{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Create a Learner for inference"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "hide_input": true
   },
   "outputs": [],
   "source": [
    "from fastai import *\n",
    "from fastai.gen_doc.nbdoc import *"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In this tutorial, we'll see how the same API allows you to create an empty [`DataBunch`](/basic_data.html#DataBunch) for a [`Learner`](/basic_train.html#Learner) at inference time (once you have trained your model) and how to call the `predict` method to get the predictions on a single item."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "hide_input": true
   },
   "outputs": [
    {
     "data": {
      "text/markdown": [
       "<div markdown=\"span\" class=\"alert alert-info\" role=\"alert\"><i class=\"fa fa-info-circle\"></i> <b>Note: </b>As usual, this page is generated from a notebook that you can find in the docs_srs folder of the\n",
       "[fastai repo](https://github.com/fastai/fastai). We use the saved models from [this tutorial](/tutorial.data.html) to\n",
       "have this notebook run fast.\n",
       "</div>"
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "jekyll_note(\"\"\"As usual, this page is generated from a notebook that you can find in the docs_srs folder of the\n",
    "[fastai repo](https://github.com/fastai/fastai). We use the saved models from [this tutorial](/tutorial.data.html) to\n",
    "have this notebook run fast.\n",
    "\"\"\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Vision"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "To quickly get acces to all the vision functions inside fastai, we use the usual import statements."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from fastai import *\n",
    "from fastai.vision import *"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### A classification problem"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's begin with our sample of the MNIST dataset."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "mnist = untar_data(URLs.MNIST_TINY)\n",
    "tfms = get_transforms(do_flip=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "It's set up with an imagenet structure so we use it to split our training and validation set, then labelling."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = (ImageItemList.from_folder(mnist)\n",
    "        .split_by_folder()          \n",
    "        .label_from_folder()\n",
    "        .transform(tfms, size=32)\n",
    "        .databunch()\n",
    "        .normalize(imagenet_stats)) "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now that our data has been properly set up, we can train a model. Once the time comes to deploy it for inference, we'll need to save the information this [`DataBunch`](/basic_data.html#DataBunch) contains (classes for instance), to do this, we call `data.export()`. This will create an 'export.pkl' file that you'll need to copy with your model file if you want do deploy pn another device."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "data.export()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "To create the [`DataBunch`](/basic_data.html#DataBunch) for inference, you'll need to use the `load_empty` method. Note that for now, transforms and normalization aren't saved inside the export file. This is going to be integrated in a future version of the library. For now, we pass the transforms we applied on the validation set, along with all relevant kwargs, and we normalize with the same statistics as during training.\n",
    "\n",
    "Then, we use it to create a [`Learner`](/basic_train.html#Learner) and load the model we trained before."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "empty_data = ImageDataBunch.load_empty(mnist, tfms=tfms[1],size=32).normalize(imagenet_stats)\n",
    "learn = create_cnn(empty_data, models.resnet18)\n",
    "learn.load('mini_train');"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "You can now get the predictions on any image via `learn.predict`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(Category 7, tensor(0), tensor([0.6870, 0.3130]))"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "img = data.train_ds[0][0]\n",
    "learn.predict(img)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "It returns a tuple of three things: the object predicted (with the class in this instance), the underlying data (here the corresponding index) and the raw probabilities."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### A multilabel problem"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now let's try these on the planet dataset, which is a little bit different in the sense that each image can have multiple tags (and not jsut one label)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "planet = untar_data(URLs.PLANET_TINY)\n",
    "planet_tfms = get_transforms(flip_vert=True, max_lighting=0.1, max_zoom=1.05, max_warp=0.)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Here each images is labelled in a file named 'labels.csv'. We have to add 'train' as a prefix to the filenames, '.jpg' as a suffix and he labels are separated by spaces."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = (ImageItemList.from_csv(planet, 'labels.csv', folder='train', suffix='.jpg')\n",
    "        .random_split_by_pct()\n",
    "        .label_from_df(sep=' ')\n",
    "        .transform(planet_tfms, size=128)\n",
    "        .databunch()\n",
    "        .normalize(imagenet_stats))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Again, we call `data.export()` to export our data object properties."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "data.export()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can then create the [`DataBunch`](/basic_data.html#DataBunch) for inference, by using the `load_empty` method as before."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "empty_data = ImageDataBunch.load_empty(planet, tfms=tfms[1],size=32).normalize(imagenet_stats)\n",
    "learn = create_cnn(empty_data, models.resnet18)\n",
    "learn.load('mini_train');"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "And we get the predictions on any image via `learn.predict`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(MultiCategory selective_logging;cultivation,\n",
       " tensor([0., 1., 0., 0., 0., 0., 0., 0., 0., 0., 0., 1., 0., 0.]),\n",
       " tensor([0.1923, 0.6070, 0.1862, 0.3969, 0.0471, 0.1786, 0.0220, 0.2634, 0.3177,\n",
       "         0.2928, 0.1829, 0.5932, 0.3870, 0.2521]))"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "img = data.train_ds[0][0]\n",
    "learn.predict(img)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Here we can specify a particular theshold to consider the predictions are a hit or not. The default is 0.5 but we can change it."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(MultiCategory selective_logging;primary;partly_cloudy;cultivation;clear,\n",
       " tensor([0., 1., 0., 1., 0., 0., 0., 0., 1., 0., 0., 1., 1., 0.]),\n",
       " tensor([0.1923, 0.6070, 0.1862, 0.3969, 0.0471, 0.1786, 0.0220, 0.2634, 0.3177,\n",
       "         0.2928, 0.1829, 0.5932, 0.3870, 0.2521]))"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "learn.predict(img, thresh=0.3)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### A regression example"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "For the next example, we are going to use the [BIWI head pose](https://data.vision.ee.ethz.ch/cvl/gfanelli/head_pose/head_forest.html#db) dataset. On pictures of persons, we have to find the center of their face. For the fastai docs, we have built a small subsample of the dataset (200 images) and prepared a dictionary for the correspondance fielname to center."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "biwi = untar_data(URLs.BIWI_SAMPLE)\n",
