generate-story-classification-training-data.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loading from cache\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "(38109500, 14)"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import os\n",
    "import polars as pl\n",
    "\n",
    "if os.path.isfile(\"data/hn.parquet\"):\n",
    "    print(\"Loading from cache\")\n",
    "    hn = pl.read_parquet(\"data/hn.parquet\")\n",
    "else:\n",
    "    from datasets import load_dataset\n",
    "\n",
    "    hn: pl.DataFrame = pl.from_pandas(\n",
    "        load_dataset(\"OpenPipe/hacker-news\")[\"train\"].to_pandas()\n",
    "    )\n",
    "    hn.write_parquet(\"data/hn.parquet\")\n",
    "\n",
    "hn.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of stories: 285603\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div><style>\n",
       ".dataframe > thead > tr > th,\n",
       ".dataframe > tbody > tr > td {\n",
       "  text-align: right;\n",
       "  white-space: pre-wrap;\n",
       "}\n",
       "</style>\n",
       "<small>shape: (5, 14)</small><table border=\"1\" class=\"dataframe\"><thead><tr><th>id</th><th>type</th><th>by</th><th>time</th><th>title</th><th>text</th><th>url</th><th>score</th><th>parent</th><th>top_level_parent</th><th>descendants</th><th>kids</th><th>deleted</th><th>dead</th></tr><tr><td>i64</td><td>str</td><td>str</td><td>datetime[μs]</td><td>str</td><td>str</td><td>str</td><td>f64</td><td>f64</td><td>i64</td><td>f64</td><td>list[i64]</td><td>bool</td><td>bool</td></tr></thead><tbody><tr><td>38107297</td><td>&quot;story&quot;</td><td>&quot;kristianpaul&quot;</td><td>2023-11-02 00:26:02</td><td>&quot;NASA+: Our New…</td><td>null</td><td>&quot;https://www.na…</td><td>53.0</td><td>null</td><td>38107297</td><td>11.0</td><td>[38109407, 38109358, … 38108262]</td><td>null</td><td>null</td></tr><tr><td>38107413</td><td>&quot;story&quot;</td><td>&quot;bexsella&quot;</td><td>2023-11-02 00:39:42</td><td>&quot;High-level emu…</td><td>null</td><td>&quot;https://github…</td><td>66.0</td><td>null</td><td>38107413</td><td>28.0</td><td>[38107957, 38108038, … 38108658]</td><td>null</td><td>null</td></tr><tr><td>38107537</td><td>&quot;story&quot;</td><td>&quot;lxm&quot;</td><td>2023-11-02 00:55:32</td><td>&quot;As child care …</td><td>null</td><td>&quot;https://www.cb…</td><td>80.0</td><td>null</td><td>38107537</td><td>145.0</td><td>[38108043, 38108132, … 38108475]</td><td>null</td><td>null</td></tr><tr><td>38107711</td><td>&quot;story&quot;</td><td>&quot;lagniappe&quot;</td><td>2023-11-02 01:16:22</td><td>&quot;Window-swap – …</td><td>null</td><td>&quot;https://www.wi…</td><td>74.0</td><td>null</td><td>38107711</td><td>22.0</td><td>[38108389, 38108527, … 38108699]</td><td>null</td><td>null</td></tr><tr><td>38108873</td><td>&quot;story&quot;</td><td>&quot;g42gregory&quot;</td><td>2023-11-02 04:05:21</td><td>&quot;Yann LeCun: AI…</td><td>null</td><td>&quot;https://www.bu…</td><td>59.0</td><td>null</td><td>38108873</td><td>28.0</td><td>[38109384, 38109151, … 38109307]</td><td>null</td><td>null</td></tr></tbody></table></div>"
      ],
      "text/plain": [
