX2mR48e2rt3r5YsWaJz585p1apVWr9+ve6//35FREQoLi5Oa9eu1dmzZzVr1izNnz9fvr6+iouL06+//qrx48fbzvo8duyYUlNTJV2ZXBMSEjRv3jz16dPHFotVG6zqAyqK4swdhZ05arUvF3WftFLac5Ukpaamavr06UpPT9fmzZv15Zdfqnfv3te8z5E6zp07V2AdKSkp8vX1VZ06dZSSkqKZM2cqMzPzmh/DVqzqt9pOcExpj3Wr8Zzb7t279eqrr2r27Nm68cYb86wrzrgvrjp16qhr16764IMPlJOTo0OHDtluHWG1DigPSnt/l65cWj9v3jzbk3fbt2+vefPm6d5773Uq2VHU79iujMHKbbfdppycHH399dfKzs5WUlKSfvjhB7Vq1col5QMloSzmhNJU2G/eX3/9VevWrbO9Py0tTTExMXmuAlm3bp3i4+NLNe5GjRrpjz/+yHPiRGGys7N15MiRYj3nAGWrcuXKGj16tFauXKmMjAxt3LhRy5cvL/oDR81/rV271gDGGDNr1izTqlUr06lTJ/P++++bf/zjH2bXrl0mMzPTjB071tx9992mZcuW5sEHHzQHDhywfW7Xrl2mT58+JjAw0ISGhpolS5Y4VN/y5ctN9+7dTUhIiBk9erSZNWuW+fbbb40xxmzZssVERUWZli1bmoiIiDzjNDY21kRERJjAwEDTp08fs3fvXmOMMTNnzjT+/v7X/L3yyiu2zz7zzDOmU6dOJicnJ08sVm2wVx+AK+zty8XZJ62U5lwVGxtrHnjgATNlyhTTunVr07lzZ/PZZ58ZY4xJSkoyzZo1y/N+qzomT55s7rzzTvPjjz9e89kLFy6YkSNHmuDgYNO9e3czd+5cM336dBMZGZnnvQXVOWDAALNy5cpC67eac+GY0hzrVuM59ziYMGGCadasmQkJCbH9derUyVZOUY/Rjn4/HDZsmAkLCzNt2rQxwcHBJiwszAwbNswYc2Vcjxs3zoSGhpro6GgTFxdn+5zVOri3iS9PMQm/JdpeJ/yWaCa+PMXpcor7G6e0j22nTp0y/v7+5quvvjLGGPPLL78Yf39/s2LFCmOMcepYUNTv2IXFUBhH+3zPnj3moYceMh06dDCdOnUy77zzjlN9BRSkpOeO0pwTTp8+bfz9/W2vDx06ZEJCQmyv9+zZYzp37myMcW5usGL1m/ett94yYWFhttcHDhww/v7+5tdff7Ut69atm5k5c6ZD7XOl+Pj4PN/3C7N7924THx9fghGhNCxdutSEhoaaFi1amMjISPP9998XuSwPY67cgC42NrbQ05UBABVTQECAEhMTi/zfypUr69KlS2XdjDKzbt06vf/++7YHMJVH/z2cu+ysHkDi+yHKzkcff6ZlMSv0zttvSJJGjX5e0Q9EaviwIU6V425juDwcC9ytz1G+uNvckZOTY/f+9LmVh7mhpG3btk3Vq1dXs2bN5OVV8B0ks7OzdeDAAV28eFFt27Yt5QhRnnHPUZSoAwcOaM6cOXbX33fffbrvvvtKMSLAPV29Z2hR/1vaidHSnjsKq69evXouq6ukXM9fdq9nHCeBgj0xfKiMjEaNfl6S9ECvCKeTG+VNaezvHAvg7irS3OGKOcGRxKh0ZW643r9ztG3bVgcPHtTGjRvVuHFj1a5d23Yf0oyMDCUlJenIkSOqX7++7rzzzjKOFuUNZ44CAFCIinDmKFAS+H6IsjZ33peSpEcfebhIn2cMlz76HOUBc4f7unjxoo4cOaIzZ87YnmLv5+enmjVrqlGjRqpatWoZR4jyiDNHAQAoRLdu3dStW7eyDgMA3E5RExsA3Btzh/uqWrWqmjVrVtZhoILhafUAAAAAAAAA3BLJUQAAAAAAAABuieQoAAAAAAAAALdEchQAAAAAAACAWyI5CgAAAAAAAMAt8bR6AAAA2BUbG1vWIQCoYJg3AAAVCclRAAAAFCg0NLSsQwBQwTBvAAAqGi6rBwAAAAAAAOCWSI4CAAAAAAAAcEskRwEAAAAAAAC4JZffc/TSpUs6f/68cnJyXF003EilSpXk4+OjypUrl3UoAAAAAAAAuE65/MxREqNwhZycHJ0/f76swwAAAAAAAMB1zOXJURKjcBXGEgAAAAAAAEoS9xwFAAAAAAAA4JZIjgIAAAAAAABwSyRHAQAAAAAAALglkqMAAAAAAAAA3BLJUQAAAAAAAABuieRoOZCamqq7777bpWVmZ2dr48aNttcHDhxQ37591atXLy1fvjzPOgAAAAAAAMAdkRx1wAcffKAff/yxrMNwyqlTpzR58mSlp6dLkj799FN17dpVMTExuuuuu/KsAwAAAAAAANwRyVEHJCQkKCUlpazDcErDhg21evVqVa9eXZJ0+vRptWzZssB1AAAAAAAAgDu6LpKjiYmJGjFihCIjI/Xggw9q/fr1Sk1N1SOPPKLZs2crPDxcX3zxhSTp4MGDGjp0qHr16qVhw4bpyJEjkiRjjEaNGqWoqChFRETohRdeUHZ2tl566SVt2bJF7733nqKjo/XHH39YllOUWHOzF4cxRh988IH69eunqKgoDRs2zHJ5amqq2rdvL0l644039PPPP2v69Onq16+ftmzZYltn1RZ7fQgAAAAAAABcD7zKOoDiysrK0t///neNGDFCUVFRysjIkDFGly9f1t69e9WjRw8tW7ZMlStX1qVLlzRq1Cg999xz6t69uxYsWKBp06bpww8/lIeHh5599ln5+/srKytLvXv31vr16/Xqq69qzJgx6tKli3r37i1JluUUNdar7MVRu3ZtLV26VIsWLVLNmjV14sQJeXl5affu3QUuz+3555/XL7/8osGDB6tbt25KTU21rSusLfn7EAAAAAAAALheVPgzR3fv3q1q1aopKipKkuTn55fncvHBgwfL29tbHh4e+umnn5STk6PQ0FBJUrdu3bRr1y5bctLf31+S5O3trcDAQLtngxZWTlFjvaqgOHx9fXXhwgX9/vvvkqT69etLkt3ljnKkLbn7EAAAAAAAALheVPgzR5OSklSvXj2H33v27FnbGaDSlWTihQsX5OPjo7lz52rlypVKS0vTuXPn1Lx5c6fL8fPzK1asOTk5BcbRtGlTjR8/XhMnTlTNmjU1atQotW3b1u5yR1m1BQAAAAAAALieVfjkaK1atZSUlOTwe5s0aaIFCxZcs27t2rVav3695syZoxtuuEGvv/56kcopbqzr1q2zG0dkZKTCw8MVFxenMWPGaOHChWrUqFGByx192JJVW3Jffg8AAAAAAABcbyr8ZfUhISFKTU1VXFycpCv39Tx16lSB7w0ODlZSUpJiY2Nty44ePSpJOnnypKpVqyY/Pz+lp6crISHB9p5q1arp8OHDtvuDWpVTlFi9vb1VuXJlJScn243j4sWLOnXqlDw9PdWmTRv5+Pjo7Nmzdpc7qqhtAQAAAAAAACq6Cp8crVKlimbOnKnPPvtM0dHR6t+/vzZs2FDge6tVq6Z33nlH8+fPV1RUlKKiojRnzhxJUlRUlG666Sb17t1bo0eP1s0332z7XHh4uJYsWaK+fftqz549luWkp6erS5cu1/wtXbrUbqw+Pj567LHHNGjQILVq1arAOI4fP65nn31WvXv3Vp8+fXT//fcrKCjI7nJHWbUFAAAAAAAAuJ55GGOMJMXGxtoeylMcZ86cKXYZwFU1a9Ys6xAAAAAAAABwnarwZ44CAAAAAAAAQFGQHAUAAAAAAADglkiOAgAAAAAAAHBLJEcBAAAAAAAAuCWSowAAAAAAAADcEslRAAAAAAAAAG6J5CgAAAAAAAAAt+Ty5KiHh4eri4SbqlSJ3D0AAAAAAABKjsuzT76+viS1UGyVKlWSj49PWYcBAAAAAACA65iXqwusXLmyatSo4epiAQAAAAAAAMClOMUTAAAAAAAAgFsiOQoAAAAAAADALZEcBQAAAAAAAOCWSI4CAAAAAAAAcEskRwEAAAAAAAC4JZKjAAAAAAAAANwSyVEAAAAAAAAAbonkKAAAAAAAAAC3RHIUAAAAAAAAgFsiOQoAAAAAAADALZEcBQAAAAAAAOCWSI4CAAAAAAAAcEslkhxNSUnR5MmT1b17dwUHBysyMlILFiwoiaoKtW3bNg0ePFh/+ctf1LZtW40YMUK//PKLJOnw4cMKCAjQk08+WWL1nzx5UgEBARo2bJht2bJly9SqVSu1aNFCktSrVy8FBATo8uXLJRYHAAAAAAAAgLy8XF3g+fPn9eCDD+rIkSPq0KGD7rjjDm3atEkvv/yyLl26pKFDh7q6SruWL1+uUaNGycPDQyEhIUpLS9OaNWu0efNmffPNN/Lycnnzr+Hl5aU77rhDDRo0sC379NNPlZqaqvHjx0uSGjduTGIUAAAAAAAAKGUuzw5u3rxZiYmJGjhwoKZOnSpJOn78uDp37qyYmJhSS45mZmbqtddeU05OjubMmaPQ0FBJ0syZM/Xuu+/q448/1siRI0s8jlq1amnlypV5lp07d06enp4aMWKEJOmDDz4o8TgAAAAAAAAA5OXyy+rT0tIkXbmcfdOmTcrOzlaDBg20d+/ePJfWz507V926dVOLFi3UuXNn/etf/5IkjRo1SgEBAdq+fbuk/7ss/X/+538kSXFxcYqKilJwcLAeeOABrV27tsA49u/fr+PHj6tt27a2xKgkDR8+XJMnT1bnzp0L/Jy9uCRp8eLFCgsLU4sWLRQaGqo5c+YUui7/pfsRERH67bffdPnyZQUEBCg9Pf2ay+rttfHnn39WQECAxo0bpzfffFODBw+WJD366KPq2LGjjh496tA2AgAAAAAAAFACydEuXbqoVq1aOnjwoAYNGqQ777xT/fr100cffaTMzExJVxKnkydPlre3t4YMGSIfHx+99dZb2rp1q6KioiRJsbGxkqQNGzZIunJfzvj4eA0bNkznz5/XsGHDlJWVpSeeeEJ79+69Jo7Dhw9Lkpo3b55nua+vrwYPHqzIyMhrPmMV14EDBzR+/Hh5enpq5MiRqlGjhv75z39q1apVluvye/vtt9WwYUN5enrq3//+t3x8fPKsd6SNcXFxWrhwofz9/SVducfriRMnuDQfAAAAAAAAcILLL6u/+eabtXLlSn399dfasmWL9uzZo507d2rnzp2KiYnRypUrVbNmTb355ptq3bq1jDFKSUnRwYMHdfjwYf31r3/VjTfeqLVr12r8+PH68ccfValSJd1///36+OOPdfnyZQ0aNEitW7dW7dq1NXHiRC1YsECvv/56njiMMZIkT09Ph2O3iqtmzZoyxsjX11d33HGHIiMjdfDgQfn7++vw4cN21+XXvHlzVa1aVR4eHrrrrruuWf/111/bbeOQIUMkSZUrV9bGjRtVuXJlSVJMTIzDbQQAAAAAAABwhcuTo6mpqbp48aIeffRRPfnkk8rJydHWrVv15JNPKjExUb///rtuvvlm7d69W5MnT1ZOTo5q165t+3zlypUVHh6uRYsW6dChQ9q4caPatWunOnXq2C4bnzJlSp46Dx06dE0ct956qyQpISEhz/KsrCx98cUXuvHGG9WqVas862666Sa7cXXs2FGDBg3S119/rZEjR8rT01OdOnXSpEmTLNc5y5E2tmvXzpYYBQAAAAAAAFA0Lr+sftGiRercubM+/PDDKxVUqqR77rnHdnl79erV9dFHH2nBggUaO3asdu/ere7du+cp4+ql9e+++65Onz6tXr16SZItWblgwQIlJiYqMTFRv/zyS557mV7VsmVL1a1bVxs3btSOHTtsy7/77jtNnTq1wHuVWsV17NgxhYWFae3atZo1a5Y6d+6suLg4zZw503Kdsxxpo5eXy3PaAAAAAAAAgNtxeXI0IiJC1atX1/vvv6+//vWveuGFF9S7d29t375d7dq1U/369W2XvK9du1Yvv/yyFi1aJEm2e2a2bdtW9erV09KlS+Xl5aX77rtPkhQdHS0PDw9NmzZNH3/8sf72t7+pefPmmjt37jVxVK1aVS+88IIkafDgwXrsscc0dOhQPffcc/L29tbw4cOv+YxVXNu2bdOQIUP07