    "fn2ctr = pickle.load(open(biwi/'centers.pkl', 'rb'))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "To grab our data, we use this dictionary to label our items. We also use the [`PointsItemList`](/vision.data.html#PointsItemList) class to have the targets be of type [`ImagePoints`](/vision.image.html#ImagePoints) (which will make sure the data augmentation is properly applied to them). When calling [`transform`](/tabular.transform.html#tabular.transform) we make sure to set `tfm_y=True`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = (ImageItemList.from_folder(biwi)\n",
    "        .random_split_by_pct()\n",
    "        .label_from_func(lambda o:fn2ctr[o.name], label_cls=PointsItemList)\n",
    "        .transform(get_transforms(), tfm_y=True, size=(120,160))\n",
    "        .databunch()\n",
    "        .normalize(imagenet_stats))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As before, the road to inference is pretty straightforward: export the data, then load an empty [`DataBunch`](/basic_data.html#DataBunch)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "data.export()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "empty_data = ImageDataBunch.load_empty(biwi, tfms=get_transforms()[1], tfm_y=True, size=(120,60)).normalize(imagenet_stats)\n",
    "learn = create_cnn(empty_data, models.resnet18)\n",
    "learn.load('mini_train');"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "And now we can a prediction on an image."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(ImagePoints (120, 60),\n",
       " tensor([[ 0.7982, -0.5515]]),\n",
       " tensor([ 0.7982, -0.5515]))"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "img = data.train_ds[0][0]\n",
    "learn.predict(img)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "To visualize the predictions, we can use the [`Image.show`](/vision.image.html#Image.show) method."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAANMAAACgCAYAAACWsVm2AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvhp/UCwAAIABJREFUeJzsvVeTJbmZ9/cDkOa4cm1megx3yCG177sRCkkXitUn0L0i9PF0o1t9CZmIXW5otUtyueQM/bie6Wlb/phMAI8uAGQi85xTVbNixFutaHRUn/SJBPDH4x8oEeFdeVfelf/vRf+XrsC78q78/6W8A9O78q78lco7ML0r78pfqbwD07vyrvyVyjswvSvvyl+pvAPTu/Ku/JXKOzC9K+/KX6m8A9O78q78lco7ML0r78pfqRT/pSsA8L/9r/+LKAVKAQKocFxrhdYKYwpmsznJW+Plyxecnr6haTdoDSBUVQkq3FMUBcYYjCnRSqOUQSmN1mFbK4UxGm36uUSHB6GUQimVV6M7np8Pu6nC6RqNyu9SaVu6+1MJ3zL2PhFEwp/3Hu89yhhEaZbLawAWiwVlWQHgnKMoCpzz8R168B2Q6qq7eqdzGo348H7vPUVRoLXu3q21xntP0zQ45yjLclCvfX/W+VA3P7zWOdf9WWu77dQWXgTprhe8eMT37UFsLR2/QbK+MMZQFEXX72m7qirKsqQoCsqy3PoDKKuKsigpyvz6sJ9+Hz1+Muy8PeVegAnR9ENXMkCpwfE0GAIo8vM+G/zhWDdW0wBCgfSDK12Tniki2Xb/+n2gCk9UiCID0KjN480qPiiBlK5afd2kOyOAR8WJAaVBKbQ2ABiTBj1oHfbBISJoncCUA0dnwOrrX+iiw7KIYIzpwBTeY7o2ERFms1kH1F2gEhGcczRtAIh1bieQ8r8x2PZd770fNKmIhE6KfeWcxTnLZtP1UhgFOnx7DqZ8Gxgcq6pqJ/AePX7CXcr9AFMHitFgFgWiEa/iDBUaVZtIPbwHrbOBw3AmExkAZutdGWEI474fdAIoGcFDJQD2j+qBlGC3C1AZAOODpbszDnIS0CQCL2wrbXAiHZi0NmidBjqRgoRtY3QET6SQETwKk22HUhQVuV9mAFNoZ0UPJmstxhgmk0mgaFl7522d9jetBcBat0VlEzByymSt7f7GIEvbiYLloM3P7QKdFxBnUShs27LZrAOXonXkeFJ7hsk1gShRtQSkqqr47/77/4G7lHsCplDSTNgNT4FAQcIg7sCkFNponHcoD/Wk7AZfvBMRn8324TnDd2VjPN2q8vduM2FqsDWm/Gr0wB3fRwR8Tssy9qubbVEodFYpYUhp+nOBGqVtkwFJdc9WZNuRSpmi7ICQWCWtFF48WkUWWATT2jDIyhKtdbxOo7RCR6qVt8EiDtIEogSoXaBK223b0jQNm82G5XLJZrMZTIgJTKkN9lG6m0BmW9uPr5zl1RqdWGGtOsAZozHaoI25sU/zcq/AlEpgnUbHJGfzTEZtAj+ncnJC9qc8oCOZUdlxwStBqyhjDFitnOLcUM8R2rTaBlmOL5WAkfYytitcqwasVToW2NMkC/UscbhXo5Sg0Ch0/J6hnJTe01GtSGHSTJ72tdbg1YDl06bAFCXaFIEqGhPkTTUEqFYKpTVFlOeC/NYP2l2sYU6d2rbl+vqa8/NzLi4uOrZVRFiv1wDM53PKstwpe40pG+wH3ZjaBWrZdudSW2k9lKtvK/cGTDlYdg7jLcVE/5GJnRvO+dv3JRCBIEoQ5UGnQRsGSGLdZECeRpqRjhXbrun42C6w7DoeqGm6ZiibJQqUrlUZaAMF8xEkJjuvB4N9/Een8ellizCA6Fi5vK75xKE6dkn31ErrYb10kPX62b6nkqE5e8rjve/ANJlMmM1m3bvbtmW5XALw+PFjiqLowNe2bQeiHExj0Nwkh+26bkw971ruDZj2lW6mptfkKK0HwrB4Aa0G98Strecllke2zklPvTI8iKeTI7SWMEiifDNWTsBYm5dtZpq0cY36PyFRmvAJHqVMvCr+JjZPJJOPeo3lgBntwKMz6pSoXGIVY52ioiOwBeEaIbLV2kTg7ZeVum9R4/frDqy5vAURcIAXjykrtAmUr64nEHuo2Wx48OABAB9//DFAB6L8d7zdNE23vY8FhJ4y5bKYtXZAPe9a7gWYcqp00zWpM3XWOanRk0B+h7cBucCfyp6bg1ajAw8qqc0D6JKKNjxaI+ih9i97dF+/4WDMqjV+eSbzhIGnlUERBnqiRlpL1yZD0G6rxSGwYypjX5Q2kZLowBGroLAQUShtMEWBMUWQIyIV0hnLeFvfpfqMr00U0sQJozBBuVJVNSB4H1i8xGpN5wsQYdLJYVEmy8CwC2T5Xw4yoFN+JOBsNhvW63V3PueAbiv3AkzbJQzXwYDLNrXqWQwRNwDalubuhn7ereXL9iNF2jPOt56lRA0oU6avG/z2Mh0BqJK0SoHtCINj+KIxm9fbzcbyUQJRBiwyto8EFkPEJ8oY0CYSRxX2AVEuyExlGeUlHbWJPduG1h1LRwJlqmd3Xf/bcXkQtLFECpUN2qIoumd476nqOrRBpgyQru3ixOQ9hQhVmqhGio9kw8qBBYHKbTabDmTL5ZLr62vW6zUi0tXlLuXegWlblb19XGf8eq6CHZfEse2eOMOgE0nPTPeoTEZIHSxbs+o2j3cnstixrAlYqd5ah5nf2Q1t2yIilFWJF4v3LdrErlISODJRAwD19iXoKZruZn+F6pUtyYjdUUyTnQv7kmppCpQuUEZH0CXgKCSxgxHM0mkaAZ3AZAKQkgyVAB4HebIDaPrnaZ0pgpSiTAM6yS9qqA0lgjq7aQtsiXpVEoCVgJwbpp21rNdrlstlBzbzNmvzcl3BQHbJAJOANLR3JI5MhsfzZ2d8f3rbgP3Zw64MlQG7wX4bmHpWMA2AnVfFehNtRxrnW5y3mCINdh8Gi3KIEjyeKNClxw+/KcpxgzoHotvpFQO50H0NlI5McFDKiI8mPwWi4qBH9/dlMld0qkB7gsSlJN4Yp6bE9ypFEP0kUqwAxg6siaIojYkGVrI23MMvjFqyl8sYgSJXvdfTqGF0llnTMJvPES+divyu5Z1v3rvyrvyVyr2gTAMFxJAIZEa//vqBXYQ0ywyvCdftoCLda3Zp9PLLVDeD5zeO5awBtVNqr8p86/2ja27bz0704l2yqWnA9EZtQSLr5hFFpBrZM8caSyR+a26LU3iR6NKUqFjqlEimSOQqUBRTVL18qJLcFNkwAtVSPrxRRxYwsBS6e7+kflEKD8G+VQTKJNpk1ZaMOI2287KD/af/kjCWEhspJVVZUdeTML6KYp+MsLPcCzDtLGM2Lzs1BhNCx+bl1+QAuHubqOyv4zh2aK3667pz27rybe3VjnO5PSd/j4iEsTqywQ3kJOnr2tUrgUWDKEFpiWxUbC4FZmSny99PfLYXj9Fl9rzEnqlOoaCTC1N0c8q1iT07nmoWWavYsCJ9k4kE8HrpRaPgLWZIDJT3eR/3/UOmaFLhg9hdctthVpeoaEr35WKE0ndn3u4tmJQeUYJRA+3yEdvdhr0NZKA2VgwGbqeC3tLq7ahbB9QdVE/6QdsZKHOt4ghcOe8+Pt5T3G3KnWvuxmC6yXtjl4p6DKY0mJI3QK7w0VF7p7LBlvoiXd+3Uf7O/tvTKZ9kXR1gZp3HeUF80G4KwUl1WNfh9ww5lrSfgWvQFKrrjs4S6dywj5RCmzgpqB8mBd0bMOXAyDs8Hfd+ONjGBsBUhlq9XUbb9BsHcjogPTsSQNCr57cp3r5GVuNx3femGoV15JySJGW5jAaOD54aAwNp1gZKk9vnOy3kqP3GavPxJLDLVpSHYuTtPZi1B30mIMH21T8TOgoguatU0DMMnEzogZC0f1qgKMvd1CFvk6zO44NbYOs7PPyvdNa/ikzFeTO7vqPcGzDdVFIH5DNosjUNZ9XhIIgT+wAQQtzvBlXqqH6QJZ/ubpYb1mZn/cJvN/fd8DF9PVSqj5JeflOxQ0XtHiTk36tRufdBOt7VKQ2IsT8fHSXJj/XPVZ1XwA8ZUCrNR4RPMAkUSuWcVe+1EjV5yfgKdEbhEKflojZtRx3yOudtl5dE9Xc+IRwZG2WVksEl+Xx7W7lnYNr92d3Z/Duz2B2y+XkrBGOHn0Ni6cL+rmbuwTFUQOQhGD3LcNeyT2bq6pDkEZUE9z3f5KPQ3sExq3k3ohM16v31hgC7uaQJKR9suyjYbd/acwIMvqcHstC0LW3TorTCaIOzLc62GFOgCpOR4/75yT445j6GJoC8DrvlqDg1ZY/ez1LeVu4FmGQ0u6ZjuW1G1PDjcuG2YxdGSojRW4a7SnW+fmF3NwUaimy5jEUiK9nx3WDZV9LgzFnb3rOjj1fqlA0kdkl1Xtv73jeuz1iW2aUYyeuQy0u3geiHskM5i66UwjuH9y66FQmbTfA+qOuaui53E6Z8mkyKjIytTBPf/jr3T8rvyp+Xzt718+4FmCBnxXa2XCDlOZs3oExDWSvtp2t/WOkHYV+f0UAcsIbDa26mrduDuAeO6jywk9VdaaLW2fTuRKOnSxRAhsqVcbRtzgLuBuG4Xoky3fQNN7XtWLGSTxrje/Nrk4/cLkVJfi3kIeyjaAEB1NAVbJfZZG/d92zfVu4NmELZ/YW553gqSWbq5SnYNZQHSgZgW7Dc3t4+P5inSENaBvJJv98JuQN2MLBtqYb5jJ9ryqCPctVGITEHhtbRXy3zfQv3mM4Np3tbNhC3BqXaz27m22Mw/Ueo09gbJf9NJUXeJrZPJET3lmW5931jIOZAHf8qdnEs+ZjIFFsZS9+xfz9gLr5nYLqhyHB20dGDuW/YbSB1pzKKlx+SOwGpp1LRw607p7JX9oN1m2cnTQQd2GLWCvEZpxi1fToDl9Gg+2C99J6h/KN7w6hAUjYMKUC6Nm+BvJrb4Mo1ebsG9T7qNAbNPg4hHU+e3ul8CpNIwYG3lbwe48DKve8GejlbAcOYpazXIxLLW+sB98idaJej6uB4GoVxN7FE+cw3VIPmIOjHdT/5hMG1f8aNAzQbuFuAG7CBu2exrWfHZwrBzpIq5xFaZ/EiITxC9/VL36mUykIfshCIjt1VA8qTKxyGoNo/46e/fWzeXVi7VBIodp3LKWDaT+/LMw2N46Z2sYfjOuR9uvv6NAvmXAf9ORVU/QqH4i2LZ7pbGdpb8ghPGGq8bmRH0nm5Vbrp+HCJ5KPTtKleuN0V07NvJr/pmu786E8RMhAlO1swKIZvttYGIEnP3g0/aRvgO6eNDERJ+TA2wN7E5t2FxcspRy4nQq8O70LlM9lxX5vlIB1Tp5vaOKkttlVedNga3bXzm3eVew2mrZkoO5fHNIVr+3v2FbVjsOXn+neNZmQ19ioYskzdfXqsHsjO30C19g1G0NHOkYdP9CHlIiEU0UuvdetNY+mayNAOed697wUGbNZ/RE5K51IdBzLM6Jtz0Kb9XRPlrvfdVLdd9eo5ipzBz9idwQ3wQ4AE9wRM3phOdkjEV0dSPPicDDAhUK3X6OWUCe7W0Crn/8IRUhBdf0R1r75pQO6UmfL97N5+M/Ge20bHngLmLMn4uxQxjqGrY/+OjCqQD+Ze9soD31IiSmCgTRvXZ1/ZxYLldRhzEDmYEqC892w2G7TWVFU1ePdNZcu3cEfp2lLtAs9wnKUWv413GZd7AaYknEskw0rREWOlJNhUkjAIeAmhztoU3UydN9DQyLn7nd0svuP4YFvSIE7Hds/Wu56XA2xwT3c8uzjKSAHPqgNv0ETtkgGGwEgDYqxU6VhHCRMFg2r0gz0f4NbaLW1jui7dtw84t5WB8Vn6NF5pu2maLkHk+J59LN8+Krj7fh+UWZ3qTnVavPH0HQj8W8bmDWfgpM6O7kGEtB5KgY3D1QoYNEU1xSuNV5EdEk2hCwRNcOX/IfNKP/vvq2MawHvZDTXaZx8QRrOtygZ99iylFfj4jMzxtydk4/ru2E8OgKr/vtsG/1gOyUE3vk5kf2j3bSx3Uj4ksKYI4+BGZAbX3lTH8TX7gR5VwgqU5M8cb6sBl3fXUXQvwLS3DOis7CAzIfFiAo4XBbrEK8GKAjEoQgKSEF5tbgTaWGDuXL46FuuGuqodwr7KFRbbLOhtbNSgiwd580J77GRrVPxPfphXRl7PPnfc8Np9rNw+cObfnctGOdXLqZC1dpCieFf5IXXYUoLs/vL+VyJlV4qoGOXuULrnYBIks+VkmpiMpKckG3k7dqwE6T7VhVuLMngVMvEo/E5BVHXAiDO7GqB6qyiVMQe3DNZ9+/vuSdRQD8Akg/P9dgJ+um9IXW4a+Om6Mcu3r9yklh5fk4Mqr0eSjVIOBmMMdV13jq43PX/XhJRfs5NySezvjOwMQvd7BAF0SvG7guRegWlsqd6WhdJvLxMZndiB2GGjPr5Ndhpd3QEpE00QgRDFmp/bIkNbT9s9W++4l212sFcWJMPsLpPgNgspO4C0q4wHfW4U3mVjuon63HT+rvc656iqagC68b03Afy2iSLWhG21QnxP1349wO40ZLJyb4y2+8tufjk1XG/QjMfTP8nZwn3NchtlSIM/1wAlB9T0O6JMMn7GbXLNTe/PvB4Yq+fpjg3eseP9PYW73VaUs3ndIzOlwfh339+uMjbA5s9IdfDed8baW80cO47d/u4hJepYucjiCb0iSakf0ltvBZjelXfl7ShvJZjymaxPTBjnkFto876Z6zYWouOp+6MdyzbQ5o0es63W3Z6dd9Vhi4oMeM94rRp/7vCavqR8drezS/sowj5N3q5z+fldFCu/fuxtkWSnXdRv8O0Mqem+53ft3x0gpC7zIS5MyBylZbhKVv57l3KvZKabSlKXh+2+kUyWfDEPbd/xhO1Dii5MWfJx2Mk18eDOwTIat3vG5y42Lxf08yT524Ny264zEOYHLIkiT83cy1w3ezHk9Uws3j5FwvD7twfuPq1lum48+IHB+1Ku731q8Vy5sKseu5QR/a+gxAdtrvSOrQqw1mMleOkXhena1Mc885i7MXv3Aky5YuGmfu/n9rHMlD9r/2x5a1EqGkx7ZKmk+RkTJxV47FyOyQf2eF/yX0ag2KHtQgQlDq3CbJlsMt4L6CLsS0a9kvvIFsr3l/HAz4X4uwBwF7W66fl9OoFwPF9pwhiDc26vJi9/5k3KjL33iEvpjggZa4l1EEQptApJXJz3OBdSnFkrOAf1/G5ZXe8FmIZlqMGD7fGRd9I4NCEHZALdfrDuepfE5PgZ6wYd6ZK02WmB0mW9tJq0QUptAyzck1xbdrBlOcuYaZX6bwgUxyuN86BU0QG00GHtJJSE3BCZCjinCjd5g9+2hMqYgt6Fbd51TwJXTpnSKoXjlMR3AfZNRSmF8gpU7yTsXJx0VQjnUVpwXlguW9Zrh3OWb775hu++/Y7/+X/6H+/0nnsDpryz8wG+qyETm5coUw6KMAuNO/l2rVCn8crEyP7d/fNzlXV/IZF4bYMjt1kF6tQ/L32pj3+K4JPoI9AGLaFGv9nxwoRkiSKCJyz+PFz0LfzftC1lVQWZRIVw8dSeHrr837tkkv/IgN4lEyZWUmu9tQZSWvD6LmWfKnwvR6KCd33KHeHiQta6CP2z3rQ8f37KH/7wZ968OeXq8pIXL17w7dNv3i4w7TO09SzHblAF7+KUVy2b+e78YjUYrbkKuS8juxPb16h0wV619LZyQbLFqlP90/sT+yjDh8Vv682yA4UEob1s26IRyqqiMClHRly9XAWq6ABEurzggo5uS9AqgxQlXhuUKTt/SQ2DvBNjGWj8jeM+zX93yUtJZqrr+k7g/UGKpDiLibOhybSmKAP1cx6uLld8/c1Tfv3vn/OLX/ySly9fcnF+zvXVJVcX5zfWIy/3Akx5uYnlHzeUUoqyCov6rlbXlGVID1Vqc2NnDFiObMqXTkOXUSqVfOGEfU6uWYX21nPfwEvnhsBi2AgDUCbNU1/rtm3Q2lAWhqpIbjiCdz5QKAnzRlVVeO/YrFpQiiK67NSTmrKsadoW5xu8aLwyEMHknYO4qLJDoSIlMYMJcP9kuK/tc01eolhjG9MPlY92FgXihLa1lHWF0grbhFUuzi9WfPf9C/75//lX/vHnP+dXv/o3vHUhTRkenL3za+4FmEKjJ8F0qJ0JbFuQR3bx3Uab2AHZ8xgO1j7Dj+qE/3Cyl3nSto/baX1a6ViytD2kWuGFmaNqdnjLoMoQWDcd756n9KjOw3s1ClOYThUiPi6M7V2HOkEwRUFpDGIMVrkQ1ZvYPO9REuQs17ahLtLLT+IdWhd4AR0HemrzNPl1mar3sAVjxVCuOTTGDGKoxvn8xsC6iXLto0xpjCmtsa3l1atTAP7y5VN++9nv+Mef/5zPP/uc68tL6rKitS2IRfu3PNJ2S+EQD+5SraakIomCDDm+sbq5H/AigsrUxn0nxPioDAjJKp7km3whaIXqFRIjMEGmrFB0dezJy25WUEUQBWXHEETpPYNkiUlLJSDicd4izlJoQ11XlGVB6xzr1ZKqqim0AtG4OGib9RpnbWA9gbIsKYuoJY2NqlDgQ84KYzRlWWLjGkZde0bv/f5bEsVKjdIDRcTjfVgd0Biw1g0Wpu6fOZxgxgAb9t3u4yp+lykMzntevDzlX/7llwD88t/+nc8+/x2fffY5V1dXIB6cw9kGsQ3av3WUabid1nTdN/uMtUPG5DFNuUNsf33c2npW74UNjFJ4dQAR6N0e4yDpgBZBovr7MhXfQMuX5K9EBVW6NJCeJAjRp0nugaR2ADXd4pyloEAXJiwMJpq2FZrVmuXVRXyPxinFbDZlMplSlfWA2ikdvOk14bgCUrpjYwylNjhncd6hVHFrxtd9Mk1YdyqxdekPmqaNlNUNJyfpY7BylnIw2e5hCcOEGf/pQOFfvHrFL37xS/73/+P/BOB3v/8DX339DW/enDKdTFivlth2Q4FQG810WnHXck/AtFsV3h0Xug6Iux2gEqs3mJ18D54x8DquLqM643Pdcwhs1L6lZ8ZyT68SH/0O5DJBSbdMYdAedmq+yM7GygxMi9kzu9mfMPF47zBVRaEVm/WKi7NT3rx6xfnZKcvlEu8c88NDDk9OmE6nTCYT5osD5vMFAGWd1pD1iJcQ1uJsWAXDGLQuMUZ3RsyU2+4uBtuxLS2xcvmK50n5UFUFIr5vD99T9K5PdP9ckWyyUnSpogeYiqDVhcFax5///CX/9E//xD/8wz8A8OrNKWdn5ygUpTZ4Z4NMKC5MJD8AIvcCTLcWYTALCsPMN2nVgpw9GINor46v50tGLiRJo6Ygi8oMlw7lprTynaIf6N3ZHcDKZ8yeCgZvhVCBYPdIbN5QHkssYqDCRmsOFjOWl9e8OH3Ny+fPef7sO77/9ltevHjO+fk5bbOhni84ODzi6PiYBw8f8v777/PRR2H18kfvvcd0PsM5T9O0HB2dIAjWg3gTVqQoik5jmMCUZ8S9zYCbgymXl1IJBttyADCVvBXyNvTBiBDuzSl36gs1jJjxEv5ay3rd8NvffMYvf/VvPH/+ItxqwuLXCmjbJrCyziHiwHiU71nZ28o9BdPNbN64JF7bezfoNAi8efjtr+8HNn2jD0rk9UljN8RCBTN5LyMluWqgyr6hyvlsPXYjSoZU7/LAPOl4uWT/Uh0bGGrobMt6ueG7r5/ypz/+nr/86U98/eUXPP/uO0Qci/mU2XyO88J6teL0zWu+/+5bvvnykG8/+BqAT3/6Uz75yU948Oghx4cHlIXGlAXOebyE5Smdtd036Mgy3eRlMg4wzAEkIgPKlFY812iSvB9kQr+l2AnPyxZsIC7/Cd1iDSprL4+isQ22bVitVnz1zVe8fvE90q4B0L6g1oJtmjB5eYe4BieeSk85ms/3d+io3BswDdXVdJ21C1BjymO2sujkFCrXIJFNZsm1Z/z87fcpNWZj1OgvsWajGKAOCNk1N5ScdR1WQw2SnZRlGRQGXri6uuLpV1/x2W/+nV//6le8/P57NI6TowWz6YS6qiiMYe2D3KREWF5ecvr6Nd8+/QaAb776ik9/+lN+/OlP+Ojjj3n//SccHj+gKA2t9XjnsbYhDwm5CUjjb8rvySlb8sOz1gb1RNLkRTlVRdkxjzBPk0xnn4oTjiSKryAFUCqtKEuDeM3pmwt++ctf8tlvfsPLF9+D3YTnSUvtBbe6RIlQmuBFUpqCR4sFDxYHd/pOuFdggsDmSJxVojJhpMWDbf67X7GuB82QOiW2aMR67ARSJvjvLEl7sK3Kln5neH02u3Z1Uj316xQQ9NeIxJVl4rHkYtM0DXVd45Vis1nx7OlT/u+f/yP/9qtf8eblCx4/POHJex/w6MEJ8+kU8Z7NasWb5TqotpUG8bTO0jQNAFfnp3z15z9ydXHK5fkpOEddVcwPD/GaKHe4jpXK1a03gWrcd7mxNv1prbv8fxIFRsn6SnVCZN/OeT8n1i4pgtJ5pRTeejYbi0Lz7bdP+eUv/pVvvv6Si/MzitRFoqmM5uhogZEwArV4ZlXNw4MFi3qy9/vG5V6BKXWW1rtC0XufgMgA9C44eyhTvn37RLoLRJk2gNiRMRllfu22ep2d++lYGib9B+ah9TqsOi4eZQzJ0Sg9K6wK7lACz779ll/867/w+WefcXH6hllV8cnHH/GzT3/C+48fUReG1fU1L1+9gmKJdY6wFlLI790WoZ2ttbSra549XbJeLtFKUdclH5WfIKqIC4Il2XHolBva9nZAJYqklOooU87mhWU/u5Yh5y4GbbdTFovvSsCK8qu1lrZtEIGvv/oLX/zlTzTNisWihmhjW69WzKoZB/MZlTbgPL5tWUwnVHWBvpuPK3CvwLSLpes9qMPg69ewdWKwaRGvmOU0sA3BAziFmW8rIvKnEx1l0vt3XRTe36tkVXc4/N5Mzbao1x7WdXwsLMFkUCrk4fb0s7pzjuXVNb/77HM+/+y3nL5+xbSu+NEHH/C3P/2Un/74Ex4cHeJdy7nR2GYDpqZtGlrXYm1Ls2loor7QaUU5mTCdTUEZ/vz739Os1xRFyYP3nqBNGYV0NYiAuwlUu74zeTgGl0vPAAAgAElEQVTkDq4QwJTciMK93VMGbb2tAt9OiS1CZ6ZQSihKzZvXr1ktrzk/f8NqfQU4tA43Ns2KyYMjFvMJlTG4tqFZtkymBfXEUBS7+3VXuVdg6n4z2WavLUNicFdcnU517vOWaV2Ca1FaUBiUikBJ7FZMqDJAzx5lx5gxZGs/gozI2+/A7U2A2munUZKW3wPA2TCTFkVB27Y8ffoNv/vdZzx/9gyjFE/ee8zPfvpj/ubDD3lwdMBiOgEpMd6hxDOZO9brFev1mmazZlOuaZrQ/V489WTCdD5HmYIvnz7lD7/7jMeP3+Pg4BhTe2bzOU3rsE4oy21n4P2uQynWrAdPvp0S90+mU5z3O6akjOUbTWT9+TTnDdu0rCqazQbvHdP5hNlsQl0VTIoF69USgMmkYLGYMJ2UIfGpF4oKqkpTVYbybQPTmHz3y2DuAdKoA402URaITS+pcQXEkzpEQdDGaY3XJrAvZKrYlAgyR3OOji2gZBQpyTdq5537Wb5sts0Ns/33a/CCiz5ixhguzs/5/LPf8vXXX2Oblg/ee8gnP/qIj95/wnxao8WjxQdhej6lLEoOXbBBrVZrNpsVTbPB2iAzeRGKsqSIclLTNnz+xz/xi3/9F5589CP+9u/+juXyGlEGbcqBomQ8MWx7LIQPGrsRJZYv2auU0jG4s78ucCV+a87bOSS09OCNF8oGVqsV08mEJ++9x3/+T3/L9fkpD0+OePp10GS+KDTeWaraoLzHW9C1oSihKKGs3rLV1reVBdu88qCTsvtQw0WLd7N0w1Xm8od0uxm3JglQKguU6/CVOryXIboHdixGX++dtcmOJ8F5WBQoE2UAjeB7cxjC61ev+OUvfsHqeokxmsVszsOTE44PD6iKYHAV26BNybQqmdQzWjFsmgmb9ZrWzvHOIZ0eGnRZUJQlB8cnmKLk2+++5/Wrl3zxlz/x5MMPcUpzcHjCLiFizN7lJU/ID/16TCncIgUEDsEUx0AXW5WZr5WKi0snEAdnr3EYlojCti2r9Yon77/Hzz79Kcp7/tPPfsbhYs7P/+H/AuA3v/411xfnFIVGo3CtRpclugBHQ+tuju8afOudr3xX3pV35cZyLyhT0vCkss/XarAf1d9Bm2cGK+ptFdm7s7s+uw4oQElYGT1P/p7ksWS1koy3Hz8pp5xqGPyXNJNpPxkgO3tKVL40mzXPnj3j66++YlZPmNQlJ8eHHC4W1FWBFol/Du2DzUQUKK0xdUVdaMTXaN3nEDcm5uUzhmIy4W8++pAff/Ijfv/nL/jyiy/40Sc/4ZOf/ozDoyPOzi/wIl2oRM7i7eImxjbBnDKlUHWlNVZ8T2liv/buZP1ztY5OEEjwdldxnatRv2pVU5iS2cRg1w6tCj784GM+eP8JCs96eQXA5ekZz777BsRSKIWrCwqtMEbjvEXJ3SnTvQAT3K4Jyksv9BLYvGxFjO6ujI2T7v9hg+dM2u0VHF2sshMqCEpaem1juCTKULINrt6QO9KIhZ1u3zlPXVXd/sXlJc+fP6dpGhaTKYcHC46PjphNa5R4XNsgGMRpcBZtqi6KtlA6hGuIClrPFDeRuFgdzA/TuuS/+umnfPPdM55+8w1fffUlP/3PfxcS67cNRVHs15DusSulPs1XBoS4xpQphh4TkuTmEbsfn+/jM9KE4FO4SdY/znqqqkIpzdXVGmMMB/MjvG/xtuHHn3wKwJ8//iNnZ6e0myV4GydmhSgJpgT1loVg7JKTbi2q9642xqDTej4qKJH7CAUZ/PzHy11gtw3PBBqV/1PJ/UhCiLlsX6ejVc1ZS7VYdHaRy4sLzs/PwoD2jtlkymIxpzRFCHgTh3EFrYZKgSqLoPVUPkTKRg2NOIf3CcAe5wXRBmlblk3LkydPeHBywtk33/Hy5UtefP8cpcuYzGmbCt1ma8oVDmMwlcUITOm5EmTE3nUIvI+uTRI0m0ap4Hluo6+gAvEKu4FSa5qmYbXaoBXUpeHrb77GuQ1Pv/4SgG+/e07beIyp2DQtzofnifYo5VF3DKOHewKmVPYpH3ZSKumv09pgkqCbGXY7NXuc6e4GqEwT0b1rl0p83713LwNKJcmSHzSZxhmapmVaLFCuSFii2TjWqwZjDG27CT5lzhN5wpBhFsHahqaBaV3hlAtURwPigysSjqQGcd7ROEvrPE4UVmn0BA4PDkCE87NzXr16yfHDR9TTWRi8ezxT8pL6Lfd2yI21yaWoIJg1EpgSRQqmIBmweSKCbRwgOOuwaelOYxAvNG1DuwHXGJr1Bd473rx5w5df/oWnX3/Fs2dPqSrN1WUIRz8/e4NIcDlqGmGzceAdhYaiyNNv317uHZi6Bt1hvxj/SpYnr5ObdkZGytZ2b6TNtHGD7X1lB9j+SqWbTBCceExRImjEKvChU7012MZjW8tkaqjniukUlNqgVUFpAs+PD1GzbdPQKE1RABrEx8A3sR31tt7SOseqabAClDXOXlLXNc5ZLi4vuL5edhQmuTgNfAlH39C1ViY3wXBVwrQ2E1oQXPBySQFeEg2y0k+CRUzs38