       "shape: (5, 14)\n",
       "┌──────────┬───────┬──────────────┬──────────────┬───┬─────────────┬──────────────┬─────────┬──────┐\n",
       "│ id       ┆ type  ┆ by           ┆ time         ┆ … ┆ descendants ┆ kids         ┆ deleted ┆ dead │\n",
       "│ ---      ┆ ---   ┆ ---          ┆ ---          ┆   ┆ ---         ┆ ---          ┆ ---     ┆ ---  │\n",
       "│ i64      ┆ str   ┆ str          ┆ datetime[μs] ┆   ┆ f64         ┆ list[i64]    ┆ bool    ┆ bool │\n",
       "╞══════════╪═══════╪══════════════╪══════════════╪═══╪═════════════╪══════════════╪═════════╪══════╡\n",
       "│ 38107297 ┆ story ┆ kristianpaul ┆ 2023-11-02   ┆ … ┆ 11.0        ┆ [38109407,   ┆ null    ┆ null │\n",
       "│          ┆       ┆              ┆ 00:26:02     ┆   ┆             ┆ 38109358, …  ┆         ┆      │\n",
       "│          ┆       ┆              ┆              ┆   ┆             ┆ 38108262]    ┆         ┆      │\n",
       "│ 38107413 ┆ story ┆ bexsella     ┆ 2023-11-02   ┆ … ┆ 28.0        ┆ [38107957,   ┆ null    ┆ null │\n",
       "│          ┆       ┆              ┆ 00:39:42     ┆   ┆             ┆ 38108038, …  ┆         ┆      │\n",
       "│          ┆       ┆              ┆              ┆   ┆             ┆ 38108658]    ┆         ┆      │\n",
       "│ 38107537 ┆ story ┆ lxm          ┆ 2023-11-02   ┆ … ┆ 145.0       ┆ [38108043,   ┆ null    ┆ null │\n",
       "│          ┆       ┆              ┆ 00:55:32     ┆   ┆             ┆ 38108132, …  ┆         ┆      │\n",
       "│          ┆       ┆              ┆              ┆   ┆             ┆ 38108475]    ┆         ┆      │\n",
       "│ 38107711 ┆ story ┆ lagniappe    ┆ 2023-11-02   ┆ … ┆ 22.0        ┆ [38108389,   ┆ null    ┆ null │\n",
       "│          ┆       ┆              ┆ 01:16:22     ┆   ┆             ┆ 38108527, …  ┆         ┆      │\n",
       "│          ┆       ┆              ┆              ┆   ┆             ┆ 38108699]    ┆         ┆      │\n",
       "│ 38108873 ┆ story ┆ g42gregory   ┆ 2023-11-02   ┆ … ┆ 28.0        ┆ [38109384,   ┆ null    ┆ null │\n",
       "│          ┆       ┆              ┆ 04:05:21     ┆   ┆             ┆ 38109151, …  ┆         ┆      │\n",
       "│          ┆       ┆              ┆              ┆   ┆             ┆ 38109307]    ┆         ┆      │\n",
       "└──────────┴───────┴──────────────┴──────────────┴───┴─────────────┴──────────────┴─────────┴──────┘"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from datetime import datetime\n",
    "\n",
    "stories = hn.filter(\n",
    "    (pl.col(\"type\") == \"story\")\n",
    "    & (pl.col(\"text\").is_null())\n",
    "    & (pl.col(\"url\").is_not_null())\n",
    "    & (pl.col(\"score\") >= 30)\n",
    "    & (pl.col(\"descendants\") >= 5)\n",
    "    & (pl.col(\"time\") >= datetime(2010, 1, 1))\n",
    ")\n",
    "\n",
    "print(f\"Number of stories: {stories.height}\")\n",