LPP2i6zlyQ/Pz/Ldc5ypo1X8UAmAAAAAAAAwHkuT47eeuut+vLLL9W9e3edOHFC//nPf3T27Fk9/vjjmjVrliRp6NChatWqlXbu3Kk//vhDL7/8surXr2976FClSpXUq1cvGWPUsWNH3XTTTZKk1q1b691335UxRjNmzND+/fs1fvx4Pf744wXGEhUVpXnz5qlVq1batWuX9u/fr06dOmnhwoUKCQm55v1WcQ0YMEDPPPOMTp06pbfffls//PCDoqOjNXbsWMt1znK2jRIPZAIAAAAAAACKwsP893TJ2NhYhYaGlnU8AAAAAAAAAFAqXH7mKAAAAAAAAABUBCRHAQAAAAAAALglkqMAAAAAAAAA3BLJUQAAAAAAAABuieQoAAAAAAAAALdEchQAAAAAAACAWyI5CgAAAAAAAMAtkRwFAAAAAAAA4JZIjgIAAAAAAABwSyRHAQAAAAAAALglkqMAAAAAAAAA3BLJUQAAAAAAAABuyausAwCA8uLSpUs6f/68cnJyyjoUVGCVKlWSj4+PKleuXNahAAAAAAAKQXK0gomNjS3rEIDrRmhoaJ7XJEbhCjk5OTp//rxq1KhR1qEAAAAAAApBcrQCyp/QAeC8gv6hgcQoXIWxBAAAAAAVA/ccBQAAAAAAAOCWSI4CAAAAAAAAcEskRwEAAAAAAAC4JZKjAAAAAAAAANwSyVEAAAAAAAAAbonkKADAIampqbr77rtdWmZ2drY2btxoe33gwAH17dtXvXr10vLly/OsAwAAAADA1Uo1OXrs2DEFBweXWn0xMTHq0KGD1qxZU2p1AgAcd+rUKU2ePFnp6emSpE8//VRdu3ZVTEyM7rrrrjzrAAAAAABwtevqzNEdO3YoICBAx44dkyR98sknevHFF9WjRw+NGTNGTz31VBlHCADIrWHDhlq9erWqV68uSTp9+rRatmxZ4DoAAAAAAFztukqO+vv76+2331bNmjUlSWfPntUtt9wiSfr73/+u9957z6nykpOTlZWV5fI4AcCKMUYffPCB+vXrp6ioKA0bNkzZ2dkyxmjUqFGKiopSRESEXnjhBWVnZ+uZZ57RF198Yfv8hg0bNHDgQEnSwYMHNXToUPXq1UvDhg3TkSNHHIohMTFRI0aMUGRkpB588EGtX7/+mhgLisUq9oKWp6amqn379pKkN954Qz///LOmT5+ufv36acuWLbZ1Vm1JTU3VI488otmzZys8PDxPXwAAAAAAYKVEkqNbtmxRVFSUgoKC1LdvXx04cKDA961atUo9evRQixYt1LlzZy1YsMC27r333tM999yjoKAgDRkyRMnJyYWuM8Zo3Lhx8vHx0fDhw3XkyBENHz5c33//vb755hu99tprtjISEhI0cOBABQcHKywsTMuXL5d0JSHasWNHzZ49W+3bt9ehQ4fstjM5OVnt2rXT999/ryeffFLt27fXxIkTZYwpTvcBcHN79uzR0qVL9dFHH2nZsmWaOnWqvLy85OHhoWeffVbLli3T0qVLtXv3bq1fv15hYWFat26d7fPr1q1Tjx49dOnSJY0aNUqDBg1STEyMunbtqmnTphVaf1ZWlv7+978rMjJSy5cv1yeffKLWrVvneY+9WOzFbm95bs8//7xatGihcePG6euvv1bz5s1t6wpry969e+Xr66tly5bp4YcfLmrXAwAAAADcjMuToydPntQTTzyhkSNHav369WrVqpVGjRp1zfsyMzM1evRoDR8+XJs2bdKLL76oSZMmKT09XZs2bdInn3yijz76SGvWrFFWVpZmzZolSZbrcvv444/VqFEjffzxxwoLC8uz7uLFi3r00UfVvXt3bdiwQS+99JImTJig3bt3S7qS9ExISNDmzZt1++23W7Y3LS1Nv/zyi2bPnq1Vq1Zp6dKl2rZtW1G7DwDk6+urCxcu6Pfff5ck1a9f37bO399fkuTt7a3AwEAdOXJEXbt21f79+3X69Gnl5OQoLi5OPXr00E8//aScnByFhoZKkrp166Zdu3bp8uXLlvXv3r1b1apVU1RUlCTJz8+vwEvbC4rFXuxWbXKEI20ZPHiwvL295eHh4VTZAAAAAAD35VX4W5wTExOj9u3bKyIiQpI0ZswYNWjQ4JrL0z08PLR06VI1a9ZMknTLLbfIw8NDaWlpunjxojw8PJSRkaGWLVvq888/t/0AtlrnqLi4ONWuXVuPPfaYpCs/sqOjo7Vx40b1799fly5d0oQJE1SrVq1Cy8rMzLRdvlqjRg0FBQXpzJkzTsUDALk1bdpU48eP18SJE1WzZk2NGjVKbdu2VU5OjubOnauVK1cqLS1N586dU/PmzXXDDTeoTZs2+uGHH9SoUSPVq1dPDRs21O7du3X27Fn17t3bVnb9+vV14cIF+fn52a0/KSlJ9erVs4zRXiz2Yre33FFJSUl22wIAAAAAQFG5PDl67NgxNW7c2Pa6WrVqGjp06DXv8/b21tatW/XKK68oPT1dt912mypVunIia7du3TR27FhNnTpVx48fV8+ePfXEE0+oadOmluscdfToUcXHx6tjx455lkdHR0uSKlWq5FBiVJI8PT1t9ziVxBlLAFwiMjJS4eHhiouL05gxY7Rw4ULFx8dr/fr1mjNnjm644Qa9/vrrtvf36NFDa9asUePGjdWjRw9JUq1atdSkSZM8tyxxRK1atZSUlGT5nnXr1tmNpaDYGzVqVOByRx+2ZNWW1NRUp9oHAAAAAMBVLr+svk6dOjpx4oTtdXZ2tj766COlp6fneV9sbKzeffddTZkyRd9++61mzpxpS46eOHFCnTp10ooVK7Ry5Ur5+PhoxIgRha5zVO3atRUSEqKNGzfa/r755hsNHz5ckmxxOIJkKABXu3jxok6dOiVPT0+1adNGPj4+Onv2rE6ePKlq1arJz89P6enpSkhIsH2ma9eu2rVrl2JjY223EgkODlZSUpJiY2Nt7zt69Gih9YeEhCg1NVVxcXGSrtyD9NSpU/L29lblypWVnJxsNxZ7sdtb7qiitgUAAAAAACsuT