g6vQzbtjRNg49rO11dXeFtRbM+58WL53z//TNev3rJ6ekrTt+84vzsDZtmiYmhHHUZ7EmFFrSuUFis20RWUnXG3buUewGmXZRnl1FzF8XK1/hJqujEbImkvOM3vX0MpjEFzAxHg/N3k7h+kFwWi4/1ruuS9WpDQUkaUcvrK66vryJ7qykKRVkDqkUpCTY3LxTOUKspalNSTxxaFM5C61wwAIslBTxKZPu0UjSbDW3j8CaEQ7St5erqirZtwqp+RRk9M3Z/VX48dx9K++mvKAqstWB0DFsKigfJktxITGGRQN84x2Z9hW0Ca2iM4fp6zdnpKadvTnn18hUvX74AN8dZxfX1NWfnp1ycn7FeXbPZrPFeg5S0TbDbiXM4K2GyEYVzBueCw21rQ36Nu5Z7AaZUckPgrhlvez/wxFVZBrlJaZwEbwFd5P56I/uFZCGBNwFtACTY1kAoUph7r4sbTQo3oGmf3OEVeC14I7TiqLTvYoeWmzXXy2tEHMaUUU5UYVFAExI4GilRlKiiQimDoUEJWEuQCcwcQXUDxdkGowWhQVxDaxtaFWZ6L8LZ+TkvX71ivV4zPzjsWej8O0cUKIEoXwy6o2yZ90MgQoHaDbKTCUHFKZnvnQi2dfjW49oWkZYX37/mu+++49Wrl1xeXrFerSkweOdpmgaNZlpPKbShMCUKFZ4Rw9FFG7xXWBs4HecLvC9jAkqLyFsMph9yXly0pBMCBFEqqEm9xwiUug+vjtZRYI/6fPBggDFF2r2kS/+7h/X7gWRJRMAonIGVa5BCEDwmLf6lhKZZY22LUhPKsuiW7fRKYbXqstC2ELLC2havHBsLrddQmJAVVoXuF6PRRYHYS2b1Aa69pFku8efnmNWS641lohRlIPeZoXubaxg45GbxS7nslGv2pFBBGxdjlLRousw2AioCCkLiSEUIVlxdr3jx4iV/+dOfePHiJc2mARG8Ezb2mtIUTKqK+XSKtS3L5TWrdc10OmExX7BZrwBo7QbE4Z3DSQidV1FLaO2GpnnLVOM3eT/kFGmXHUOpkLy/LMpeWRG9uju5SHbRjKS6Ds8sonq2bduQySemzBLZBb7/mMx0m8q/K0bhtGe9XlGbGqc8ughgrqoSa1vaZoPigLKsQBSNtSAKhUG5lgLLpPBURmH0GicKRwCQt+ClV9YUxlCUhqquQAyNa2m9o7Utrm2ZlhXvnZwwUSooL2KuwoFLUfrG/mO7b85tSklOSvYmFb0giOkzA9gI3+IDqJJs7KyjriZ48VycXfP1l0/585+/4tWLl2g0ZVUiXlherdhsGqxr0VpTFgVFUXQmlLqeZPFhE5QKS4C6NvjxeW+xtmW9WoG6vnP/3hsw5WxO/jv2X8tLsCuFWaSsqpDEXwSlTJilUh62gXmUTDsUZjIAq4J7izEBZMko2baWup5kCxmPAZbNzvS2kF3lTkCKj/Q4PB50lKGi8bAqC3AO31q0QBHVv64VlBeW5RrbePCGUleUylLqhrKaMJkdoqspm7ZhuW663HFaayZVQVUoSqPw4nEartdrGu84eXjEZD5DFSY49mT90StNYhvkYec6ZECy0TCbNHkqxjA559BiYh6+aPBObF0EEl6R7Ka29SyvzsFDs2mDrslr2o3j4uKcy8sLqspQV/0SnsYUaC341uHWjrZtA3BsG+sZai8iaCUhM1NZUBQTZvOKsprerc+4R2DK/bfSsXEZsxUShcQuR3VRDESg5PfFyArfPYO0ikawd4iExvTes1yu0Np0gnLn3p89ItmLEl9/V6zcVqxYlBYKEw2SSrO+Dl7OrtlQKkNlDKXWGBTNesNms6bZbFBorq/WhCxcmrnxHNQwPzjm4WPD5MBwfrnm9ZtTLi8uw/uahrrUHB7MmM+mzOczxGhOry5oxHP83mPmJ0eoqsTTq65vK7n3g4h0wCqUorUWEEzTokwBqsB7cD42pEjM36B48fwVAN98/ZQ3r06piooHxw8py5on739IXc64OL/k8vKCwmyIhpAw0RpDWZYopWnbltVqRdu23Zq5SRu4aTbYtqW1jsY6ms0mhtO/hTJTrplL+7td+kNRKgR/BUOlp4wW/cTaBR7dbXVoeEbPlhgTmsA6y+XlJW3b8ubNKefnFzx8+JBPPvlkJxUKfHwK47gbivbpIsbfaW2LKYsQNr5swChev3gNwIun33E4m7Koaub1hMqUuE3L9eUl51eXWKu4uFzRtGE1h6NSOJkVHFuhWhzjqwlnFxe8fP2ai4sLANara0oNh5czDg4XPHr0gHI6483lBd4oHrz/HtODBaTJag+Qxv0DvSo8sXgDNyKgEIW0kQNpNd4qTBEWiV5trnlz+ppf/DLkuPvjH/7AyxevKHXBpz/+GceHD7k4u2S92jCZTDk4mCNyEVbSMKaLwFZKgQ774j1N22JbG9vaRmpl8RKiib33rFfrIGdtVnfqW7hHYEpli3rcIGc477leLlmvVxzMZ6QQ50AlNMjY5pQsuInE9Oyf1prlcsmzZ9/z7bffcX5+zkcffcTJyQl1PQH6XG3iCencxtW6RQ+enxrLYgPK6T2lMmgF1q5Q3nD+MszOL55+x4PFIW+qCcfTBYtqAjjEerwLM7v1ntZ5Nq1lIo618bTOYZ1js2k4v7rk7PyMZYzp2ayWaDzONXjlqeYTDuqa5WaDKkqOHj5AlSUuq+euiW4XmHr2WDrWLrDPLYVSiCicC4koq6pGG8V6bXn58jXfPXvKV19/wa9/E8D04uV3nJ+/oW3WPHv9Fz54/0MePXif6WSBKSZo41GqoDQldVVTFYHLsM4F5YUCU5aUVe4OpTqw5+JF8JxY0bxtSShD2HJyjLz74sQ6upKsltfUZdGtb5tTp4CbbTYvcfpNlBuU1kync8oyrBJRFAXr9ZqLi0seP36/q8dwVbRR/X4QmxcZkVylHAFWaYOyDruxFCi0F7QNVLu9XvFgfsDBZMrJIvy2vqUuKmaloykUzhuqWqitZ+6vqeuSg4NDjo6OQFeIc7RNkzxG0Urw1rLZCJtNyWq1wlQTPFDUNYujY6wEO0+hzLDOWRvsUpcnVnucay+x5o0H8YGjcLRcLRsuzq/45ulTfv/Hz/niiz9xvQwUtNmsQYKT7Wp9zfNXz3BiOT46QSsDhQp/Injro8wV+kzHf6UJyqXKBDbPOY+1rktUM51Oqes65BecThHzlsUzlWVJn5ugpa5rikJHkpuE72jQIzBpGoMpDJO6Znl92QW4aW3Cig+SWIykGh9mKlIiELU4AKYomE6nLBYLTk5OmM1mGFPQNG3mNzbKaS4JELu/KxmRb8LYLg1mShxfiMaYmsprJrHzZ1XFrKx47+QB7z94xLSsmOoCxQMm9Yq1g9nM03poHZjzZ8zrikcnj3j84DEb65nVNdO6po5JWqwJauCyUBTagIR2mU7n2LJmcXgY15SKKZt7lV0n3qR2HnuR3+RG5ESFNMzasFyueP36Od8/f83lRTC2Xl9dMZlOUDpQh01zjbUb2jZE556fnbO8WvF9+SIonQqPMRptgu3Pi+Ccx3mHiwk2vQR/yLqsw9gzJYUuKHRBVVYspjPmswWz6RStC5q3LXF/mLmC5my1WiEiFIWhLAuqqqKqSlB02XS898znc4yaIoRMo21rmVR9PodhxtHkMR515NnAzamg1rrLeBqC0Ir4qzvKEYyMKa9fuvMGBcQONHVaSnpTb16UB+MVWoW0xCWKaRU6//jwCDZLalNQaE2lDfVsynQyYTFvWVuhEc26cVytNjw+DIbdB0cnLKZzytby4PCI9cmqM1y6JmwrHcwK06oCpTk4OEQmUw4Pj6jrSQCFhy7PuQTLgo9tE9pgqOkLk2GgAGlSstYG9kssShmsbXn58jV/+MNf+MMf/sLp6TlFqdEalqsrmjawo4XRTCcTikKxXm9YLddsVi2TiacsK85fvEab6BkfjdkiAUDh3cEZWmnNWkVZSPWG9yD+KgpTMKlrpqqi+AEhf/cCTN6HTKJnZ+d8883XnJ6e8vDhAx49eshkMuHo+AilFN9//wyAV69e8ejRI378k09RKmplrIVqQtNY6rrAWs9kMolrCmm2nVUB6VNopQEdll3xvHz5kqqacHJykrF34wck8N6kDr+BNiUEjqhToUs0Bi06OsCaLpXxo0ePOH3xHccnx8EMIDApS0w5Ze4VrSgar1i3ntV6w/vlEZumZTaZoj0YFI9PHlCbgnYTfNzWyyu0Dmu6Ou8opjOWTjhcHFAsDplUEybVhNamSWnbjOGj/jrhTJEoQeC0Qn6O5EYUdIKC4vziksuLNd8+fcHX33zLs2ffs1wuMYUCPLNZFcwBgFIVWgtKTdEnBWen57x5c87V1YpCbyij54N4hzKCMQVlWVGUVZy8VBf3lurrI7fSZUaKhsnNZo0Rz7J5yzK6JnvQwcEhT548Qeugxjw/P+f6+jrmoa54+fIlAF9++SVN0/DhRx9TTapOn1AURTeocwEzKEnDDCRJtRuVCCYG3aW4mKIo0Frz/PlzZrMFP/nJTwJ/r3v/uFRuY/P692f3QFwyaWhDy4sxBUoCtW5aizZQT4O94+jBA64uXmP0EQ5L61q8cxRVQVUaSgqmqmRWK2xdM3VLnA2eBcoL0jpKNMezBXoRAOoOF+AtZVWgCoPVhudnFzx+YHj40ScczA8pdEnjm0Hb5nJSD6p+UHbcgQuOtK1z6KKgdSFuaLNu+Oqrr/nu25dcnC9pW8vDBw+Y1hMur865ur6kqhXzaR3bznJ1fcFqtUIrg7U+hqIoChEmRdXFI+m4mIOKC+JpbdBGU5RVCNeJspDRgXV1XrDORmWFAy/UqsKWbxmYnAu87mKxoK4rFosFy+WSzWZD2246Fe7yOpBmawM72DYNk1lvie8XF44d2zlaSjR8JtkpyEsqsxspCQDbrBqODo75b/7r/xalVCdTMZqRt8s2BdpJk6S3gezy7khU0KOwNuRpKIsSXYfVGKrFFKoCbSqapuG6WaGuhLqpKIuKqp5RloaJrvBFQbsOySbF+Rg3BBttMEXwmAYwkyKAqSywwNpDZUqOFhPee/iYSpe067YDSv8pO2x30rf9lozkPV6E1lpKbXj56iW/+8Mf+OqLb0EMk3IWpGGtQIL/3asXz/EuZFXVWlEaA/Wkk5vEK+aLBXJ1jVu3iA4AU0VYt6rQBo1BiUZ5jXaCQVEoE58Z4+CMwhmPT/KVE5QoJpN6T39vl3sBJkWgRMkmUVVVZ1QzRtO2Dc45Hjx4AMCTD56EFLZVMMZ1LJYKaYS9t8FXzQtKRSVEUkYQB7gkj/K+840xHB8fE04HrVHrbNTUCR43iJfSpBS9MZxPQh12WKW6e3aV8eze2pD/z3qhLCukKKL/GjCpkLqgXXo8jstmSevXzOsp8+kco0sMLUYrxFusCIJBGUNZ1VgXV5sST1qHU6mwujgSnDutCxmfjhdHHB8dIy7YXXxVjqipZJMDsc17AKUJrvN8UH0CyuvLS37/xz/y5ZdfcXZ6yWyyQPtgf9qsN9hmg459d3b6Bgj2t3pSM53OcE6wraVtLGXhKUWhy5rCGExRUpQFhSkihQpKFWsttm2gEEwdekejMWXQ9hmlcCgMjrVtceL4AYG27xKqvCvvyl+r3AvKFOxDUW4RBwTvhbYNavLkGHFycgLA48eP8eKZTaf4eF+wzAd5Q8Rl6toYhjz2glDDHAOJuqX0U9576rpmMpt2s+xQOBK88vTB6AJEn0DSrBd/U8L7faQpf6oINq0nJJa6qnBK4SOPX8ymVAdzVs05q1VD6y1eguZzGtdtF9diW8/GNVwsrzm/uKQRj9ea9bphs1oxrUuqIrAwXjw2hlusreW6dYiuODw64ejoBGNK2h32ul0snovUzjpPUZionnadkqZt22B7KoroxtPgI4fQNBvW6zUK4ehowdFRzXJ1wfV1YPNNVfDo4QOOjx/y7NkLzt5cUhjDYjrjwdHD6KTas/zGhMXfApejOm8XIeWzJ+ZKjN4aIngV1ee6Csblt82dSKI9IAgyiqIoqaoauA4NUgb2wmQGNKODzSc5pRKFem00WEjq6viGzDC6bVgEurS9q9WKN2/e8OrVK370ox9xeHy0ZUTu7u02kjw1xFt6k5PkABvuSp4avSNS7iwrYTFmEwaiKkpa26Ki1/j8wTHv/c3HWFYs12c4cVglnVMshEA/11rOr855+up7zs7OKV+/4fnrUzZNQ6kVHz15j/n0UWxLhXdCYxuuVxsuNg16dsRkvqCezcGYsG7RSPGwyzF5GAgYFmCwzgc5RFI8EwieyWzKZDqhWVvC8qEtShyLxZQPn7yHKeHq8oz1Jnhu15OaJx9+xOPHH/Dbz37H2dkV19cr2vWGKycUpuiWF0KpoFTYNKw2DSlRS1mWVGVFXQcxYjKZRHV7EVXojtWm4WK1DqEe6zvMgLHcCzBZ23s9eNePyIODI7yPTqzGcHkZZqg3b14zmUz4+OOPUUZH9XZcOWLkMLuvbK+AEe5dr9dcXl7y9OlTHj16hDGGyWRC27ZbMk+vYEiGyj3vGntMxPdLd2/UOKqgUq7Y4G2DNgYnYdGxFDJw9PAB9RQ2mzMuzr9HtwotHq/7pP6iDFopWjzLZsP58hp3ecWrs3PapmE+qVnMpxwezACYTUqct2xsy6rZsGotM2OgKGi8D3Fj3QLQYz/HIZhy96EErLZtu2ucc1jX0rRrDg4XHB0dcnF2BeI4PjpkWtcoPM5tqOuShw+POD7+EQAHR4dMZwtWawsinJ+fsdlYvGloiiLISzGSIEUTKNVrc73fpqhKaSb1lEldM5nUwTSiNV4pvIfVarO7U3eUewGmUHrjaFXVCEFNvVq1aC1RsRAa4cWLl4gIi8WC44eP+xxz3ncs1f7X9HFQSveDPLmTpHNt23b12Zct9lbvoT0mpqGPXtQsjuiUdw5TlDSNhbLqEocIBWX1gOOTD5nPv+P69DnaBHeo1rZY3VKXNUVdUtkaU5WI1lgXZv9yUrM4PmJ6MCNlJmtcS9NuWNuGjXhUVTE9OqScz2iVQhkV7TH9Z43BtA9Y+W9Slztv8d4xmU44PDqkbb7g8nqF8h43nyHeAi32aM7Dh8ccHQUV/mw2RYDV8prNZhMnUUtRh1B6Yu6/sqooTVg6VBGorvjwF5QQLVqHoV+YoEq3PtjlVpsNHoUpK84vzjg9fX1bL3flXoEp91ULEoeiqqqO/53Nwkx6fHzCarViuVwzP7QURUnTrGmt7ewH3VNHVCH57PVJW/rzaWkTgPfff5+Dg4OYBai9E7Xbws4NttqOPqmeSo7z+u0uCnTBZHJAWcxp1jCdVZhqwhrPZFax0sRAN5gUE+qqDmEFzlLWFbODOXWMTwJw0rKyDY23SKlZHBwHT/HDA5yGojC4pl8gO8igamhqiB/WraYe9/PUXp3tSXlMYTg4WPDRRx9w8eaUP37+R16/es76ekJhYDYtsDOFtRPauMqfcxXagJfg7/fo0SOapuX4+KTzFC+KoMUz2nSpDBQhJkpEAuUWQWKaM60VWvU5F23rOIsRA/WkYj5/C9dnCqXPXAohTDmpN70XJpPwYR9//DFN0zCbzTpwICmuKRjofLRtaK2iUiJzJiUHVW+0vb6+xlrLmzdv+PDDD5lMJlRV1Ssgbqv9Dc4OybLeK5H734Ejrwghp0iv7s/lvDSAjVZopTpKnFYuv16vWeqW5WpFcWmYyYSPDh6xqde0rgUDlXja1YrGRHmxUMEjXCuKouLg5ITDBw/RdU3jJSzFmUAi28qcfKJpXa/QUUBrfa+OV8GmKIWjtQ2C4uT4mE8++Ruuzs754