    "display(stories.sort(\"time\").tail(5))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAApYAAAGwCAYAAAAT7yW8AAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjguMCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy81sbWrAAAACXBIWXMAAA9hAAAPYQGoP6dpAAC6T0lEQVR4nOydd3hUVf7/XzOT3htJKKGGFqpUI8rSJCB2dC0oiIo/WdAF9quuuy6uui67roquYlkb2LvugkgRKSrV0AmETgKkh/Seub8/ztw7MyEJSUgycfy8nmeembn33HPPBJ288/6UY9I0TUMQBEEQBEEQLhKzqxcgCIIgCIIguAciLAVBEARBEIRmQYSlIAiCIAiC0CyIsBQEQRAEQRCaBRGWgiAIgiAIQrMgwlIQBEEQBEFoFkRYCoIgCIIgCM2Ch6sX8EugqqqKXbt2ERUVhdksWlwQBEEQfglYrVYyMjK45JJL8PAQydMayE+5AezatYsRI0a4ehmCIAiCIDSB7du3M3z4cFcv41eBCMsGEBUVBaj/MNu3b+/i1QiCIAiC0BDS0tIYMWKE8XtcaHlEWDYAPfzdvn17OnXq5OLVCIIgCILQGCSNrfWQn7QgCIIgCILQLIiwFARBEARBEJoFEZaCIAiCIAhCsyA5ls1IdXU1lZWVrl6GUAuenp5YLBZXL0MQBEEQ3BoRls2Apmmkp6eTl5fn6qUI9RASEkJ0dDQmk8nVSxEEQRAEt0SEZTOgi8rIyEj8/PxEuLQxNE2jpKSEzMxMAGkZJQiCIAgthAjLi6S6utoQleHh4a5ejlAHvr6+AGRmZhIZGSlhcUEQBEFoAaR45yLRcyr9/PxcvBLhQuj/RpIHKwiCIAgtgwjLZkLC320f+TcSBEEQhJZFhKUgCIIgCILQLIiwFARBEARBEJoFlwrLV199lYEDBxIUFERQUBDx8fF8++23xvmysjLmzJlDeHg4AQEBTJ06lYyMDKc5UlJSmDJlCn5+fkRGRvLQQw9RVVXlNGbDhg0MGTIEb29vYmNjWbp0aWt8PEEQBEEQhF8VLhWWnTp14h//+AeJiYn8/PPPjBs3juuuu44DBw4AMH/+fJYvX85nn33Gxo0bOXv2LDfeeKNxfXV1NVOmTKGiooLNmzezbNkyli5dysKFC40xJ06cYMqUKYwdO5bdu3czb9487r33XlavXt3qn7etcdddd2EymTCZTHh5eREbG8uTTz55njBvLBs2bMBkMklfT0EQBEH4taG1MUJDQ7U333xTy8vL0zw9PbXPPvvMOHfw4EEN0LZs2aJpmqatXLlSM5vNWnp6ujHm1Vdf1YKCgrTy8nJN0zTt4Ycf1vr16+d0j1tuuUVLSEiocw1lZWVafn6+8UhKStIALTU19byxpaWlWlJSklZaWnpRn9sVzJgxQ5s0aZKWlpamnTx5UnvllVc0k8mk/f3vf7+oedevX68B2rlz55pnoc3EL/nfShAEoU3wz39q2nPPNeqSrMIybe6HO7XNR7NbaFF1k5qaWufvb6FlaDM5ltXV1Xz88ccUFxcTHx9PYmIilZWVTJgwwRjTp08fOnfuzJYtWwDYsmULAwYMICoqyhiTkJBAQUGB4Xpu2bLFaQ59jD5HbSxatIjg4GDjERcX16jPomkaJRVVLnlomtaotXp7exMdHU2XLl2YPXs2EyZM4H//+x/nzp1j+vTphIaG4ufnx+TJkzly5Ihx3alTp7jmmmsIDQ3F39+ffv36sXLlSk6ePMnYsWMBCA0NxWQycddddwHw+eefM2DAAHx9fQkPD2fChAkUFxc3ar2CIAiCiygogEcegf/7P2jEd/ebP5xg+Z6zPL82uQUXJ7QVXN4gfd++fcTHx1NWVkZAQABfffUVcXFx7N69Gy8vL0JCQpzGR0VFkZ6eDqgdbxxFpX5eP1ffmIKCAkpLS43G2Y48+uijLFiwwHh/5syZRonL0spq4ha6JtSe9GQCfl5N/2f19fUlJyeHu+66iyNHjvC///2PoKAgHnnkEa666iqSkpLw9PRkzpw5VFRUsGnTJvz9/UlKSiIgIICYmBi++OILpk6dSnJyMkFBQfj6+pKWlsZtt93GM888ww033EBhYSE//PBDo4WwIAiC4CLy89WzpkFeHvj7X/ASTdNYuS8NgH1n8qmstuJpaTOeltACuFxY9u7dm927d5Ofn8/nn3/OjBkz2Lhxo0vX5O3tjbe3t/G+oKDAhatpHTRNY926daxevZrJkyfz9ddf89NPP3HZZZcB8MEHHxATE8PXX3/NzTffTEpKClOnTmXAgAEAdO/e3ZgrLCwMgMjISOMPg2PHjlFVVcWNN95Ily5dAIxrBUEQ2hznzkFoqKtX0bYoKrK/zs+Hjh0veMmBswWk5JYAUFZpJTm9kP4dg1tqhUIbwOXCUi8aARg6dCg7duzgxRdf5JZbbqGiooK8vDwn1zIjI4Po6GgAoqOj2b59u9N8etW445ialeQZGRmGk9YS+HpaSHoyoUXmbsi9G8OKFSsICAigsrISq9XK7bffzo033siKFSsYOXKkMS48PJzevXtz8OBBAB588EFmz57NmjVrmDBhAlOnTmXgwIF13mfQoEGMHz+eAQMGkJCQwMSJE7npppsIlS9uQRDaGv/7H1x3HSxeDPPmuXo1bYfCQvvrBhZnfmNzK3V2peaJsHRz2pwfbbVaKS8vZ+jQoXh6erJu3TrjXHJyMikpKcTHxwMQHx/Pvn37yMzMNMasXbuWoKAgI3QdHx/vNIc+Rp+jJTCZTPh5ebjk0djdZfRq+SNHjlBaWsqyZcsaNMe9997L8ePHufPOO9m3bx/Dhg3jpZdeqnO8xWJh7dq1fPvtt8TFxfHSSy/Ru3dvTpw40aj1CoIgtDi6YfHzz65dR1ujpmN5ARzD4LGRAQDsTslriZUJbQiXCstHH32UTZs2cfLkSfbt28ejjz7Khg0bmDZtGsHBwdxzzz0sWLCA9evXk5iYyMyZM4mPj+fSSy8FYOLEicTFxXHnnXeyZ88eVq9ezWOPPcacOXOMUPb999/P8ePHefjhhzl06BCvvPIKn376KfPnz3flR28z+Pv7ExsbS+fOnfHwUAZ23759qaqqYtu2bca4nJwckpOTnXJNY2JiuP/++/nyyy/5wx/+wBtvvAEoFxpUQZYjJpOJUaNG8cQTT7Br1y68vLz46quvWvojCoIgNA7djXMUUkKjHcsDZws4lVOCt4eZeRN6ArA79VwLLU5oK7g0FJ6Zmcn06dNJS