45GREQoLi5Oa9eu1dmzZzVr1izNnz9fvr6+ed6XkpIiX19f1alTRykpKZo5c6YyMzOVkpKidevWqV+/fkpISJCPj4/8/PxUrVo1SbJc56guXbooPj5eixcv1rlz57R9+3b17NkzT6LBFX799dc8D0lJS0tTTExMnkTxunXrFB8f79J6AVRsx48f17PPPqvevXurT58+uv/++xUUFKSoqCjddNNN6t27t0aPHq2bb77Z9pnq1aurVatWuummm9SwYUNJV87cf+eddzR//nxFRUUpKipKc+bMkSSlp6erS5cu1/wtXbpUVapU0cyZM/XZZ58pOjpa/fv314YNG+Tj46PHHntMgwYNstWVPxZ7sdtb7iirtgAAAAAAUFQe5r+PVo+NjbU96KK41q1bpzfffFOHDx9W8+bNNXnyZAUFBenYsWMKDw/X3r17dfHiRY0ZM0YbNmxQnTp19MgjjygpKUnr1q3Tv//9b7300ktau3atsrKyFBQUpFdeeUUtWrRQVlaW3XXJycnq0KGDLdkYGhqqGTNm6K677tKMGTN0/vx5TZw4UdKVh3tMnTpVv/zyi+rVq6enn35affr0uaYMKwW99+GHH9Yjjzyi8PBwzZgxQytXrtSaNWskST///LMiIyP13Xff2R70FBoaqujo6AIfWlUQV24nwJ0VtC9xv2C4Uu5brgAAAAAAyqcSSY66s/92Z4ldbs92AlyD5ChKGslRAAAAACj/XP5ApuvJgQMHLC/bvO+++3TfffflWcY9SAEAAAAAAICKgeSohcDAQM2YMaOswwAAAAAAAABQAlz+QCYAAAAAAAAAqAhIjgIAAAAAAABwSyRHAQAAAAAAALglkqMAAAAAAOD/t3fnYVWX+f/HX+yrKSZuBCqZueQFSZq7gULDkEBWXpl6WS65zOTWQt+padEZq3Fr6meW65T6tSanNHVMSQi/WmbjIJkbKoqibNIBEUS28/vDPIkcDqBHj3Cej+vyKXaMxQAAIABJREFUqvP53Od9v+9z7IQvP/fnAIBdIhwFgF85ODjYugU0Eo6O/O8VAAAAABoC/vQGAL/y8vIi1MINc3R0lKenp63bAAAAAADUgbOtG0D9JSQk2LoFoFFycXFR06ZNbd0GAAAAAAC4RQhHG5iwsDBbtwAAAAAAAAA0CuwfBQAAAAAAAGCXCEcBAAAAAAAA2CXCUQAAAAAAAAB2iXAUAAAAAAAAgF0iHAUAAAAAAABglwhHAQAAAAAAANglwlEAAAAAAAAAdolwFAAAAAAAAIBdIhwFAAAAAAAAYJcIRwEAAAAAAADYJcJRAAAAAAAAAHaJcBQAAAAAAACAXbop4WiXLl3UqVOnm1G6mqeeekqBgYEyGAw1jtmzZ49Gjx6tBx54QL169dLEiRN1+PBhSVJ6eroCAwM1adKkm9ZjVlaWAgMDNX78eNOxr776Sj169FCXLl0kSY888ogCAwNVUVFx0/oAAAAAAAAA8BtnWzdws23evFnTpk2Tg4ODgoKCdP78ecXHx+v777/Xl19+KWfnm/8SODs7695771Xbtm1Nx1asWKH8/HzFxcVJkgICAghGAQAAAAAAgFuoUYejly5d0ltvvaXKykotW7ZMYWFhkqR3331X7733npYuXarJkyff9D5atGihLVu2VDlWVFQkJycnTZw4UZL0wQcf3PQ+AAAAAAAAAPzGJvccPXbsmMaNG6cePXqod+/emjFjhnJyckznP/74Y4WGhqpLly4aMGCAFi1aZDq3d+9excTEKDg4WHFxcSorK6txngMHDujs2bPq1auXKRiVpAkTJujNN9/UgAEDzD7P0vzr1q3TkCFD1KVLF4WFhWnZsmW1nrt2635kZKSOHz+uiooKBQYGqrCwsNq2+qSkJEVHR6t79+4aOnSotm/fLkk6dOiQAgMD9eKLL2ru3LkaPXq0JGnMmDHq16+fMjIy6vYmAAAAAAAAAHbuloej58+f11NPPaWdO3cqKipKvXv31oYNGzR+/HhVVlZqz549evPNN+Xq6qqnn35anp6emj9/vn744Qfl5ORozJgxOnr0qKKionTw4EHt3bu3xrnS09MlSZ07d65y3MvLS6NHj1ZUVFS151ia/+DBg4qLi5OTk5MmT56spk2bas6cOdq6davFc9dauHCh/Pz85OTkpC+++EKenp5Vzqempmr8+PEqLi7W+PHjVVpaqmeffVb79+83jUlKStLatWvVoUMHSVJeXp4yMzPZmg8AAAAAAADU0S3fVr9161adO3dOM2bM0HPPPSdJqqio0ObNm5WSkiIfHx/NnTtXISEhMhqNysvL09GjR5Wenq6UlBQVFxdr5syZ+uMf/6jz588rJCSkxkDQaDRKkpycnOrcn6X5fXx8ZDQa5eXlpXvvvVdRUVE6evSoOnTooPT09BrPXatz585yd3eXg4ODgoODq53//PPPVVFRoVGjRikkJES+vr569dVXtXr1aj399NOSJBcXF+3atUsuLi6SpE2bNtV5jQAAAAAAAABsEI6eOXNGktStWzfTsa5du2rz5s06ffq0+vXrp5SUFL355puqrKyUr6+vadyVLeM9evSQJN1xxx265557TN88f6127dpJuryN/2qlpaVas2aNmjVrZqp1RfPmzWucv1+/fho1apQ+//xzTZ48WU5OTurfv7/eeOMNi+fq68o6Z82aVeX4yZMnTf/+4IMPmoJRAAAAAAAAAPV3y7fV+/n5SZIOHjxoOnYl3PTz89OSJUu0evVqvfDCC0pJSdHgwYNN41q3bi1J2rdvnySpsLBQx48fr3Gubt26qVWrVtq1a5f+85//mI5//fXXmj17tuk+nlezNP+ZM2c0ZMgQbd++XYsXL9aAAQOUlJSkd9991+K5+roSyK5evVppaWlKS0vT4cOHtXr1atMYZ+dG/V1aAAAAAAAAwE130xK2iooKvfTSS1WODR48WBEREZo7d64WLVqkvLw8nT9/Xhs3blS3bt0UFBRkukfn9u3bdeTIEW3YsMFULzY2Vu+//74++OADZWdnKyUlxeIXMrm7u+t//ud/NHPmTI0ePVp9+vSR0WjUzp075erqqgkTJlR7zpWt+Obm37Nnj/785z/rgQce0JAhQ+Tj4yNJ8vb2tniuvmJiYrRmzRr95S9/0aOPPqqUlBT9+9//1p/+9Cf169fP7HPGjBmjY8eO6bPPPtNdd91V7zkBAAAAAAAAe3PTrhw1Go1at25dlV8HDhxQ06ZNtWbNGvXu3Vvr16/Xzp07FRMTo2XLlsnZ2VnPPPOMevToob179+rUqVN67bXX1KZNG+3fv19t27bV8uXL1a5dO61fv14BAQF66KGH1KZNGzk6ml9KdHS0PvnkE/Xo0UPJyck6cOCA+vfvr7Vr1yooKKjaeEvzjxgxQlOnTlV2drYWLlyoHTt2KCYmRi+88ILFc/UVEhKi9957T0ajUQsWLNCBAwcUFxencePG1fgcvpAJAAAAAAAAqB8H46+XSiYkJCgsLMzW/QAAAAAAAADALXHL7zkKAAAAAAAAALcDwlEAAAAAAAAAdolwFAAAAAAAAIBdIhwFAAAAAAAAYJcIRwEAAAAAAADYJcJRAAAAAAAAAHaJcBQAAAAAAACAXSIcBQAAAAAAAGCXCEcBAAAAAAAA2CXCUQAAAAAAAAB2iXAUAAAAAAAAgF0iHAUAAAAAAABgl5xt3QDqJyEhwdYtAI1GWFiYrVsAAKsoKytTcXGxKisrbd0KGjBHR0d5enrKxcXF1q0AAADcMoSjDRCBDlDd8ZMZau/fps7jk5KSbmI3MIe/3AGs59qfBQhGYQ2VlZUqLi5W06ZNbd0KAADALUM4CgC4ZfjLHeDGmfuLBoJRWAu/lwAAgL3hnqMAAAAAAAAA7BLhKAAAAAAAAAC7RDgKAAAAAAAAwC5xz9EG6LFFp23dAnDb+VuUg61bAAAAAAAADQxXjgIAAAAAAACwS4SjAAAAsJr8/Hzdf//9Vq05atQofffdd7dNP+Xl5dq1a5fp8cGDB/XYY4/pkUce0ebNm6ucAwAAwO3tloajZ86cUffu3U2PN23apL59+yo+Pr7W59Y09tqat5It54bk7e6of/3B36o1HR2k4AB30+NAXxf9/ak2+mB0W/Xv5Fnl3M1w7fzX4+rXxRr10LA0ts9ZALgdZWdn680331RhYaEkacWKFXrooYe0adMmBQcHVzkHAACA25tNrxxdvny5/vSnPyk8PLxeY2fOnKkpU6bcgg4brjmPtVKvDh5Wrzviwabq0a7xhm0tvJ30x8HN5eFy+f6Vj4U01Z60Yk1ZdVapmZeqnLsV899u9dDw8DkLANbn5+enbdu2qUmTJpKkX375Rd26dTN7DgAAALc3m34hU0FBge666656j33uuecUEBBwM1tr0Jp7OSngThftO1Vi9dr+zV2Uc77c6nVvFzmFFRq/8qzpcVNPRyUdKTV77lbMf7vVQ8PD5ywAW0tPT9fbb7+tU6dOyc3NTaNGjdKwYcNkNBo1ffp0nThxQmVlZQoODtbs2bPl7OysQ4cO6a9//atycnLUs2dP5ebm1mmutLQ0vfPOO8rIyJC3t7cmT56s4OBg0/ma5nRyctLixYuVmJioS5cuqWXLlvrwww9rPH7hwgX97ne/0+7du/W3v/1Nhw4d0rx587R48WI9//zzmj59unbv3i1JOnr0qObMmaPc3Fy1bt1ar7/+uvz9/ZWfn6+pU6eqb9++Wr9+vUaPHq2RI0felPcAAAAANbsp4eju3bs1Z84cpaWl6d5779Xs2bPVtWvXKmMmTJig06dPa8KECXrrrbc0ZMiQGutdO/ann35ScXGxXn311Wpjjx07ptdff10//fSTWrVqpRkzZigqKqrWnt9//3397//+rwoLC/XAAw9o3rx5atGixQ3VtJXegR7678mLKq0wSpL8fJw1YWBztWrqrOJLlfr0hwLlFZXrxUhfTf7kcnDm6CB9PsVfo5dmqLjUqBEPNlWvQA+5Ojkor6hCb6zP0XND7lSQv7u6tHHToyF36K8bc5VZUK42zZw1YaCP2jRzUVm5UV/tO69vDhbJ291R7zzeSsdzStXNz10Xyyq15vt83XeXu0LaecjRQVqSZNB/Tl6sdU3m1nAo61KVMS9HtZC/j4ucnRx0OPOS/h6fp0qjzK6lpuOebo5a+nRbjfgwQ2MHNFOgr6vGDvDRiN5NtWKHQX8a6qsRH2ZIuhwUT3rIRz5ezjp3oVyLtv+i7PPl8nZ31CuP+Co5/aKGdPXWhuRCbf7pt6113u6O+n8j22hHapGC/D3k6uygf+4pUOLhy6/ZlfktjbM0/9WurffO4610KPOS7vNzl4eLg77873mtTy6scz3cPhri5ywA+1NWVqapU6dq1KhReuKJJ3Tq1CmNHTtW/v7+6tmzp6ZPn64OHTqotLRUsbGx+vbbbzVo0CBNmzZNkyZNUkxMjI4fP66UlJRa5yotLdVzzz2niRMnKjo6WhcuXJDRaFRFRYVpjIODg9k5fX19tWHDBn366afy8fFRZmamnJ2dlZKSYvb41V566SUdPnxYo0ePVmhoqPLz86usf9q0aXr++ec1ePBgrV69Wn/5y1/00UcfSZL279+v8PBwffXVV3JxcbHSqw4AAID6sPq2+qysLD377LOaPHmyvv32W/Xo0UPTpk2rNm7p0qXy9/fX0qVLLf6BvT5jS0pKNGbMGA0ePFg7d+7UK6+8opdffrnWH6i/++47LV++XEuWLFF8fLxKS0u1ePHiG6ppS306eur748WSJGdH6dWhLfXtkSJN/uSsXvkiWwfOWr6itFNrV4V28dJrX+boD6szTSHj3+PzlHK6RKu/z9cfV2cqs6BcTo7SK4/46oe0i5r8yVm9tTlXIx5spm5+bpKk1k2dtWX/BY1feUbbDxbpuSF3ak/aRU365Ky+2HteT/drVut66rqGVbvy9YfVmfrDqrO6t7WbHgz0qHEtNR2/2or/y1dabqmW/59BM9Zm6URemenclXV/ta9QU1ad1Z60i5oc2vy317CVqy6WGTVl1dkqwegVTdwdtSftop7/NFNzt+Rq3EAf+TWr/ncVNY2rbf6atG7qrK/3X9CUVWc1e2OuRvZpJi83h+uuB9toiJ+zAOxTSkqKHBwc9MQTT0iSAgICNGzYMG3btk2S1KFDB0mSq6urunbtqtOnTys5OVklJSV69NFH5eTkpE6dOqlZs9p/XkhJSZGHh4eio6MlSd7e3ma3tpub08vLSxcvXtSJEyckSW3atJGkGo/X1b59+1RZWamwsDBJUmhoqJKTk6sEtqNHj5arq6scHLgFDgAAgC1Y/crRTZs2qXfv3oqMjJQkzZw5U23btlVpaam1p6omKSlJvr6+Gjt2rKTLP4DGxMRo165dCgoKqvF5JSUlcnBw0IULF9StWzf94x//MP3QaqlmTEzMTV9TfTX1cFSgr6v+m345POzcxk2Xyir17a9XG5aUGS09XZJ0sdQoDxdH3eXjokOZl5R3oaLGsfe2dpOM0tafL0iSMgvK9c3BC+p/j6fS8wokSYcyL1/hue/URUXf30Qppy/3lnK6RM8M8Km1n5rW4O1e9Q8RZ/IvX+VYXikdyylVq6bOyjCUm11LfdZYU0+ODtIPaZevev0hrVij+zaT41Utbdxn+YsYfj5z+XVJyy3T4cxL6hXoqfiDF+o0LjX7Uq3z1+RYTqnpnyVlRrXwdpa3u2ON9a4NjWF7DfFzFoB9ysnJqXZrDz8/Px05ckSVlZX6+OOPtWXLFp0/f15FRUXq3LmzcnNzdffdd9c7LLyybd2Smubs2LGj4uLi9Oqrr8rHx0fTpk1Tr169ajxen54KCgoUGxtrOtamTRtdvFj7rhkAAADcGlYPR8+cOVPlPnUeHh565plnrD2NWRkZGUpNTVW/fv2qHK8txAwNDdULL7yg2bNn6+zZs4qIiNCzzz6rjh07XndNW3kw0FPJp0p0qfxyouXj5aRz9Qz+Tv9SpqVJv2ha+J06X1KhT3blmwK6a93p7aSsa7ZeZxWUq12L6l8GdaWnqx/X5Y89dVmDg6TYHndoQCdPebs5ysPVUSdyS2tcS33WaE5zL6fL295H/XYFSW5huVydr++qj9O/lKmZZ+0Xcl8ZZ635S399TyzVq0ugjlurIX7OArBPvr6+Onu26r2vMzMz1bp1ayUmJurbb7/VsmXLdMcdd+idd96RJLVo0ULnzp2r91wtWrSo9d6kNc0pSVFRUfrd736npKQkzZw5U2vXrpW/v7/Z43X9sqUWLVro7rvv1urVq6udu3r7PQAAAGzH6uFoy5Yt9fPPP5sel5eXa8WKFRoxYoS1p6rG19dXQUFBWrt2relYTk5OrfdwyszMVP/+/TVy5EhlZmbqo48+0sSJE7V9+3aLNYuLi2/aWq5Xn46e+uaqqw8NRRXy8XSqNs4oyd1CkLYjtVg7jxbrgfYeejnKV89/mmX2/pOGogq1bFL1t5FvEyflFdYvkLWkpjVc7cG7PdSrg4de/SJbxaVGjRv42xWpNa3F3PGi0so69ZRfXKHTv5Qp7vPsaue8neofkN7p5WS6orMu46w9v6V6uP00xM9ZAPbD1dVVLi4uOnfunIKCglRSUqINGzYoJiZGZ8+e1Zdffql58+bpp59+koeHh7y9vVVYWKhjx47pzjvvVHBwsEpKSrRlyxZFRkbqwoULKiy0vBtDkoKCgpSfn6+kpCQNGjRIpaWlMhgMatKkiamfrKwss3OWlJSooKBArVq1Us+ePeXp6amCggL5+vqaPV7XcLR79+7Kzc1VQkKCaWt9RkZGnb8oDwAAADef1e85GhkZqaSkJG3fvl0FBQVavHixVq1aJS8vL2tPVc3AgQOVmpqqdevWqaioSD/++KMiIiJ07Ngxi89LTEzUE088oWPHjsnT01Pe3t7y8PCod80jR44oMTHR9Pj8+fPatGlTlR/oExMTlZqaaqUVX74f59DgJvJwcZC3u6M6tXLV3qu+4OhI1iU18XDUA+09TOObe10OL73dHdW+hYucHaX7fr1HqCS5OjmouZeTKo3Sz2dKdLG0Uk3cL/9WuVRuVNtml0MQRwfpcOYluTo7KLTz5ffXt4mThnTz1rdHiqy2xprWUFZuVHml1MzTUS28nVVSVvnrdnkHBTR3sbgWS2usW0+l8vFyUq8Ov10h2+qOmgPcq9+nK1o2uTy+3Z0uCg7w0HfHzIft5sZZmv/q1+VmrQe21RA/ZwHYD09PT40dO1ajRo1SVlaW3nvvPW3cuFHR0dGaNm2apk+fru7duys6OlrNmzdXbGysZsyYoTvvvFOS5ObmprffflsLFizQ73//e82cOdP0c1lhYaEGDhxY7deGDRvk5uamd999VytXrlRMTIyGDx+unTt3VumnR48eZuc8e/aspk+frtjYWA0bNky///3vdd9999V4vK48PDz097//XatWrVJ0dLSio6O1bNky67/oAAAAuG5Wv3K0ffv2ev/99zV37lylp6erc+fOWrx4sRwdrZ7DVtOsWTMtX75cs2fP1uuvv67WrVvrtddeU8+ePS0+74knntC+ffs0fPhwlZaW6r777tPcuXNrrXnmzJkqdTZv3qwtW7YoNDRU0uWtr1OnTtXXX39tusJg9uzZiomJUadOnayy5qYeToq9/w79dLpEHVu66qeMkirboMsqpLc252rCwOZ6un8zGY3SV/vOK/5AkdZ8n683Y1rqYplRKadLlPPr1Z6+dzhpRngLubs4yMPVUd8eKTJd1fh/qUWaEdFCPTt46IPEX3Q485L+uilXEwb56PEH7lBphVGf7MrX0exSedcjbPRwcdBHY/yqHV+506DEw0U1ruGLvef1tydaa87mXHVs5ar/N6qN8ooqlF9keS1+Ps5mj9e159Jyo97alKvxA3005tcvlTp49pIWJfxS6/uU92tvT/f30V0+LjJKeu+bPJ27UGF2fnPjJNU4/6Vyo+l1eefflrcXXu96YFsN8XMWgH2ZNGmSJk2aZHpsLhBs0qSJ5syZY/b5ISEh2rZtmy5duiR3d/cq53bs2FHjvF26dNE//vEPi/3UNOfVV8RfERgYaPZ4s2