o9/ZLlsePzgiAcnB5SFYrW84PWr56GPjGKxOKEwhtl0wpP336P1PixwHVn8sihDgF8R/kLAZ4F3Hmdd8JAQCY6xBPY5yFBViK720DYtRWm4urri6vr81n5P5V6AaThQe0fJ4CRp8F76QD/g6OgYrcMsZ32fiFLFGSYgRpHLRaHT++w4SoUVtXVk9cqyDDaOzYZXr17x6aef0rYtVVVlGsBbADVCU76XlCUykJhU51KUPN23ss523u+5bSeysyJAihcKy+lcXV3TOsd6tWHeTPGujYMHShOch9k0XJ+fdwCp5zOK+UPOLs4QX/LBySPMZEbrLFY0NA3aq65PUnuMF3mG3s6U6txai+vi1AQrHnEN18srsFDqiocPjvnRxx+wujjj+uqc+axiNitBeUQc61VQQCyvrymrGUZrirKgrArWyyWb6wZi3sGgvKqoqiD/FK6ME63rdKxaaSJRRmHCCu7GIEaH5Xemk+gbaphM37oklEMhXGLD6JTLYTQ7Jo1NYiGMMai6DoAgqEcRupj/yOVn7NWQDYEwCMoyzGR///d/T1mWmahGp08AACAASURBVPcDnWPq7d+xe0+pTPCNxt1uM8PV+G4JAsrAU0KJRytwNgS5WS3YQkdzgmPdNDSbDa+efh+AZBSl0VG2Uljf4jVIHFG+KCiLgtYK5aRmujgKKzNaS1GWCI7WBWfSxKql/kqTVKrbGEzJqbUDIhZloCiLqNRxTKcTPnjyHquLU7748wXPnz9Fqw0nD48wRXCEBVgurzBFjfWG9WbJ06ffsG7bIM/53r9Oq+DMqpSJ7mGmExW0DixgUugk7V5VVdRVRS1mELXg3FvmTpRTpt6FP4EAiOrvRJlCJ/nOdlGWJVqXTKqChjRbSucdEXe7ZyUmKkv41YUyHx8fR2WHZ7PZDAZHAkB3DN0tZpgnGEllP9cnXd2H14XjaYlRGejZe7lERMB7XBss+pvoghM8soNNZ1LPOPmbAzbrNU2zxjsH4rHeslyvOL++YtEETdUHkyluvUFQzGYLdFFgyhKtChrrQztiaDZN5BSC/S6YE4SmcR3FTwM/Ua2QRSnKXc7haFEq5L2jFHzjqEzBg5Mjlk/e4/LsJd9/f8X5+Rtm85LJbEJrQztcXl2CLhFVsVouefXyJa11eCQsYh0DIIP3StGx/Cpq+ETo3MpS25dlSV1NqKqaSVUz0yVFafDOIkp6z5M7lHsBpmGJw1yl1FDRCIcegCmxfWVpEKWo64JJVRCSKzLw3ZEOTfFgB9j0vnBNWZZMJpMQfBZzBYiXTqmxDYw+waV0sT2Z2pUetrvKGHwDOSq/S6RXpIuPq0QISgWqsF43SEzs6AW0KZjUEwyCNELjWsRaJpMJB4tDprMppixoXL+mbLPeoIqaxeEx3oPRBRaNtS1lWUWgeIpC45xntVrHSQfevHnDZ7/9nNVqFaOSoSzrGGQ3YzabUpU1RWHQpcP4kCzTaE1rNyybDcZ7DhYzfvTxh9jmkvPzN5yfn2HtjOnBEQBOVmizRJuQE88YxfVVH1Yu3ocVQLSJ8mRgp7212AQ26zoHWIB207DW6zCWlMEAzlk2mxVlWWCKtywEIx9qCUQQJmORmGFIwsoG4aI+jbLWBq+IWWVKbLR2J7YpsIF5UJtEBUSiNL3RNryzV+1aa3HeYYphmMaYkt6snMhkHQIVtW2f8iqs3pFfrVINd7aQ8oKIC6taaI1Rmk0T2D1TlKA09TTEeAkeUdA6y+r6iuX6GieWsiqpZjOUBCqybizOwWJ2xHx+iIimsR4nIKIRR4hcjQgP/RImtJcvX/PP//zP/Pa3v2V5vey8VVL8Ul1PODwMPn6HR0d88MGMJ08ecXIcvFkcmnJWs7q4xJc1Bw8f8+j6CutcCPr0nunBIRCULBcXF0ynQRyYTac0q03HtXihy8FXFAVFUQbjt08ypeucnrv2jeNARCjQGAVKFcxmIYTebaU+2F/uBZh8Wqg4fqATHyqmQo4AHcPRbRcVq7pwC+8dPnN07VdqT1oujzHJSJIb64ZgSlQv5SlIa60mCrjLQ3ow4O+g7ROREBKu+1CRzqgY952zAUzGoMIqq+EtuQLCObyzMUtpwdI6rHi0FURptKnxlaBLw/xgQVkVbA4WeO+oqwqPsNpsuFoFlXMrCimEh5Mp88UhZT2n2XiseIqqpt3EQegd6xjfUyjD2ek5v/5/2zuzJ0mS47z/IiKPquqqPufa2cFiASwJ7BJ8lcwo8EUm/dGiDqNMFM1E44MAkDiJBbC7MztHX3Vl5RGHHjwiM6tnAczSxkyzUjlsMNs91dlZWeER7p9//vn//jH//W//lqZpYlit995X27ZstxtevnyBMZrf/Ivnk08+4cMPP+TRw0eU5YRdtaPpOpQ2FPNTHjz5Ho6SLz/7PU1VsX4pKkrzozlZ5tBqy5FRHBUFYb4giwRca0WFNoTQs2jyokAh/VS2S0wM+hpjyqdUHPxgtPR0EbyM9hxNLfxT9m4400hGNy15T4hDugydiwyJxK1SCh8iy9dZgk7t7UOPkI7zaf9YzCtR3hDmpQJtSlb7ImM/KnLIm+RE+jrvcgj3BvQxoWO+39HzosCFgAsjwCOAiTB2rkTIv2taijyjLAqMNrSt5Aoe4fb5ENB5zqwsKYpzSah9oGpqNpsN22rHciMDv4IuIPeCjOqMLJ9QNx02BLRXtG0XNTFavPcURc5yueTnP/85P/nxj3nx5XOKIo/PTjYzk2WCOI5yYO87Xnx5ie0cT794xl/91V9x7949+dy0wSiY5jl0jvcfv0cWPNevXhISP66zTIwm2204D44XAapdhXET8rIkU5rMFHROIHAVAkWExqUnTRoUBd2MzmQGZ0LHyMF2+CDSAfpreMg74Uxj2PuuhfBVw6KFdyWdkzkmUzGeL/C2I89Lum5HbrIYkuxX6cctG+Pfk/4t0ZeSg3810VXu403qT/2rVTxR2Xfg5Eha68hjk4EvSstQMqNVr9/W1ltCvaGuKzJjWMzneNvx8uUl212NySfouqPKdmicbEhxMmDaPJwPbOsd20ryDZUH6m1N8cVzHjy54X45x3nwQdF2Nh6Kqkc767rmV7/6Nf/wD/+Lzz//jLLM+w3M9E13snl47/qQz9qarrN8+ulvefr0GR9++F1OT88o84JgPV3bUNW1tM4fHXHv/n1817LbCI1MRTROaUW985ycnJHnguJK/5eP0gc11a7hfHJCUU5lmFoRZb2sxbpOTh8QNkZEJIuskLxOE6fKC/z+pvbOOFPqX0q2F1L9ifXaFw+RBVsUJXW9QalC5INR9AL7/ZSF/Yt+lcD8ePxJr9nwFTlSKlC+qY2VeuR3arJMhg+0rcN6D8qgEL2E9XZLFecJud0GbSv8bk3wnulkips71qsdm90tTVVT7Tpcl7PdCGv6aDbhaDaTlvGuY7XesN3VuDTszDleXi1Zd/Dko4+ZndxHF1MCiq51KOdpfMN2u40h20v+8R//kX/+p3+SnNIY2raRXFYnoERijPQcnfcoAgaDbTqm5ZTlzTKGvJosz6m2FV3XYAATHHk54fzePVYRBAi2Q2fSUYwpOQ4dqMBmvQVtKMsps9kRk8mMrKjwBDKTo00OQZOpwCRXmEzgcBB0z7oompNBwOKwKD1jMi0ov2l1JmvjRIKwH6q96a4/prMoJY1tY8mv0StjGDkosSZs24WcSVFQNxVGZ4CiaRqEKyiF23Gn6eth3oDE7dudwrEPfUgK9LWP/n0o4mAykcdaLpdcvXpFFwuXUxMwXU23WWNbGYl5NJ9zfNqybSy364rNdofttuSZwigRIikLWUBaa9mdg6KcSQfrpm64Xa5YXDzCZAWd82RBWklCZ/Fdy3J5y6e//S2/+uUvefrsGc++fMpmvSLLc3xuYivDUHPyvutBnJR7llo0GhSKPMvZVTs2m4rpdCJqSdOpnM5tI1oUec7RySnEEHd1dUXjLVrlFJMp55l8xjc3t6xWN7ggwv4ozbbakRUlTWcpyylaGwFVrGxUWdy88ywXGB1DQY6zHkUG2uNC4OT87I3WILwjztQPvfpX2jiMU0oLwVGlHqcwWvgqwsz7+RKIQ/s8x3vR7w5BsVqu+NaTgA7CLkCJnsBd9HH01R++x9FPpZMu6U2IulJH0zRsmlrEIlvHei2nwHq5ZDGRD//40T0KnVMvLcFLX06uc6bTGSen53hd4tWG7eoFWoHR8gy0ChQR+i/KgunRgqOFONNt1ZDlBadn59x78JCinNJ0jta2tNWG3/zql/zkJz/hs88+4/LyktV6iTGao/mcuq7w1kv5IOihVOATdy+Os0C6Wm3boY2hrhs224r1eiOds0GBNmAMpijwnYAxSmt0PEVUUbDbdti6Jnee0+NTzs/PyLKMbVWz2lasNxuyrMAFj0fRtE0vzLmtarrWYrJShhQgzqSVzPnVCnZ1Resajhczvvu9D3n83vtvvA7fSWf6WifTiD2QRkMKwXFE0Qn7A8/GUd6QPimss31fy9XVJU+fPo2Su56TkxOU9mgCHklWdZ5TNy0EKLJM5tPuV4iG66db1fFEZJDmbZou6l10XK3X1E1DVdWst1s2qzXNbkeIhMvdYkpeGskzgkcFmSaRZQXT2RynCnQ2wfuK3CiM1nhncV2LNppyMmVxfMLZxT2OjmXXVeWMbDrn4sEDprMZyhiW10tul2v+4e//ll/94mc8f/6CereLAiXyPruuxphEdYp0rcjNk7co4ApBnltIvUwhsF6tWS5XdJ3ce9N0kjtqjVEFzjt29Y5g2170ZrJY0IVAvdng25p20mCyjPOLc5RZcn17i3OW+WLBg/ce0naW2WzBbL5Am5y2szR1S2t9P2DAWheHA8hamS2mmFyT54rHTx7z+MnjP70Go71TzjTOk8bh3h+hEsRVGj9MH5PgTOLwlEeNLppKUP2PJkusB61kAdpIitxsNtzc3EilfDJBGQNxzGWiymR5AUrmokL8gBAHLcsCY3QcnCUoUtu1dJ3I7242G6qqlkkWzlF1HTYIjItSMibFWtpaYOwvnz6lO54y1cK1M8oAmjx3FIWicJrcBh4+eowKAR8s3nZ425FlhsV8wcnJKdPFgtML0dQ4f/SY43vXWJXz4vIVjoyf/vwX/PSffsbV55/y4vmX7HY7irJgUhYIYdXhnO1FHnMtcz5TmAcpGpC6WHCCUE7KEus8u6ZhuVxhTCYi/lqTlSW+ldGaNgQh3ypNlkeE0Chya+l2Ld7X3NzeoLQmzwuqqsJ56RqwzlJMJuSTkvOLC87vPcCYnLrtqOuGtu1wblgZZTaRDbEQ9kPb7ei6mslswvToGzYFY3wyfeWJpPa//1WvSTveGHZOzgOjklLYv256gVKx21dJX9WkLHnv8XssFgtCCNze3jJfLDgpCoi5gFGiIGu06RHHLM4CSvcp6JLt+2+KouDy8pqrqyuausbF9vymaXDWosoCZTRd66iaBtd2BHz/jG5v11BvOMo1s8IwLSYy2chk5IUmc4qic5Af4WyHsxpdGLJiElsKZuSTUu415g3zkxMq6/nycskXT59ig+bT3/yGn/7kx6jdDdvtNrKrNQRL23R0Xbv3vI0W1LHfD9VAhrXWoiwok0VOXo1WistXr1iuVtx/+Eg6A4InKI0NAbRhOl/QtTV1LSKUwXvMZIKzluX1Nc456ZQ9OqKYTDg5O+HFy0teXb1CZTnz4xNmbc12J5y+zloaa2ldJ6NIEaZHmWm6tqOtW3Qb2FZrlApUdcVqs37jdXwQVDnYwd6SvRMnU0LzgB6KHo+X2W/mu5v0AwwyuNKGEdG8P1K9jlfqc5oeAkdkmK2zLBYLjmZzjuYzqmrLerXiaL4gj1PIcyVUfes8Up/UsaYiu17TNFRVRVVV7HY7JpMJZVny/PlzNptNPxNKTrCOpm2h6frvubire2fR8WRyLrDebGlDR1MY6nKKMjlOZQSVY0yOyTO6rkUyFY1WiqLImM4mTGdHTKdT8smUNo5VWW7WVLuatmvxJqeYzXn/yWN+cPsRN19+yuVlwFkX80+PivSt9D6N0WRaxyJthMbjNm1jhJAIr85a0bRQmvn8iKurK77/iaFpO5IUm8lzUAFrWxwDM8Z7kV9b3Sy5Xa3QRrOLbfmnZ2fMjhbYYHlx+Yr1dovJck7Ozlkcn1BMp2iTgdK42HkLYFuHDhkmM7RdjY653sX9M56ox7jwDWNAJOfZY1yPOHLp769il4N8mEpLq7dSWaQaadHPC6lNgNf4puOmikRBMVqxqWuZkRq/Pz+ao7ViuVyx2Ww4n0zJs6JXhfXx+sEH1us1m8gs2G6lvbppauq65vj4mNlsxs3NjQwvjm0kqcblRwk6SFiTnN7GZraQzdh0HUWm8buKsunQoaEwGXleCormwZhcolotpKQsN30IGuLGdH1zA4A3a7Z1h/OBMjNMphM++PDbnJ8tuPnyEb/73e949uwZNzc3LG/lvU0mcZ7u6LNIFKz03ExUVTXK4OkQ0EJa1rNCctRXV69EX15JDN4/BwY2TCIvOgX1dsfV5Ss22zUoIdTmRQ7rjNYHimnJ/Yf3qT5v8AqWmxW327VwHmPh3blhYIMOGfPZHELAecd8ccR77z3kvSePef+DJxTTb1inrY5jGbMs23Oeu440Zins51By0vgQ+qRdJJY9mUlvMYyoeK+zuPcGrREoypIiz/o6SVlMgDV13QCKoixoug7XBrKiwHWO7bri8vKSm7hIUwtHWmSz2VEkXAYkwk4OP/QDOb+fP6b3OWbMG2PIJxOWTUtlhTWulANk2l2mFIXWZDqQKyiMxjuFqi3OrdBZhvU3bGppwZgen9B5hcXQWZmuV86OuH/vjPMZXFyc88EHT/j88y/47W9/y3otktVt12B01kP84+ghFby9l/YIraGzLc57slyk3bbVhl21Zbm85fTsfCQLJmKfabu08bNpO8vl8+d8/uwZu6oiyzPm8wVFOaG2kmMWkwmT2YxiNul1IZLWeRz2SZZp8qhOpLzBFIZdVXF8csyf/fmf8fHHP+CDD55wcX7aa5K/ib0TzmSM6Z1pzDyAgbc3ttcAiLib982DfZh3V/QDSKdI/H7iaKV+G++EAjObTSmLnFcvX3J1dc10OkUpHRkVOjYginZ58LBarbl8ecnt7S1VVfW/M72H5FRpRxxYFkOh07nQzxfoWR3hdccvikFCetxSn5J96z1JoSot8DRzalJkFMZCcL0gvq89ShtcZON3toO2Ic8deZHx8NF9njx5zA9/+EOePfuS3//+Mz777HOePXtG29j+9wppNAqVxIKqtLFIGFzXNcVkgjIZ2+0OtJZQeLvh4uJehKvj55gWcZDQEKBrW169eMlyu2G325GZjA7I6pqiKJm2DUU9AaU5OT3BBwVaoSIFSRsp6Drv+w4E18U2Hz3j/oMLPv7kB3zve9+lKHI6774W//KdcKYk1F+WUaB9tAjHXycbo3q9xV0QIstc6/4ASovZe9937Crv0cr3jHVlDCHIglda2A9FlvHee+9F51BMpzOOjhaAYrdrKCcSh683G168eMnly0uapu7vKzHbk+M0TdOHWkkOOt2f7MgelN5zpK9qDS+KgpOTE25ubmJPV75HfbLWxU0j9M83PbemU7iQURRTzExUfzZNQ1HkLBYLbBOgbcjIhB+oDCg59fM848mT97l//z7f+c53+Oyzz/ny2QsuLy+5urqhruv+d+V5gdaR9a8gIA6V5UZoSl3LdD5ntVrSti3e21hMFzkBYYt40etL+aJ1bJYrRLtIxPbDrgEalN5iNsLwyPIck+Vok0mZpJdJVkQGFd6GeE35DLLccHN7zdOnX/DJJx8zmZY0zS7qOb6ZvRPOlGac3g3x/pAzjcMfiOyC8clkJBkedvYhHh/+DEPQ5CJD565ikAKbz+dRIcmgTUZWlITY3Cbxt9BZrq6u2Gw2OOd6aFxr4b+nyXR13fQiLVVV9S336XTJshwXRi3ed5wK6EO+yWTSh1ZlWfazYpVSZFr3wiHjMDGRd5PSU1J7Sj8fH6YwGX2grQNtT+YNaC3UpDyf8+i9BRcX7/G97y55/uI5v/zFL/niiy/Y7Xb9fUrbhkdHYCErMlGPtdJThYKbm2t2TUXXtdJT5oXtHaITOWf7CYcqBJbrNa0PODRoQxdEkTY4j+os3oPONKTWdRRBadLUEa21bLaj/rUQ6V3Hx8esNktul9c8mj4ivB4U/FF7J5xJGSOC6tG+Kld6va50x6HCcJKZOFYkhXbJoe6e2Pu1YIU2SoZ3pToVqelQtA98nE4RgiIzOc4G1ps1N9dLdru6B1HS/SeJsHQKad3ivXSCdp3kSqnxUamo4jpW94kf/hicSY41fi5J7y/PhSyLG07iZKm1JE0KKcuSo6Oj/pqbzaY/2RIwYq3ba0fxHrogw7TF+QvuP/oWH33/h/ybf/vX/M3f/Cd+9rOfAbDdyLDtDAWuJQSFUZqubQnI0IW63rEoCq5evuKj73xXkLMQ6V4hCmo6i4ohX7XZ0DqPReGV3JcXMqOgcEG6j1Fp7GcXHSU+T637Piav0qltpGbmZQO7vr7mxYsXnJ+fx1lS3zANCKsCO2epXEemNdNMlHe8s70+9NjGTpYYEsqMTqZM9NIC6QS6q6X3OtxO1IzIjCaEyJlz8kFkmSgfKS+hBiaQ5QXVpuLq8orlctV/aNIkJ1cURxm044qi6O87z3OqqiKEpncOZQZwZRyajUsFY0dK/52knXsRzdFuO95wxg6aTrR0zfR3+r2yEQxI41iDY+ykXdfJbK3jY77/g4+l2xf4/PPPefr0KZvNJjLDM3CeoFphp/gA3qHVILEmjqz6gntwFtcOBdbtZgsmQ2UZwToJ2UabT/88VLxO0IR++tpQvXdhDPrIP5XTkq5tWS2X1LUwUibl5Cu24D9s74Qz5VlG0widRlhfssM4azEqSBztjBAhAbQRBzG6fxjBS2McEOtMuj+VxmFdT08SFmx/DyFE1InBySQkESUeWUcJDpR2jmpbcXl5TVXV0dE8XTsIWSok2U00pclkSpZlrNfrfkFaa6MgzNACcteh0v2NXz9GNtNp0rexOE/Qr4fK4w0khYXpmnc7ir13MecJeydjula6XqJBATx48IBpHMp2fn6OMYZ/+ZdfYVuhR4Wg0DqPeZSiyDTVquPyy1uuXy558q3HdF2DQoT/ZbyN7QGI5XotCKoPBKViW73vQ3XvHJnJITgSvSyFakqrvo6ICmRJ10NpijzHKENjHev1WjZiL+vv67TWvBPO9PDBQzbbivOzM5SCTInMVEJTQoh9Mj3CY+J/alwQSaci26AB17VkRpMl3YawD0CEICCEjvOJegBC60jbGRRpUniQFpTWmXwISmaqLpcrlssVzon0Vlp0Q85kIsdPplqUxQTvYLVcs9vt+jzJmEwKmt69tujlPobTJWmfhxDihJDhd/Y6FsqNnGK0iURLgE+6z+122/8eGE6e/frXEAImS8CJUor1eo1zjuNj0WsoikLQwzLnV7/4Kct2SddZiqJE5nH50UkYqHc1tnUQNNa1satX0qUuDmleL5dU6xWua/tWfmVkirpC5jVI3hXDek0UVQHi30GBMUOnrfcOlEhLA5RFie2kFBIia/1N7Z1wptPTMx482PLg/kN8cLS7inpX0bVpMUWwIBYuRXCQeLBEeNfLCea6DqMystygo4b0tJgQhLcsPxS8sCNGu1oiuOJdrwZrRqFW6rTV2tDZwO1yxfX1dZy/mkeBR9UvaKCvUQE9L09qaeJoKceRUNBj8gyjB+dPNgjQy72kcHEymfS1rFSjUxHNlGfi++uP/+R5TtM0fZg3PqXkeYdenDN9PXbG9DxEQVfAjOvra7bbbX/Noij46KOP0DjWy0t2u5q6FofL4j0lZ6qqitVqRdM05LnBdl0ftjlnqSPJt91V7Nar6NQZzjrpvFUCQmQql7BMjTYG5/buWRn5/LsRGVm6RAT9NUrhrMV1VhSOvsY6fiecqanr2FIQCNZRbbfsthu6piYUGQR6VgOA0hYdcpR2KJzUSIKD4OIHnPICYTH3UV30QGml6DCAiuozrnWYfIbShqbZCENhTxwknjxBZhxdXV2xXK1jXcqii6zfyZum6V+fcpmkGLvdbvFepsUfHR31WubWWkyhByWkmGslEmxCyYqiYDab9XWjpFHRO5uGxKIaF3rHluc5dV33ctPJGdN1vgpBTNfp87s7QEvTND09CmQjKYqC8/NTnjz5Fi9fXrJeb7CdQ6t00srvub295ebmGmNEWiwhbwkEaaIzoQaqmTQiiuCkhJlanGhUtwMfCbce0r4ZPEH5PsA3JoMQKPKCum6odzUKRVs3TKclaVjEm9g74UxPnz6T4meAtt7x6uVz2rqizA0nx9JPVIRiWNzaoHULOtYQdIbH4pSnbRuyTI0+8LhDZwPcPA59UiqgVZBnD6xXG549e0YgMD2acHFxT07FCIpUu4r1ak1Tt2iT0bYdGk8WFWTHizeNFFVKcXp6yunpae9YKXybzWbM53M611K3df9vRVFwenrKZDLpYexXr15xcXFB27a9M+61wCMARAI9BjloeaMJzVssFv33JpMJVVX1HcXy3Bzejxj4sPfs0td5nnN7e0td13taHkKjajiez3n48BEXFxdcX9/QRJpWOkmd63AuY7vdoLWmaTqyXJoz+3JJ/H15ljOZlFRVHd/LcALnuelzZLEwnFAqEOJwO6F+jTIhLdFGQihT+Nuvn2+aCOVnv/89R/MFm/Waarvhi89/h7ct9y9OOJqVaB1kqkNiB8QHoHSGisBEpqS7uWlqilKGRKcJ6gFZGAPrIC6wOGkPIMs0QSlubqT46L3n5uaGq6tr7t1/IPC9C3Tes6oqLm9XNK00tjnX0gbhBeqY08AQ2nnvqeuap0+fcnt7y/X1dR8OpWHXIQQZbpxnvaBLOp1SE6Hcp8DaaQJfnudCXI2FW9t2sQlvPyxLTpCuN51OXyMYw+ta4SmUuysAmkCHLE4ATCHt2NG0Fij87Oyc999/wosXL7m+ukGpNKt4fwbu9fUV09kkntZtDx4UEViZTIUo3DRClNVa9+CEvMeY6zKeChmBGgAV+iEOfRgdN1tnLXleMDua8uXzp/zlX36C1oGm+4ZNDvzFz37GyekZ5WSCt5YXL17Q1Bs0lovzU+azQsZzJmQmgPUKpY3wr5TCK0OmFG2XY22BUsS4f0wpGu/WkmOpCJ02bUNVN1zfXBMIPHz4kPliwWx2xHZbkZcl2mTUdcN2u4n6dqGfW9RZF9WETL9jOhRtLNju6oZjH2Kh1/WhYHI4ADLJD1NiD/T1o/SapEuxWq24vb2lbdv91+W5TBiP9aExMJEcpG3bXrkW5GTadwQvRNnRiQQD4DA+nfI877Xqxu/LOYci4LuC4+NjHj9+zK9//WuuLq8jAqkAYaB779jtKq6vr3h/+rgXi4wIOUUhzjSLojBy36pXPdpvBE0auiH+f3KoMBTjleqZbiPcMQAADuxJREFUDcJEp+cFpjB8t9tRZAbUN4wBcX19xXqzkdEf3or0opMQIniRws1z0y980UGRfMg7Kw9TaVCyo0mhLUj+oEYsgtFC1VpOJaXSThyZyc4ym82k7tB1MmM1gg/eB6qqYrlcYWP7Qtc1JBpS1+3XdcacPOccOs+lGq8NnfcYLcl2ngi+kf2wD1H7vYWeQIVxjWk8jK1tGto7YX5C+9JCNMYwnU77om1CCNOpZW3Hblf1Refx9b33dBEgSPeTOo7HwIn3Hg10raWqahaLYy4uLvj9734/CiXFYaztuL29oWmkWTJ0aTNRe8yWyWRCEsPUOk2gH6ZzpKMsKOH4DSeTxO/OB1B+D9BBroBWGtt1LJdLdrsdu6piPisZvexP2jvhTLe3t0xnMwIKoxTgMCpQbbeRs1VI7NoX4EAeUBI59BCMsINdh4+Q+l5hl2ExDom0x6tU6A2xv0ZT7bYEJUCCdY7HAYq8YFeL3NV6vY4THEysj4WIGu3XY1L4kxwh5QkD1Ax100YemUE0E9zA0yP9nElvuue/pTAvXatnbBjTF2zHJ4i1dq/AOz796lpaRFIIOZlMWCzmPZs/5T/pVAT2HCvNsLJ2QN5CCGRKkWnRV8iynNOTU6azKd57ttUaH4Vqssxwe3sjbTTe4dNgMh9wvsN7G99DB8r3IJP3ISKDEXEcMc3lPqNTIbVF5wJ5rvtrQ4peYm44Wo8Chlic/4YxILabFaenpzLRwFk26yVFrlmuVjRNy2I2BVyv8JOO8Y/++df86L/+PX/3H3/Eb3/4A7QSsmoiVIrwR+oRMj3EnhaTVgOtJHSOulpzfn5G20mrQFGWbLcVL1++4r7StJ1jvd5S16IgBIkVLToHSu03NToncr1jyk9d1z1rYLxDS/7i+1DUeRuvGfZ20jyPm41RcdPYr0XZzvahTAIP7r4mhX8pb0tiLslhUnhsjIxXKYqCo6MjiqLoHS8Va1erFY8ePerzvuRMs9kMH5nkSdqsnEyYz+dcX1/3Tq2NNHFa17FareQ5YWJhW1gSyYE3mw1t28XryfvURtOl9xyfu2xEI0cavXfYLz30zInoUGVZkmcZWWYIhD6MfhN7J5zJWcv5+Xlfc3j14kt8rqmrShLq9CASJxVQQfGj//L3PHj+ih/957/j07/4PkF5fJBFqHXqKZMuFjmhht85rruAHHKiRBTY7dIPK7bbHQlQ2m62fXEy4GOZysuUbuf6BHhMu0mwcxonulqtaNu2rxWlwqmcInK6jcGDlIuMuYLeD85UltOe9iRhkYv6fgMKNWZJJMQtzewFOUETwCHOJzU67z1VVfWnYUICz85EXiudVAlVlBA4imU6x3wyjfw+S1FknJ+f9wRdcVbdh8iLxYLlcsWuanomh1R5hhP05uaGrpPB0c4JKpdCxVTcHy8SoRKlYQ+ip/haKUGpfhBaURQYY3j08GFUt/K9Qu2b2DvhTKjAbDbl7Oycuq75xc//mbZr5UNuJZzRYah3pFrK//j3/46//m//k7/7D38tDy+k/MRGPt0dhjj7Id/Ywbz31E1FkU8io1lmpiptOJovUBiqqqGqxLkUQjcSMqTFOWntTjkNMNSPYp6SpIVT+JVqMa89jlhA9iN+3ZiVcLeoOzifhCw6rS6GZ5Zel76W8EwWbQoX0ykmjjUwHkKQE7ht25EwvtSlzs7OqOu6r6WdnMj4l/nRXLTNq6ovBVxcnHN0NJNiuu1Qcf6RzCRuWS1XLJdLzs8vMCanaWq8t/17bZo60rvk2eiYN4/zL0ECG7RRXJxfcHV1Ka9Nk1SczK9Kn1Ea2ZqaALXWPHz4QNpO4kb4pvZuONO/0n7z8Uf85i/+XNRE/2/fzMH+v7d3Qp1IEB3RRDiaT8kyg/cyBa7tOunN8QNBVcIrRQg6CgjGP8H1AATs64YnyFn+vtvbNJBGA55hmoPm5OSEIi97FoKcPqGHYrMsjSLR/WkzlgQej6hJbIfxyZVYDul1QyH5dRGZMYM8JdBjkCEhV4H9wioMxelkbVuTZYYsM7HQbeJz86/llulES3lWgpDX6zVN03BzcyMIWKqXJVSx7XDO96danuc8ePCArmv60zC9t91uR1VVtG3XI6shDq5r24a2bfo8U5DUEENd379vIebKJPd79845Pz/DGFHMTQCOipSwFC2kZ5oZYbAsFgtm0ykycMB+89A8pQJVtcX7DqMMs9mU1W0dkaQ2siMGTTaFOJQC0OJI6BTmeUF/lCjcyMKKc5z8uEZyBz+O6I8k/ZD2GR2F5buupa6b0SKOCjdGU5Y5icqWEDygd460cBIgkSDpcbVdFqyAJd4rmQ0Uhk0g2X6Yl4qx8lT6+spXvLVUVojfoSjyvr0eQv98rE2LeegFGztmCvFS/mWM6QGJMdOibVuMEmet6x2TaYmNufHx8TE3NzfxWqZ/f1VVoZSOCCiA62lFkHTZDV1ohR2OijB5OhN8JNvOefLkCV3XUpYTuq6K60zC4D2gKM5PVvHzOjs5kZEyI4rVm9o7cTIFAuvtiqADNljKaYnOFJ3raG1DkpcK6X8BKTbFPyroeDqJIIkfwcopZ/LB7y3OuydTclRJTuWuFJCZjBAUVVWz3Vak5j6to6yV8qD8HuqWTqaU1Dvn2Gw2fPrppz0j4i4tZx91G7XZ37nXtLj61gofRrvzPnAx/Bkzx+VrARBqmqYWSpJ3oORUSzSs15/RAPGnU2i5XPZcv7quWa/XrNdrVss1WZb3Dpr0HWazKWfnZ0NVI24IQr5t43MTZFQrv/dsxqItY1ZHel/OOayznF+c8+1vf8B0OomvHZZ5D1T0j2tIEVK+NDuakedCoG3j3N83sXfCmbTWQsGPi7AoCwKIgH3XYH2HxeJVkD/a4ZXD45HxwAGCJlFaE1qzF+alxdmHMAFey7T2BwJLFV40FUS2q+2T14Ty9G0Id5jdY+fwsfB8eXnZM7ZlJqsZ1WX8nZ8TZnwCAcZw9xiQ8KPN4O7CHztkCnETCdREGWHrLFmu+81gQD7DXsH57umYGBTppE3OlWx6MuVme8POtpAZrHcEFEUZZ9vqgd2dkNosy3n58iVpNKkxvu/fGnq+hsF4Y1QuPef5/IgHD+5LeB6h/P1Y7atnaiklEgXf/uADTk9P+nX0dewdCfOk1kFc34kVDWA7ySm06Xsu4s8Iw0GUb2Qno24pioy2qlGLBZkWqBhS1+3QgZqQvbAX/YweYPxvkxEpRKIEpJVGGR21GiAVWSflEWmeUrpGKmqmsG86ne4RUMfTCeV7KjLek7TXwIBOC+A1Bafgh/cQxoxp9hb30Fkqi9E6y9C3NbRDhBBP9xFBdlznSo427u6tqoEtkWDtEISiZWK9RmSJPSorOD69R1E+p653BDy+c+gY1joXnwuuZ6ykMC6FlgpNml2b1gLytDg9PeH8/EwaC63ti/d3LZV3079kWcaT999ncbyI6GOL1qpvw3kTeyecSbhQkWYfpCOVuDCloOgoJ6rPV8anTiAQvEYpj9Eao41M3UZR5iJ6j1Yor2O7xYhJcccCqVmMWDEXx9hVOzbrjYSVCTKORMskGhL80KSXnCc51m63wzknhemRKlFapC4WgIUpLQ1xEgYNodu4jX182kiheiSPFfY1BhMTYBwiKg12BISMu2WFILoPbIxh89Renr6eTqf9STsuCyRmSAJhUq5XFDnHi+OYxxlxpuAIQROwTGcaazcyTrSBzJTsqiY+T7l+17qY0xic831dTKnAxcU58/m83xTKcoKthlBNyhqvxyTGGO5dXDCdTPBe4HM9eo5vYu+EM+VxCJfWol4zmx2htFS2N9uK3a5mNpv2O1ASS1E6Bt5xWp0P9MXIXpEVJbN3EoKjpRQYI6l+19LKYL1MByeIsHKqSyQkL5DCiyQRnOG9wrqOtuvk1FJqb+GPQ6ZxnWa844qikRk6ZQFG6CPsT4Mf0EIPRsdQLKAjcXTYiUc1NURRyXmLjkOQEwMi1ZWkaLyfLyVLG0CiD2mtKcuS09NTlstl/1rXc/Nc3HRM3AR9T7A9PT3tQ7AQhjGrSQoNFXCtxXuN1gXrtTjDclnjQ44yBqVjT5JSsb6kyYuSi4sLptNE4g3QgzujulsqcLPvVJnR5EVGonKlgu6b2jvhTHH1UNc7tDZMyiQ/3GGtaHAL9WR4fQhB9CJiBVcS1nGiqjE6Rys5PeTI12iVkRlPcAGPw4YEEgR0rlFZhnUe6wJmMqHpLMvdltrayKROx77CmAyTScuzbeU4uduuAOzRd0RWOO9ZEPK7h/b88UMZis2vT1dMrGrCWF/QxkWz3ykrruSpbRuLssJJK+K81rath07b0ekbRqfcUGIIfeiaNoTb29u+HaRXQFXSEVzkRc8m6LpOBlobw8nJCev1GgJ7z0zHcZpVfSNikc6x2Q5y0z7NMo5ikiid+ODcOzvjwcUZ0zynaRtO5kdxEmAydefvwfJM9NnT+kldvwlEehN7J5xJK4V3nvVqLfNGixwb9dJ2u4bNZsvueE4e5/Rok7QIFCoTmgiICIa1jq5LI+xBYSBInB08PSnVO4fRA5ajtQw17loRvDfZ0IWKQsa8WIc2BoOGYGLbtKY0M9ANPoTXnGlco0ls7RQapR1ea83JyTFaS5W/bRs620SI1u0BDmVZ9l9b69CKGEZl4LsRgJGmwydKjRpV88URU2gnikoqnsp+bxe/m2/ked6DDU3TcHt7y7Nnz0h6gWk/KIo80nN0n6daa+najizLmc/ngiJa24M5s9lMwliXWOjyOa7jWJftdtsr3iqlCN6T5xlFJqyHoigk6gieXCnOj0+5f3bKZ58/lZ/RorVnTD46cSIq6B3lpCQBD85ZuhHN6k1MfZ2Y8GAHO9gftncCGj/Ywf5fsIMzHexgb8kOznSwg70lOzjTwQ72luzgTAc72FuygzMd7GBvyQ7OdLCDvSU7ONPBDvaW7OBMBzvYW7KDMx3sYG/JDs50sIO9JTs408EO9pbs4EwHO9hbsoMzHexgb8kOznSwg70lOzjTwQ72luzgTAc72FuygzMd7GBvyQ7OdLCDvSU7ONPBDvaW7OBMBzvYW7KDMx3sYG/JDs50sIO9Jfs/wet+2w+N6Y0AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 