0sjODiYgQMHsnr1aq688koAFi9ejNlsZurUqZSXl5OQkMArr7xiXG+xWFixYgWzZ88mPj4ef39/ZsyYwZNPPmmM6datG9988w3z58/nxRdfpFOnTrz55pskJLgmVP1LoGfPnlx33XXMmjWL119/ncDAQP74xz/SsWNHrrvuOgDmzZvH5MmT6dWrF+fOnWP9+vX07dsXgC5dumAymVixYgVXXXUVvr6+HDhwgHXr1jFx4kQiIyPZtm0bWVlZxjWCIAhthnM28SPC0plGOpa6Wzm2dyTx3cMBOJZVTH5pJcG+ni2yRMH1uFRYvvXWW/We9/HxYcmSJSxZsqTOMV26dGHlypX1zjNmzBh27drVpDX+WnnnnXf4/e9/z9VXX01FRQWjR49m5cqVeHqqL4Pq6mrmzJnD6dOnCQoKYtKkSSxevBiAjh078sQTT/DHP/6RmTNnMn36dB555BE2bdrECy+8QEFBAV26dOG5555j8uTJrvyYgiAI5yOOZe00wrF0DINfNbA94QHedAn341ROCXtS8xjdq10LLlRwJS4v3hFcR31bW4aGhvLuu+/Web6+fEqAv/zlL/zlL39xOrZq1apGrU8QBMEliGNZO41wLJMzCjlpC4OP7xMJwOCYEE7llLBbhKVb0+aKdwRBEATBpehunKNDJzTKsdyYnAXAZT3C8fdWHtbgmBAAdqfWf63wy0aEpSAIgiA4Io5l7TTCsdx4WAnL3zg4k5d0Vu3ldqWck80x3BgRloIgCILgiORY1k4DHcvi8ip2nMwF4De9I43jfdsH4mUxc66k0miaLrgfIiwFQRAEQaesTD0AKirUQ1A00LHcciyHymqNzmF+dA33M457e1jo1zEIgMRT0nbIXRFhKQiCIAg6NZ244mKXLKNN4igs63EsHcPgNTfcuKyHajv02c+nm315QttAhKUgCIIg6Jyr4aRJONyOYyi8DsdS0zQ2HFa74f2mlsrvaSO74GE2seV4DvvPXLgXpvDLQ4SlIAiCIOjUdOKkMtxOAxzLkzklpOaW4mkxEW9zJx3pEOLLlIHtAXjrR9nS1x0RYSkIgiAIOjUFkziWdhxFdnExVFWdN2RjsnIrh3cNM9oM1eTey7sDsHzPWdLyS5t/nYJLEWEpCIIgCDoSCq+bmj+LWsLh3yef32aIVatgwAB47z0ABnQKZmS3MKqsGks3n2yp1QouQoTlr5gxY8Ywb968844vXbqUkJCQVl2LyWTi66+/btV7CoLgBhQWQnP2RBTHsm5qpgXUEJZ7T+ex6XAWJhNMiItSB5ctg2uugf37wWG3t3uvUK7lh9tSKKusbslVC62MCEtBEAThl8nevRAeDvPnN9+c4ljWjqbZfxYWi3p2EOGapvH0Nwd5aOMy9rx2Fz1uvQ6mTYO77rKHzE+dMsaP7xOJj6eZwrIqMgrKWuczCK2CCEuhXu666y6uv/56nnjiCdq1a0dQUBD3338/Fbbebv/5z3/o0KEDVqvV6brrrruOu+++23j/6quv0qNHD7y8vOjduzfv2UIiAF27dgXghhtuwGQyGe8B/vvf/zJkyBB8fHzo3r07TzzxBFW2LylN0/jrX/9K586d8fb2pkOHDjz44IMt9JMQBKHN8fPPUFkJGzY035xtxbF89VUlyqrbiJtXWgq273mtvSq+cXQs1ydnsu14DtN3fUNQXjZ8/z18+KE6eeed6jk11ZjDbDbhaVYSxCqb8LgVtWfWCheHpkGJi3YV8PODGn3DLpZ169bh4+PDhg0bOHnyJDNnziQ8PJynn36am2++mQceeID169czfvx4AHJzc1m1ahUrV64E4KuvvuL3v/89L7zwAhMmTGDFihXMnDmTTp06MXbsWHbs2EFkZCTvvPMOkyZNwmL7a/iHH35g+vTp/Pvf/+aKK67g2LFj3HfffQA8/vjjfPHFFyxevJiPP/6Yfv36kZ6ezp49e5r1swuC0IbRReDpZuyJWNOxdFVV+F/+Ajk58MADMHSoa9bgiIPAPuwRRG8wfv5V1VYWrTxEp/wMAstLwMsLXnlFhb/j4+HGG5XIrKiA9HTo0AGw/6qyyvaOboUIy5agpAQCAlxz76Ii8Pdv1im9vLx4++238fPzo1+/fjz55JM89NBDPPXUU4SGhjJ58mQ+/PBDQ1h+/vnnREREMHbsWACeffZZ7rrrLn73u98BsGDBArZu3cqzzz7L2LFjaddOJXmHhIQQHR1t3PeJJ57gj3/8IzNmzACge/fuPPXUUzz88MM8/vjjpKSkEB0dzYQJE/D09KRz586MGDGiWT+7IAhtGF1Y5uQoR83Xt/nmNJmcw7+tSUWF+kyO63E1NoFd7OlDmslXCUubY/nlrjMcySzixjxbqLt/f7jnHufrO3aElBT1sAlLs1kpS9k33L2QULhwQQYNGoSfn31brvj4eIqKikhNTQVg2rRpfPHFF5SXlwPwwQcfcOutt2K2hTkOHjzIqFGjnOYcNWoUBw8erPe+e/bs4cknnyQgIMB4zJo1i7S0NEpKSrj55pspLS2le/fuzJo1i6+++soIkwuC8CvA0V08c6Z559TDva4QlpmZ9tf1bJ3Yqth+DsVevhR6234f2ETvmgMZANzqqfYHZ/Dg86/v0kU9O+RZWmyWpYTC3QtxLFsCPz/X5eU4CMALERQURH4tX1p5eXkEBwc3eJ5rrrkGTdP45ptvGD58OD/88AOLFy9u8PV1UVRUxBNPPMGNN9543jkfHx9iYmJITk7mu+++Y+3atfzud7/jX//6Fxs3bsTT0/Oi7y8IQhvH0c07fRpiY5tvzk6d4OxZ13yXp6fbX7cVYak7ll4+FPjYomK2tR04q557nT2qjl9yyfnXd+6snh2EpckQlqIs3QkRli2BydTs4eiWoHfv3qxZs+a84zt37qRXr17G+z179lBaWoqvLcy0detWAgICiImJAZTIu/HGG/nggw84evQovXv3ZsiQIcb1ffv25aeffjJC2gA//fQTcXFxxntPT0+qaySpDxkyhOTkZGLr+WXh6+vLNddcwzXXXMOcOXPo06cP+/btc7q/IAhuiqNj2Vx5lvqcMTGwfbvrhWVBQevfvzYMx9KPQm/b77e8PLKLyknLL8NkgqDkA+p4fY5lSopxyBYJp1osS7dChOWvmNmzZ/Pyyy/z4IMPcu+99+Lt7c0333zDRx99xPLly41xFRUV3HPPPTz22GOcPHmSxx9/nLlz5xqhblDh8KuvvpoDBw5wxx13ON3noYce4re//S2XXHIJEyZMYPny5Xz55Zd89913xpiuXbuybt06Ro0ahbe3N6GhoSxcuJCrr76azp07c9NNN2E2m9mzZw/79+/nb3/7G0uXLqW6upqRI0fi5+fH+++/j6+vL130LzBBENybmo5lc87ZqZN6dhSWurPWzAWS59EWHUtdWHr6UOBtdywPnFXCd5BPFeYztn+DgQPPv74Wx9Js0nMsW2bJgmuQHMtfMd27d2fTpk0cOnSICRMmMHLkSD799FM+++wzJk2aZIwbP348PXv2ZPTo0dxyyy1ce+21/PWvf3Waa9y4cYSFhZGcnMztt9/udO7666/nxRdf5Nlnn6Vfv368/vrrvPPOO4wZM8YY89xzz7F27VpiYmK4xBZGSUhIYMWKFaxZs4bhw4dz6aWXsnjxYkM4hoSE8MYbbzBq1CgGDhzId999x/LlywkPP39/WkEQ3JDmFpZWq13I2SIyRlV4QYEKtd92W8srobYoLI1QuK8hLCtyctl/Rq1vQnmaGhcbC0FB519fS46lWarC3RJxLH/lDB8+vNZweE2eeOIJnnjiiTrPm81mzp49W+f52bNnM3v27DrP6+HsmiQkJJCQkFDrNddffz3XX3993YsWBMG9ae7incJCo8+iISx1x3LnTjh+XD1uvBF++9uLv19dtOFQeImXr5FjWZKVa+RXDj1nE4y1hcGh9lC4WYp33BFxLAVBEIRfJs3tWOrzeXtDRIR6rQvL3Fz7uD/8oWVzL9uyY+npQ5mvaqdXlZPL/jNK+MaeOaLG1Va4A/ZQeH6+8ZnMUrzjloiwFARBEH55VFQ4b0TRHMJSd0BDQ+29iHUBqfeV1O/197+ff312NqSlOR0qq6xmzYF0KqocdiezWuGxx+C112pfR1sUlg7thjp1U62YrHn5pOSqf4Owo0lqXF2Opb+/2n4TjHC4EQoXy9KtEGEp1MvSpUv5+uuvXb0MQRAEZ2o2Ds/IUGKzOeYMCalbWOrO23PPwZEj9muPHYM+faBfPyc389UNx7jvvUT+s+mYfeyHH8LTT8ODD9pD7zU/i04bEZaaLSRf7OVLv74qTcCjUB3r4W/CkpysBtYlLOG8PEuz9LF0S0RYCoIgCL88dBEYFKS2ENS089zCRtMQx/Kmm2DSJCVir79e3bO4GG64QY05dw4OHDCm3HpcXffTUdv1ZWXKrQS1z7ljiF2nDeZYVhXYi3eGDOwGQGB5MWgaE6qzlECOjLQ3lq8NXZTb8ixlS0f3RIRlMyFbUrV95N9IENwIRxGotwa62HB4bY5lebmzAAwPV/tgd+wISUkwejTcfjvs22efx7armNWqGe149p7OU/0aX3nFqTLaaZcdUELWMX+zjTiWFefUOqr8/OjaQ23J6GmtxreynJH5toKcwYPrb8VUp2Mp383uhAjLi0Tf4aXEMddHaJPo/0ayK48guAG6CGxOYVmbYwlK6OmOZVgYdOsGmzZB165w9Cj873/g4QHx8WqMTVieyCmmqFxtM1tcUc2xwynwt78537OmsHQMg0ObEZaV+cqxtAQFYQoMxGpS8iGwvJjeabYdd+oLg8N5wtJi7BXe7MsVXIi0G7pILBYLISEhZNq+HPz8/IxtqoS2gaZplJSUkJmZSUhICBaLxdVLEgThYtFFYEiIPfzanI6ll5d6VFQ4C0u9AKV7dyUux49XuZb//rcKB2/ZAocOAbDvtBKFoSX5XHFyF0Ez/qHW3a8fBAfD5s3nC0s9DB4YqCqxi4uhuhpc/L1lteVTegYHgclEuX8AvkUFBJUXE7UvUQ269NL6JzkvFK5+V8rOO+6FCMtmIDo6GsAQl0LbJCQkxPi3EgThF46jCGwJxxKUa5mbq4SlYyhcJyYGdu9WDlzfvvD99+q4zbHcdyafyYd+5KX/PYOHZivSMZng2WfhzTfV+6ws5zXowrJXL0i0CbaCAvuaXEWhCs97hwYDUB0UDEUF9K/IxSPJllM6alT9c5wXCldvJRTuXoiwbAZMJhPt27cnMjKSyspKVy9HqAVPT09xKgXBnagtx/Jim6Q7ilVwFpY1HUsdPz8lKkFVhYOqEC8vZ9/pfO7bvw4Pzcrx0A7suGQMt/xzPtYhQ9n/ynsMhLodyy5dYP9+leOZn+9yYWkuVsLSN0wJS5+IMDibyr2Fyp2ld29VvFMfurBMS4PyctnS0U0RYdmMWCwWES+CIAitQUs4lo55m2DPsywsdM6xrIv27VWVekEB1cmHOXDmHMPOKPdy/tV/YG/H3lzVfxCJR7LYnasxELBmZDgXO+g5ltHRKlyemdm6eZYrVyqx7LDlLoBHSTEA/uEh6n2Y+hn13/WDGnD55ReeOyICfH2htBRSU8WxdFOkeEcQBEH45dGSxTuOjiUoh61KFeGc51g6YjIZ7mXm9l10OnuCkLIiNH9/zvXuj6apvMu3fzpJjp9y/irOpjvPYXMsD1h9yfHwVcdaq+XQ0aNw9dVw1VUqt9MBzzJV/BjYzia69Z+RXuHeEGFpMjnlWcqWju6JCEtBEAThl4ejCNSF5dmzqtClqdQMhQcGqmddPPn4KDevPmzC8lziXoanqtxDU3w8A7qpLSI/SzzNpsNZ5Pipe1TXEQr/7HQVZ6y2Dhat5Vh+9pmKS5eWwrZt9uNVVXhXlAMQEmkTlsHBztdecUXD7uGQZynthtwTEZaCIAjCLw9HxzIqSlVNV1ef366nMdRWvAN2YVmfW6ljE5bVSQcZmbpfHbviCi6JCQHgq10qD1R3LMmsvXjntFcQhd42EdtawvLTT42X1h9+4IvE05zMLnZyL0MjbT8DXXyDCtt3796we3Ttqp6PHZMtHd0UEZaCIAjCLw9Hx9JigQ6qafdFhcNrK94Bu7CsL79Sx1bAE3D8CMNP26qlr7iCwTZhqVMd0Q4Aj5zahWWWfwiF3v7qWAuHwv+97gi//cMyVeFuI+Pb7/nDZ3u4e9kOKvOUsK00W2gXYXNxHR3LK66ovzG6I717q+fkZKPdkOhK90KEpSAIgtC6NEf3jJoi8GLzLCsqQN/oohkcy85njhFdlIvm6QkjR9K/YzAeNouuf8cg+g/uAYB3Qb7956FpdmEZEEqhl01YtrBj+enPqQz72dYqyRaqDt69A4u1muNZxazcrPZEL/byJTzAR41zdCwbkl+p4yAspXjHPRFhKQiCILQeBw4oUbJw4cXNU7OC+2KFpT4fqMpusAtLW0PvBgnLbt2wenlh0ftWDhsGfn74eFoM1/Ley7sT0D6KatvuNWRnq+dz5wyRme0XSpG3rXinBYVlQVklp8+VcvUhVd2d8+D/URkYhF95KX2yTgLw6YYkAMq8fY3dcs5zLBuK3pLpyBHjZyTC0r0QYSkIgiC0Hj/9pJzBVauaPoemnV/BrVcbHz/etDn1+YKD7bvc6MJSdzIbIiw9PCjs3M14a3IQXc//djCvThvCdYM7EBniS66fTcDqBTw2t7LAN5AKD08KvFvesTyUVki33DPEZZ6g0mzh9YjBJHXpB8A9nMHH04y1QPWwLPf1t1/oWOA0cGDDb9i1q9rRqKyMqHPqc4uudC9EWAqCIAith77TzNmzTZ9D3+YQ7I5lXJx6PnCgaXPWFKrgvF84NCzHEjgb3cX+xkFYdg73Y/KA9phMJiIDvcm2VYYbwtJWeJRhO24U77RgjuWh9AKuOvQjAJu7DOKd5AJWh/UEYFLeMW4d3hn/ilIAqhyF5ZAh6udz662N227SYoHYWADaZygnWLZ0dC9EWAqCIAithx72TU9Xe2s3BV0Eenqqhtug9t+GpgtL2/7edO7MzpRzjPrH9+wvqLG+hjiWwOFQFZbXTKY6tzlsF+hTp2OZ7R8CYC/eaUHH8mBaAWOOq60j9106gcpqjZ87KZHut30L917elcCqMgCqHYV2167q3/L11xt/U1s4vEOGyl2VULh7IcJSEARBaD10YVldff4+2Q3FMb9Sr0bWHcu0NPu+3o1hxw71PGIE3yVlcCavlJ25NYqMGigstwcoYVnSt3+dWzFGBnobvSy1GsIyyz8UD7OJIq+WbzeUdLaA3rZcykE3TQJgT3RPrJ6ecPYsnfIzGRXlDYBHUKDzxd7eDa8Gd8RWwNMhXQlL0ZXuhQhLQRAEofVwFJNpaU2bo2ZFOKhcP735dlNcS11YDh9OZqFqBp5r8nIe0wBhWVBWyUedhvLM6OlY//NGnePaBXobvSzL9d13HITl4JiQFg+FV1s1zh05QVBFCZrFwmVXXca0kZ2ZOSEO89ChatCPP3JdD+Wsdo5p1zw3tgnL6HRxLN0REZaCIAhC66E7ltD0PMva8iHBHg7fv7/+65csgchI2LlTvS8vt/dwdBCW2Xg6X9eAHMvD6YVYzRa+mjSdwFEj6xzn42mhKFjNZwhLm9DO8g8lvkd4i4fCT+UU0yX9pHrTsycWH2+evmEAf5zcx95CaPFiPFeuAMAjJLj2iRqLLRQenaZyLCXF0r0QYSkIgiC0Ho7C8mIdy5ph5v791fOFHMsPP1TO6auvqvd796o2P+Hh0K0bmQUqpzBLqyEsG+BYJmcUAtA7OvACI6EqTG3zWJWuinY0myA+Gdqe+O7hRlW41kLC8mBaIT2zUwEw6aJcZ8IE9bxrF2zapF7XFPJNxeZYhuZl4V9eIo6lmyHCUhAEQWg9HEPhLeVYXkhYHj2qnv/3P5Xr6RAGx2Qiy+ZYZlg9nK9riLBMtwnLqAsLSy3SFlrOyoKyMti3D4CDnXozuHOIvY9lQUGLJCIeTCsgNsfWo1PPUdWZOBGWL4e//Q3uuUdVf99zT/PcOCREOcZA99wzbVpY/uMf/8BkMjFv3jzjWFlZGXPmzCE8PJyAgACmTp1KRo2tRFNSUpgyZQp+fn5ERkby0EMPUVVV5TRmw4YNDBkyBG9vb2JjY1m6dOl591+yZAldu3bFx8eHkSNHsn37dqfzDVlLayPCUhAEQWgdSkrsPSGh5RzL+kLhBQX2KuzMTNi6FfRf1sOHU1ltJae4Qp2mRo5lA0LhhrBsgGNpiYoCbNs67t6NqaqKLL8Qgnp2x8/Lg+pAldtoqq5WP7fycqw7fm56NX0NDqYV0Cu7DmFpMsHVV8Of/wxvvgkffWS0CWoWbOHw7rmn2+xe4Tt27OD1119nYI0+nfPnz2f58uV89tlnbNy4kbNnz3LjjTca56urq5kyZQoVFRVs3ryZZcuWsXTpUhY6bApw4sQJpkyZwtixY9m9ezfz5s3j3nvvZfXq1caYTz75hAULFvD444+zc+dOBg0aREJCApn6f78NWIsrEGEpCIIgtA45Oc7vm+pY1la8A0qsmEwq3O7wy9eJI0ec33/1lVNFeHZRuXGq2NPXPi4oiNxyKyUVzq6TI5qmGaHwXg1wLL3bK2Hpcy7HWMPe9j3p20EJSq+gQPvuPPn5HJw1D/OI4Rx7oQktfmrhUFoBPXVhWTMU3tLYwuHKsWzdWzeEoqIipk2bxhtvvEGowx8w+fn5vPXWWzz//POMGzeOoUOH8s4777B582a2bt0KwJo1a0hKSuL9999n8ODBTJ48maeeeoolS5ZQUaH+aHnttdfo1q0bzz33HH379mXu3LncdNNNLF682LjX888/z6xZs5g5cyZxcXG89tpr+Pn58fbbbzd4La7ApcJy0aJFDB8+nMDAQCIjI7n++utJTk52GjNmzBhMJpPT4/7773ca01yWsyAIgtCC1Gwv1FTHUg+F13Qs/fygh9qDu07XUg+De9jC3J98AgcPqtfDh5NZYBeWJV4+xuvqsDAu+8c67nxrO1ododuswnLySioxmyA2MqDWMY74duoAgHdpiZHHuDe6J33bK2EZ4u9NkZd9W0fPLT8BULxu4wXnvhD5JZVUnj5DcHkxmsUCvXpd9JyNwiYse+ScbpVQeGFhIQUFBcajvLy83vFz5sxhypQpTNBzTW0kJiZSWVnpdLxPnz507tyZLVu2ALBlyxYGDBhAlM2RBkhISKCgoIADtjSNLVu2nDd3QkKCMUdFRQWJiYlOY8xmMxMmTDDGNGQtrsClwnLjxo3MmTOHrVu3snbtWiorK5k4cSLFxcVO42bNmkVaWprxeOaZZ4xzzWU5C4IgCC2MXrhjtv3qaW7HEi6cZ6k7ltdfr/ownj6t8hdjYiAqyqgIB6i0eKp+jkBJQAhllVYST51j63F7n8wNyZk89vU+8koqOGQLg3eN8MfH88K70YREh1NuUQJXW7MGgD3tezK0ixLMIX6e9srwvDzan1bbVQYdP3zBuS/EwfQCeubYCndiY9XPojWxhcJ75J5ulT6WcXFxBAcHG49FixbVOfbjjz9m586dtY5JT0/Hy8uLkBr/7UVFRZFuaxeVnp7uJCr18/q5+sYUFBRQWlpKdnY21dXVtY5xnONCa3EFHhce0nKsqrFX7NKlS4mMjCQxMZHRo0cbx/38/IiOjq51Dt1y/u6774iKimLw4ME89dRTPPLII/z1r3/Fy8vLyXIG6Nu3Lz/++COLFy8mISGh5T6gIAiCYEcXlrGxcPiwffcdcyM9jrqKd0AJy//+t27HUheWgwergpkVqpUOI0YAkFlY5jS82tcfc2UeJYH2VjvvbjlJfI9wCsoqmffJbvJKKjmRXcwVPVUxTp8G5FcCRAb5kuMXQofCbEy2XpVHO/chTncs/TztvSyTkvAvU6ZLZMoRJYab0pzchsqvVH0kz8uvbA1sjmXXc2ls1LfnbEGSkpLo2LGj8d67DiGdmprK73//e9auXYuPj0+tY4T6aVM5lvm2lgphNRKkP/jgAyIiIujfvz+PPvooJQ7J381hOdekvLzcyTIvLCxsls8nCILwq0YPhffvr0RRVZVz+6GGUlfxjj431O1Y6qHwnj2Va6kzfLhaYqFziLTCVwm7Iv8g49iapAzS8kt5Y9Nx8krU7jw/Hc3h3+uUaG1IfiVAZJA3uX52wXo6qB1d+3bDw6J+NYf4eRkth6w//miM8yspsru95eVw001QjwNXG98fyjRaDblEWHbtSrXFA9+qcnyzW95dCwwMJCgoyHjUJSwTExPJzMxkyJAheHh44OHhwcaNG/n3v/+Nh4cHUVFRVFRUkKf/N2gjIyPDMMCio6PPq8zW319oTFBQEL6+vkRERGCxWGod4zjHhdbiCtqMsLRarcybN49Ro0bRX/9iAG6//Xbef/991q9fz6OPPsp7773HHXfcYZxvDsu5JosWLXKyzONc8T+dIAiCu6GLyPbtoZ2t1U5Dw+FVVSonU9MaHgqvLcaqO5axsXDNNXbXzyYsM2sIy3KbsMz3swvLaqvGv9cd4c0fTgBw6/AYAEoqlPPWkFZDoG/raBeWe6N7Mryr3VgJ8fWkyOZYVv+02fli3ZFdtw6++AKefFK1TmoAZ/NK+fFotusKdwA8PKj0UuLOUl7R+vevg/Hjx7Nv3z52795tPIYNG8a0adOM156enqxbt864Jjk5mZSUFOLj4wGIj49n3759TtXba9euJSgoyNAT8fHxTnPoY/Q5vLy8GDp0qNMYq9XKunXrjDFDhw694FpcgUtD4Y7MmTOH/fv386PDX2UA9913n/F6wIABtG/fnvHjx3Ps2DF66Enazcyjjz7KggULjPdnzpwRcSkIgnCx6I5lRAR06KAqt9PSVFj6Qtx3H7zzDnTrZi/6qc2x7N0bLBYlPvfscZ47P9++hp491TaQjz+uindsO83oxTs+nmbKKq2U2YRdrrcSi33bB3EwrYCPtiu3b1BMCItuHEBkkI/hWDak1RBAgLcHeQH2z7C3fS9Gd7O/D/XzMkLhnkec8yq1/fsxJSTAZpvgLCtTbqwtxFwfX+06g2bV6HvOhY4loNlSILRWCIU3lMDAQCdzC8Df35/w8HDj+D333MOCBQsICwsjKCiIBx54gPj4eC699FIAJk6cSFxcHHfeeSfPPPMM6enpPPbYY8yZM8dwSu+//35efvllHn74Ye6++26+//57Pv30U7755hvjvgsWLGDGjBkMGzaMESNG8MILL1BcXMzMmTMBCA4OvuBaXEGbEJZz585lxYoVbNq0iU6dOtU7duRItUXW0aNH6dGjB9HR0ec1DG2s5VwTb29vJ5u8oIX2aRUEQfhVoTuWurDcvbvhjuVPqiKaEyfsx2rrK+ntDWPHwnffQUKCcvR0oaCHwaOilKgEJSwdyLLlWMZGBrD/TIFRGZ7lraq8bx0ewysbjpJhE6CPJPRWDbTH96S8qprKKo1uEf4N+kgmk4nSUPtn2N+hF7+PsQvLYMfiHRsH23Wlb9ZJqvbuUxtObnZwMvfurV9YFhSgjR9P36ogug/7LQElhSq/tQFitCXQbK2UtKq2IywbwuLFizGbzUydOpXy8nISEhJ45ZVXjPMWi4UVK1Ywe/Zs4uPj8ff3Z8aMGTz55JPGmG7duvHNN98wf/58XnzxRTp16sSbb77pVPdxyy23kJWVxcKFC0lPT2fw4MGsWrXKKQJ7obW4ApcKS03TeOCBB/jqq6/YsGED3bp1u+A1u237ubZv3x5QdvLTTz9NZmYmkbZO/rVZzitXrnSax9FyFgRBEFoBXVi2a6fC4dDwlkO60/j883DokBJDdTUs/+gjuPJKJVzHjFEic/Bg5zB4Heih8F5Rgew/U8DJTj2J3bedfVEqQhYV5M30+K78a3UyV/SM4LJYtS2j2Wzi0cl9G/ZZHNC3dQSwDhmCr5e9mjzE15OjevGOjeV9R9M36yTV+w/gWVVlb+4OSljefHPdN1u5EtPPPzMOGHrIdl2PHuCiIhWr2fZZm6nhe0uxYcMGp/c+Pj4sWbKEJUuW1HlNly5dztMdNRkzZgy7du2qd8zcuXOZO3dunecbspbWxqU5lnPmzOH999/nww8/JDAwkPT0dNLT0428x2PHjvHUU0+RmJjIyZMn+d///sf06dMZPXq00Qnf0XLes2cPq1evrtVyPn78OA8//DCHDh3ilVde4dNPP2X+/Pku++yCIAi/OmqGwqFhjmVlpb0SfNo0eP11cEhXOo+ICOVUDh+umrJPmWIPFYMKg9eC1aoZxTt6nuSb1/0O0tPZHKVcvTB/b/7f6O68eOtgXr5tyIXXfgG0dsoQORbWif5xnZ3Ohfp7UeRlF5ZlHl58F6uq1z0PHVShfsf2fLYtIevEIRcvuKxIvXBFfqUNzazyW9tSKFy4eFwqLF999VXy8/MZM2YM7du3Nx6ffPIJoJJXv/vuOyZOnEifPn34wx/+wNSpU1m+fLkxh245WywW4uPjueOOO5g+fXqtlvPatWsZNGgQzz333