bNtHv3btPjFStW1Hiuc+fOWrlypdkaycnJNa4FAAAAt4aD0Wg0Sqqy3Qe3r4SEBC06dI/Zc0O6eimroLxe987Ereft7qiPx/npsUWnrTIOl/0tykHt/ev+LcJJSUl85t1i/H8GsA5z/y0ZDAYbdYPGyMen9i/MBAAAaCysfuXo9Th48KDFLUYPP/ywHn744du2/u3im4PW28oOoHGxl89BAAAAAADq47YIR7t27aoFCxY02PoAcLvjcxAAAAAAgOpu/g3qAAAAAAAAAOA2dFtcOYr6+dcf/G3dAqygru8j73fdHD+ZYesWAAAAAABAA8OVowAAAAAAAADsEuEoAAAAAAAAALtEOAq7U15erqSkpBuqYTAYFBgYaLV6AADcCAcHB1u3gEbC0ZE/HgAAAPvCTz+N1OOPP65t27bZug2zFixYoMTERJvVyczMVFxcnAoLC2+4h5tRDwCA+vLy8iLUwg1zdHSUp6enrdsAAAC4pfhCpkYoKytLqampGjhwoK1bMSs1NVV33XWXzer4+/tr9+7dNzz/zaoHNGYJCQm2bgFolFxcXNS0aVNbtwEAAAA0OISjjdDWrVv10EMPyd3dXZJ05MgR/fnPf1Z2drbatm2rt99+W+3atZPBYNC4ceM0aNAg/fOf/9T48eMVGxtb7dgzzzyjEydO6PXXX9fJkyfl5uamcePG6cknnzRb45lnnqmxtxkzZmjnzp3au3evPvzwQ61YsULt27evsUej0aiFCxcqPj5eJSUlat26tVatWqUXX3zRbJ0rDAaDhgwZopiYGO3cuVMlJSWaOnWqHn/8cRkMBvXt21eHDh2yOM7Sa3e1a+sNGzZMPXv21Pfff6+ioiJNnDhREydOrHM9oLEKCwuzdQsAAAAAAFTB/qtGaMuWLYqMjJQklZWVafz48Ro3bpySkpIUHh6uV155xTR237598vb2VmJiop5++mmzx8rKyjRu3Dg9/PDD2rFjh5YuXaoFCxaYrpY0V6MmCxcuVP/+/fXiiy8qISFB7du3t9hjcnKy1q1bpzVr1igxMVHz58+Xs7Oz2TrXMhgMCg8P16ZNm7Ro0SLNmjVLx48fr/O42l67mqSnp2vkyJFKTEzUypUrNW/ePBUUFFx3PQAAAAAAANwchKONzLlz53TgwAENGjRIkrR3715VVlYqIiJCkhQREaEff/xRFRUVpueMGzdOrq6uVb7M4epj//3vf+Xg4KCRI0dKktq3b68nn3xSmzZtslijriz16O3treLiYlOo2bZt23rV7tOnj1xdXdW9e3eFhIQoPj6+zuPq8trVJCgoSM7OzgoKCpKXl5cyMzNvqB4AAAAAAACsj231jczWrVs1YMAA0830s7OzlZ+fr8GDB5vG+Pn56eLFi3WumZWVpYCAgCrHAgICdOjQIav0bKnHTp066Y033tDMmTPVvHlzxcXFqW/fvtc1T8eOHXXu3Lk6j7PGaydJbm5uMhqNFut5e3vXqyYAAAAAAABuHOFoI/P1119r+PDhpse+vr665557tH79+mpjDQZDnWq2bNlSGRkZVY6dOXNGbdq0ubFm69CjJMXGxmro0KHavn27Jk2apI0bN17XfTqzs7PVvXv3Oo+zxmt3tdrWCQAAAAAAgFuLbfUNXGlpqZYvX67CwkIZDAbt27dPoaGhpvP333+/cnJytG3bNtOxU6dO1WuOkJAQlZSUaN26dZIuB6OfffaZhg0bdl09e3h46MSJEzIajaqoqLDYY0lJibKysuTk5KQ+ffrIy8tL+fn5Zutc/Vpccfr0aUnSoUOHtGPHDkVFRZntydw4S325ubnJxcVFubm5dV63Nd4LAAAAAAAAWA/haAOXl5enJUuWKCMjQ/Hx8erbt2+VLdoeHh5aunSpli1bptDQUIWGhmrRokX1msPV1VXLly/Xv/71L4WGhmr8+PF6+eWXFRwcbHZ8YWGhgoODq/26Eq5GR0dr7dq1ioiIUHJyssUeMzIyNGHCBA0ePNj0rfJBQUFm61z9WlwxZ84chYeHa9q0aZo/f36N9yw1N85SX56enpoyZYpiY2PrHHBa470AAAAAAACA9TgYjUajJCUkJCgsLMzW/aAWlt6nTz/9VAEBAdd9T87GxGAwKCQkRGlpaVYZh9vf8ZMZau9f91s9JCUl8ZkHAAAAAICd456jjciTTz5p6xYAAAAAAACABoNt9QAAAAAAAADsEuEoAAAAAAAAALtEOIpGycfHp073Ea3rOAAAAAAAADQ+hKMAAAAAAAAA7BLhKAAAAAAAAAC7RDiKW8pgMCgwMND0eP/+/YqIiNCgQYO0cePGOtUoLy9XUlKS1fqwRj0AAAAAAAA0PISjjdTjjz+ubdu2SZIWLFigxMTEG65prTpXW7x4scLDw5WUlKShQ4fW6TmZmZmKi4tTYWGhVXqwdj0AAAAAAAA0DISjjVBWVpZSU1M1cOBASVJqaqpyc3NvuK616lwtLy9PQUFB9XqOv7+/du/erSZNmlilB2vXAwAAAAAAQMNAONoIbd26VQ899JDc3d01Y8YM7dy5U3PnzlVYWJhOnjwpSTpy5IiGDx+uQYMGacSIEUpPT5ckGY1GLViwQJGRkQoNDdWIESNUXl5eYx1Ljh07plGjRmngwIGKiorSN998U+X8rFmz9PPPP2v27NmKjIzU/v37TecMBoNCQkI0a9YsRUREaODAgVq3bp3pXJcuXWodZ2mdV7u2XmhoqF566SUNGDBAPXr00EcffVSvegAAAAAAAGgYCEcboS1btigyMlKStHDhQvXv318vvviiEhIS1L59e5WVlWn8+PEaN26ckpKSFB4erldeeUWSlJycrHXr1mm3bTccAAAGQElEQVTNmjVKTEzU/Pnz5ezsbLaOJaWlpRo7dqweffRR7dixQ5999pkefPDBKmNee+013XfffXrttde0ZcsWde/evcp5g8Gg8PBwbdq0SYsWLdKsWbN0/PjxanPVNM7SOi1JT0/XyJEjlZiYqJUrV2revHkqKCi47noAAAAAAAC4PRGONjLnzp3TgQMHNGjQoBrH7N27V5WVlYqIiJAkRURE6Mcff1RFRYW8vb1VXFxsCiHbtm17XX3s3btXnp6eeuyxxyRJ3t7e17VtvU+fPnJ1dVX37t0VEhKi+Pj4Oo+ztM7aBAUFydnZWUFBQfLy8lJmZuYN1QMAAAAAAMDtx9nWDcC6tm7dqgEDBsjT07PGMdnZ2crPz9fgwYNNx/z8/HTx4kV16tRJb7zxhmbOnKnmzZsrLi5Offv2rXcfOTk5atOmzXWtoSYdO3bUuXPn6jzO0jrrw83NTUaj0WI9b2/vetUEAAAAAACA7RGONjJff/21hg8fbnGMr6+v7rnnHq1fv97s+djYWA0dOlTbt2/XpEmTtHHjRrVr165effj6+ionJ6dez6lNdnZ2ta33lsZZWqfBYKj3/LW9bgAAAAAAAGhY2FbfwJWWlmr58uUqLCyUwWDQvn37FBoaWmWMh4eHTpw4IaPRqIqKCt1///3KycnRtm3bTGNOnTolSSopKVFWVpacnJzUp08feXl5KT8/32wdS0JCQmQwGLR9+3ZTn1lZWXJzc5OLi4vZb72/ei1XnD59WpJ06NAh7dixQ1FRUWbnMzfO0jot9VETS/UAAAAAAADQ8BCONnB5eXlasmSJMjIyFB8fr759+1bb4h0dHa21a9cqIiJCycnJ8vDw0NKlS7Vs2TKFhoYqNDRUixYtkiRlZGRowoQJGjx4sIYMGaKYmBgFBQWZrVNYWKjg4OBqv9atWyc3NzctWbJEixcvVlhYmCIjI5WYmChPT09NmTJFsbGx1b7x/uq1XDFnzhyFh4dr2rRpmj9/fo33QDU3ztI6r+6jrgGnpXoAAAAAAABoeByMRqNRkhISEhQWFmbrflALS+/Tp59+qoCAgOu6R+jtxmAwKCQkRGlpaVYZh8bv+MkMtfev+31uk5KS+MwDAAAAAMDOcc/RRuTJJ5+0dQsAAAAAAABAg8G2egAAAAAAAAB2iXAUAAAAAAAAgF0iHMVtycfHp073Ea3rOAAAAAAAAOBahKMAAAAAAAAA7BLhKAAAAAAAAAC7RDgKAAAAAAAAwC4RjgIAAAAAAACwS4SjAAAAAAAAAOwS4SgAAAAAAAAAu0Q4CgAAAAAAAMAuEY4CAAAAAAAAsEuEowAAAAAAAADsEuEoAAAAAAAAALtEOAoAAAAAAADALhGOAgAAAAAAALBLhKMAAAAAAAAA7BLhKAAAAAAAAAC7RDgKAAAAAAAAwC4RjgIAAAAAAACwS4SjAAAAAAAAAOwS4SgAAAAAAAAAu0Q4CgAAAAAAAMAuEY4CAAAAAAAAsEuEowAAAAAAAADsEuEoAAAAAAAAALtEOAoAAAAAAADALhGOAgAAAAAAALBLhKMAAAAAAAAA7BLhKAAAAAAAAAC7RDgKAAAAAAAAwC4RjgIAAAAAAACwS4SjAAAAAAAAAOwS4SgAAAAAAAAAu0Q4CgAAAAAAAMAuEY4CAAAAAAAAsEuEowAAAAAAAADsEuEoAAAAAAAAALvkbOsGUH8JCQm2bgG4LaWnpdq6BQAAAAAA0IAQjjYwYWFhtm4BAAAAAAAAaBTYVg8AAAAAAADALhGOAgAAAAAAALBLhKMAAAAAAAAA7BLhKAAAAAAAAAC7RDgKAAAAAAAAwC4RjgIAAAAAAACwS4SjAAAAAAAAAOwS4SgAAAAAAAAAu0Q4CgAAAAAAAMAuEY4CAAAAAAAAsEuEowAAAAAAAADsEuEoAAAAAAAAALtEOAoAAAAAAADALhGOAgAAAAAAALBLhKMAAAAAAAAA7BLhKAAAAAAAAAC7RDgKAAAAAAAAwC4RjgIAAAAAAACwS4SjAAAAAAAAAOwS4SgAAAAAAAAAu+R89YOEhARb9QEAAAAAAAAAt5SD0Wg02roJAAAAAAAAALjV2FYPAAAAAAAAwC4RjgIAAAAAAACwS4SjAAAAAAAAAOzS/wcJ1ejVIEK52AAAAABJRU5ErkJggg==" /img>

  • create a new keras pipeline including a vectorizer. you can setup the number of neurons and the activation function per layer (the last layer will be automatically adjusted to fit the label-dimension). if doc2vec is deactivated, tfidf is used. In the image above is the configuration we used for our final classifier.
  • you can also save a trained classifier and load it later again by a .pipeline configuration file (either by selecting them in the file selector box or by give the full path to the clf_file box)

train classifier

  • if sample data and classifier are loaded, the classifier can be trained here

Test single predictions

  • in the playground tab you can predict single sentences and get the nearest emoji in sentiment space of a given emoji set. Also you can plot the predictions of samples with given labels (as emoji) of the validation set of the currently loaded twitter data

Evaluation_sentiment_dataset.ipynb

this is just a notebook to perform and plot the predictions on our hand labeled validation set