216x216 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "img.show(y=learn.predict(img)[0])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### A segmentation example"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now we are going to look at the [camvid dataset](http://mi.eng.cam.ac.uk/research/projects/VideoRec/CamVid/) (at least a small sample of it), where we have to predict the class of each pixel in an image. Each image in the 'images' subfolder as an equivalent in 'labels' that is its segmentations mask."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "camvid = untar_data(URLs.CAMVID_TINY)\n",
    "path_lbl = camvid/'labels'\n",
    "path_img = camvid/'images'"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We read the classes in 'codes.txt' and the function maps each image filename with its corresponding mask filename."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "codes = np.loadtxt(camvid/'codes.txt', dtype=str)\n",
    "get_y_fn = lambda x: path_lbl/f'{x.stem}_P{x.suffix}'"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The data block API allows us to uickly get everything in a [`DataBunch`](/basic_data.html#DataBunch) and then we can have a look with `show_batch`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = (SegmentationItemList.from_folder(path_img)\n",
    "        .random_split_by_pct()\n",
    "        .label_from_func(get_y_fn, classes=codes)\n",
    "        .transform(get_transforms(), tfm_y=True, size=128)\n",
    "        .databunch(bs=16, path=camvid)\n",
    "        .normalize(imagenet_stats))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As before, we export the data then create an empty [`DataBunch`](/basic_data.html#DataBunch) that we pass to a [`Learner`](/basic_train.html#Learner)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "data.export()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "empty_data = ImageDataBunch.load_empty(camvid, tfms=get_transforms()[1], tfm_y=True, size=128).normalize(imagenet_stats)\n",
    "learn = Learner.create_unet(empty_data, models.resnet18)\n",
    "learn.load('mini_train');"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "And now we can a prediction on an image."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(ImageSegment (1, 128, 128), tensor([[[21, 21, 21,  ...,  4,  4,  4],\n",
       "          [21, 21, 21,  ...,  4,  4,  4],\n",
       "          [21, 21, 21,  ...,  4,  4,  4],\n",
       "          ...,\n",
       "          [17, 17, 17,  ..., 17, 17, 17],\n",
       "          [17, 17, 17,  ..., 17, 17, 17],\n",
       "          [17, 17, 17,  ..., 17, 17, 17]]]), tensor([[[2.3323e-02, 2.2946e-02, 4.6494e-03,  ..., 3.7161e-04,\n",
       "           5.0104e-03, 6.4412e-03],\n",
       "          [2.2634e-02, 2.0179e-02, 5.6307e-03,  ..., 2.9151e-04,\n",
       "           5.1298e-03, 5.2150e-03],\n",
       "          [6.2241e-03, 7.0151e-03, 3.7897e-04,  ..., 4.0918e-06,\n",
       "           3.2178e-04, 4.4637e-04],\n",
       "          ...,\n",
       "          [9.0529e-04, 7.9842e-04, 7.6028e-05,  ..., 8.6253e-04,\n",
       "           3.6710e-03, 3.6289e-03],\n",
       "          [3.6277e-03, 4.0498e-03, 5.4831e-04,  ..., 3.7190e-03,\n",
       "           8.8866e-03, 9.6405e-03],\n",
       "          [3.8535e-03, 3.4831e-03, 6.7555e-04,  ..., 2.9442e-03,\n",
       "           8.2562e-03, 7.8639e-03]],\n",
       " \n",
       "         [[2.7580e-02, 2.4669e-02, 7.3161e-03,  ..., 1.1249e-03,\n",
       "           9.4848e-03, 1.1947e-02],\n",
       "          [2.5641e-02, 2.3801e-02, 8.7720e-03,  ..., 7.5269e-04,\n",
       "           9.4482e-03, 8.9084e-03],\n",
       "          [9.8628e-03, 9.7588e-03, 9.2763e-04,  ..., 2.3099e-05,\n",
       "           8.9190e-04, 1.2924e-03],\n",
       "          ...,\n",
       "          [3.0092e-04, 3.2446e-04, 1.6413e-05,  ..., 5.1571e-04,\n",
       "           1.9530e-03, 2.2843e-03],\n",
       "          [1.5598e-03, 1.6954e-03, 1.7152e-04,  ..., 2.1259e-03,\n",
       "           5.2503e-03, 5.7921e-03],\n",
       "          [1.5411e-03, 1.6587e-03, 1.9799e-04,  ..., 1.7436e-03,\n",
       "           4.3894e-03, 4.9487e-03]],\n",
       " \n",
       "         [[3.9882e-03, 3.1505e-03, 5.0457e-04,  ..., 1.4255e-03,\n",
       "           1.0569e-02, 1.2237e-02],\n",
       "          [4.3692e-03, 3.8382e-03, 7.0057e-04,  ..., 1.4145e-03,\n",
       "           1.2426e-02, 1.2313e-02],\n",
       "          [8.4888e-04, 7.3702e-04, 2.9357e-05,  ..., 8.8902e-05,\n",
       "           2.1612e-03, 2.8400e-03],\n",
       "          ...,\n",
       "          [4.2905e-04, 4.1327e-04, 1.9656e-05,  ..., 3.4418e-03,\n",
       "           6.5353e-03, 6.9649e-03],\n",
       "          [1.5882e-03, 1.9241e-03, 1.7470e-04,  ..., 6.1704e-03,\n",
       "           8.6600e-03, 9.6681e-03],\n",
       "          [1.6485e-03, 1.6144e-03, 1.9710e-04,  ..., 4.7098e-03,\n",
       "           7.9967e-03, 8.2181e-03]],\n",
       " \n",
       "         ...,\n",
       " \n",
       "         [[9.3113e-03, 6.5334e-03, 1.8013e-03,  ..., 3.2796e-03,\n",
       "           2.0767e-02, 2.0632e-02],\n",
       "          [8.6677e-03, 8.9047e-03, 2.1877e-03,  ..., 3.3119e-03,\n",
       "           2.2205e-02, 2.2050e-02],\n",
       "          [2.8595e-03, 2.1089e-03, 1.8231e-04,  ..., 2.3835e-04,\n",
       "           4.7750e-03, 5.3993e-03],\n",
       "          ...,\n",
       "          [1.3388e-04, 1.5440e-04, 3.7817e-06,  ..., 1.1344e-03,\n",
       "           2.8678e-03, 3.3601e-03],\n",
       "          [7.2934e-04, 9.0662e-04, 5.3639e-05,  ..., 2.7882e-03,\n",
       "           4.7959e-03, 5.4441e-03],\n",
       "          [6.9227e-04, 8.4727e-04, 5.7668e-05,  ..., 2.1941e-03,\n",
       "           4.1560e-03, 4.9784e-03]],\n",
       " \n",
       "         [[1.2492e-02, 1.1063e-02, 4.9451e-03,  ..., 3.0513e-02,\n",
       "           6.4178e-02, 7.0040e-02],\n",
       "          [1.2166e-02, 1.2005e-02, 6.0362e-03,  ..., 3.1947e-02,\n",
       "           6.7927e-02, 6.9785e-02],\n",
       "          [7.7364e-03, 7.5642e-03, 1.4550e-03,  ..., 6.8208e-03,\n",
       "           3.3262e-02, 4.0198e-02],\n",
       "          ...,\n",
       "          [1.5871e-05, 1.4742e-05, 1.9349e-07,  ..., 3.2095e-04,\n",
       "           8.5799e-04, 9.8868e-04],\n",
       "          [1.1769e-04, 1.4500e-04, 4.5852e-06,  ..., 7.2754e-04,\n",
       "           1.4678e-03, 1.6742e-03],\n",
       "          [1.1819e-04, 1.1339e-04, 5.3238e-06,  ..., 5.4077e-04,\n",
       "           1.2279e-03, 1.3526e-03]],\n",
       " \n",
       "         [[2.5760e-03, 2.1818e-03, 1.8901e-04,  ..., 1.0640e-03,\n",
       "           1.0117e-02, 1.1868e-02],\n",
       "          [2.7598e-03, 2.6283e-03, 2.9429e-04,  ..., 9.7199e-04,\n",
       "           1.1117e-02, 1.1070e-02],\n",
       "          [3.4934e-04, 3.4733e-04, 5.4813e-06,  ..., 6.2566e-05,\n",
       "           1.8896e-03, 2.6472e-03],\n",
       "          ...,\n",
       "          [6.8025e-04, 5.9770e-04, 3.0153e-05,  ..., 6.9031e-03,\n",
       "           1.1587e-02, 1.2512e-02],\n",
       "          [2.8487e-03, 2.9917e-03, 3.1098e-04,  ..., 1.0829e-02,\n",
       "           1.6660e-02, 1.6246e-02],\n",
       "          [2.4998e-03, 2.2843e-03, 2.9663e-04,  ..., 8.1416e-03,\n",
       "           1.2735e-02, 1.3245e-02]]]))"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "img = data.train_ds[0][0]\n",
    "learn.predict(img)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "To visualize the predictions, we can use the [`Image.show`](/vision.image.html#Image.show) method."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x216 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "img.show(y=learn.predict(img)[0])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Text"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Next application is text, so let's start by importing everything we'll need."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from fastai import *\n",
    "from fastai.text import *"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Language modelling"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "First let's look a how to get a language model ready for inference. Since we'll load the model trained in the [visualize data tutorial](/tutorial.data.html), we load the vocabulary used there."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "imdb = untar_data(URLs.IMDB_SAMPLE)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "vocab = Vocab(pickle.load(open(imdb/'tmp'/'itos.pkl', 'rb')))\n",
    "data_lm = (TextList.from_csv(imdb, 'texts.csv', cols='text', vocab=vocab)\n",
    "                   .random_split_by_pct()\n",
    "                   .label_for_lm()\n",
    "                   .databunch())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Like in vision, we just have to type `data_lm.export()` to save all the information inside the [`DataBunch`](/basic_data.html#DataBunch) we'll need. In this case, this includes all the vocabulary we created."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "data_lm.export()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now let's define a language model learner from an empty data object."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "empty_data = TextLMDataBunch.load_empty(imdb)\n",
    "learn = language_model_learner(empty_data)\n",
    "learn.load('mini_train_lm');"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Then we can predict with the usual method, here we can specify how many words we want the model to predict."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Total time: 00:01\n",
      "\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "'This is a simple test of these men from the \" popularity of some scenes of a xxmaj gordon - xxmaj hudson and thick parts ,'"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "learn.predict('This is a simple test of', n_words=20)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Classification"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now let's see a classification example. We have to use the same vocabulary as for the language model if we want to be able to use the encoder we saved."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "data_clas = (TextList.from_csv(imdb, 'texts.csv', cols='text', vocab=vocab)\n",
    "                   .split_from_df(col='is_valid')\n",
    "                   .label_from_df(cols='label')\n",
    "                   .databunch(bs=42))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Again we export the data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "data_clas.export()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now let's define a text classifier from an empty data object."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "empty_data = TextClasDataBunch.load_empty(imdb)\n",
    "learn = text_classifier_learner(empty_data)\n",
    "learn.load('mini_train_clas');"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Then we can predict with the usual method."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(Category negative, tensor(0), tensor([0.8194, 0.1806]))"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "learn.predict('I really loved that movie!')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Tabular"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Last application brings us to tabular data. First let's import everything we'll need."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from fastai import *\n",
    "from fastai.tabular import *"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We'll use a sample of the [adult dataset](https://archive.ics.uci.edu/ml/datasets/adult) here. Once we read the csv file, we'll need to specify the dependant variable, the categorical variables, the continuous variables and the processors we want to use."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "adult = untar_data(URLs.ADULT_SAMPLE)\n",
    "df = pd.read_csv(adult/'adult.csv')\n",
    "dep_var = '>=50k'\n",
    "cat_names = ['workclass', 'education', 'marital-status', 'occupation', 'relationship', 'race', 'sex', 'native-country']\n",
    "cont_names = ['education-num', 'hours-per-week', 'age', 'capital-loss', 'fnlwgt', 'capital-gain']\n",
    "procs = [FillMissing, Categorify, Normalize]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Then we can use the data block API to grab everything together before using `data.show_batch()`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = (TabularList.from_df(df, path=adult, cat_names=cat_names, cont_names=cont_names, procs=procs)\n",
    "                           .split_by_idx(valid_idx=range(800,1000))\n",
    "                           .label_from_df(cols=dep_var)\n",
    "                           .databunch())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We define a [`Learner`](/basic_train.html#Learner) object that we fit and then save the model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Total time: 00:04\n",
      "epoch  train_loss  valid_loss  accuracy\n",
      "1      0.328005    0.354749    0.820000  (00:04)\n",
      "\n"
     ]
    }
   ],
   "source": [
    "learn = tabular_learner(data, layers=[200,100], metrics=accuracy)\n",
    "learn.fit(1, 1e-2)\n",
    "learn.save('mini_train')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As in the other applications, we just have to type `data.export()` to save everything we'll need for inference (here the inner state of each processor)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "data.export()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Then we create an empty data object and a learner from it like before."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = TabularDataBunch.load_empty(adult)\n",
    "learn = tabular_learner(data, layers=[200,100])\n",
    "learn.load('mini_train');"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "And we can predict on a row of dataframe that has the right `cat_names` and `cont_names`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(Category 1, tensor(0), tensor([0.8100, 0.1900]))"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "learn.predict(df.iloc[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "jekyll": {
   "keywords": "fastai",
   "summary": "Intermediate tutorial, explains how to create a Learner for inference",
   "title": "tutorial.inference"
  },
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}