HmWsyAIgtDCOIbCG+NY6jv2mEwN2q8bUG7m2rXQubMSr8uWXdCxPFdSQZVtG5ieUSr0nVdaBVFRxjaPYf5eeFjMXDe4I8F+ng1bSz1kXXoFSZHdeGfYtU6FO6AcywKHUPjR8BiOh3WiwuyBpaQYPv4YgHy9cnzv3lrvsf9MPs+uTiZnuWrxt2DKfPIus7X0s+WWugLNcCxFWLoTLg+F10dMTAwbN154h4HmspwFQRCEFsJqtQvEdu0a51jqTmdYmCrMaSjBwcrZnDcPnn3WLkrrcCz1MHiYvxftAlR4+FxJBRVVVgrL1G5uEQFetV7bVHy6d+GqmS8B8IcuzsVIIQ7FOwDJEZ2xenhwLLwTfbNOKrEMfNl/HDMTl2M+cULthR4UBG+8AW+9Rcorb3PjF6eIzE3j/zJOU2m2cGDkeIL++AwkH4K+jQ/fNxf24p22HQoXGkebaTckCIIg/MJYuxYcIkj1kpdnb4cTHl6/Y7l1q/Oe3rqw1FsUNYZ771WC9OhR2LZNHbuAsIwM9CbE5kbmlVSSa3MrLWYTQT4X71I6EhWoBGzPyABC/Z1Fa7Cvc/HOkYgu9GgXwJEIW8jc9nNZ12MEaQE20bx/vxLxCxfCtm0c+sNfqKi2cnO++nme6zeYN+eMxezpoYqaGiPUmxldWIpj6V6IsBQEQRAaz7lzcPXVcMMN5+8BXht6GDwwUFVu645lWppz8UZaGlxxhSq+0bkYYenvDzWLH+oIhWcWqIrwdg7CsrzKypk8tSlHqJ8XZnPTd7upjfF9I5nUL5r/Szi/MtvLw0yVv33P8eR2XRjQMZjkiC7GMavJxO4OvUlu1xWAyt17lIC29XG+bPO3hFWXcW+lqqaPvG4yMWHOBUGuQq8KRxxLt0KEpSAIgtB41q6FigrlQta1y40jjhXhoFr+6Lvv6CFyUI5bVRWcOgVFtgKTixGWoISl3lrOsdVQDeyOpQ8B3h542ETksUxVINPcYXBQ4e7X7hxKQr/ad0oxOTSBPxzRhX4dg+2OJXC6Yw+KvP04FNkVgIwft8NXXxnnAypKWVy5H/+fNqkD48Y1+2doMm2wj6Vw8YiwFARBEBrPt9/aXx88eOHxjhXhAJ6ete++o1duA5w+7XxtU4Vlu3Zw993qdR1hcLBv5xgZ5I3JZCLETwnJY1lK4Ib5N7+wvBAVUdFUm8xk+YVQENmezmF+JLezO5a7OqocyYKe6rly5240m7DcGqP6d47+6BXlBPv4QBtqs2eEwkVYuhUiLAVBEITGYbXCqlX29w0Rlo4V4Tq15Vk6CstU284wNmF5sNrngkWfdfL443DHHSr3sA4ybc3R2wWoVnWhtnC4K4WlJTKSO255ium3PEl0sC/tAr1JDY6i3EOt5cdIJZSHXqOqvGMO78V09CjlFk/+cO1DVPv6YdK3Fhw1ymU9K2vDnmMpoXB3QoSlIAiC0Dj27DFy+IDGCUtH17G2yvB6HMuPT5ax8bBzPmdVQ/Pz2rWD995zzt2sgb6dY2SQEpYhhrDUQ+HetV/YggT7ebKlyyAORnYnKsiHiAAvrGYLy+N+Q1X7DnzfZQieFhOjr7mCKrMFD039PH7sOpiH7hmP5c477JO1pTA4Du2GxLF0K0RYCoIgCI1DD4N37KiemxIKB7uw1AUk1OtY5voGsfPUOeN0cnohA/66hnuX/UxhWWVjPkHtSyyy51gCRig8JVcV77jCsQzxtVehK2GpxO3/Tf4969YmkuMfQscQXzx9fciO6W6MDbj1Jq6/pCP87nf2ydqcsLQVQolj6VaIsBQEQRAahy4s9WrrM2cgP7/+a2oLhet7VOvFP1YrHD9uP19DWOb4BbPvjP0+aw6kU1pZzXcHM7jp1S2cPlfSlE8DqL7KhmMZ6BwKr7Y1TXeFsAz1s98zOtgbH08LQT6qBfXOFCWy9Spv/2FqJyCr2czIB+9SFw0aBI89BvffDyNGtN7CG4I4lm6JCEtBEASh4eTlwZYt6vWtt0K0rZr50KH6r6stFG7bmhd9x5izZ9XWizo1QuG5NmGp51nuTs0DVHF5ckYh1y/ZzIGzFxC4wPeHMrjzrW1kFNjvVVReRWmlEjh6KNxR1EHLVIVfiBCH3X2ig5STGmETvrtO5QHQKVQJy8BLhwFgHjUKIiPtkzz1FLz6qlGF3VaQBunuSdv6r0wQBEFo23z3nXKY+vSBrl3tO7dcKBxeWyh80CD1fPgwlJY6h8FBOZZWK5qtHVGOXzDZRRWk5ZehaRq7bMLy5duG0Cc6kOyicu56Z8cFnct/fHuIH45k83miPQSvtxoK8PbAz0s5giE1hGWYvwtyLGuEwsFeXLT3TB4AMWG2Vkr33Qe//z0sWdKqa2wqeo6lSRqkuxUiLAVBEISGUVkJr7+uXk+erJ4bKixrC4VHRSkH02pV/St1Ydm1q3pOTYXcXEy2HLxzvkEA7DuTT2puKbnFFXhZzEyIi+ST/xdP76hAsgrLmfnODvJLa8+5PJFdzOEMVeW93yGsfuZcKWB3K8HZLQTXh8KjajiWZZXq59JZb3geFAQvvAADBrTqGpuMLcdSkxxLt0KEpSAIgnBhqqvhrruUY+ntDTNmsCc1j5ORtmbdNYRlVbWVH27/HRnh7amYfJXKwwQqQsN4Y9NxjmQUqhi27lru3WsXlmPGqOeCAuNYvrc/VRblJO47nc+uVJVfGNchCG8PC8G+nrwzczhRQd4cySzitv9s5YnlB1iy/ih7T+cZ61qbZK9m3+8QNtdzN/t1CDaOhdYQli4PhQc7O5Y6MaFtYyedxqLp20mKsHQrRFgKgiAI9aNpqvjjww/BwwM+/5zyfv2Z9uY2Fh62iQIHYVlaUc2iv73PqI9eIyo3Ha9V30K5CjWvy9F4euVB/rrcVrCj51nu2WMXloMHg77jzK5dgAqD6+w7k8+ulDw1NCbEON4hxJd37hpBgLcHSWkFvPPTSf61Opnb/rPVaH6+5kCGMT41t5S8ErUPuJ6vOaiT/T6OofCW2Ce8IehrsJhNRkV4u8AawrKNbNHYaGw5liYp3nErPFy9AEEQBKGN8/nn8OabSgh88AFcfTUn0wspKq/iYGgnNeb4cSgrI89qZuY723nsP4swo7Gm56V4XDmecSVnYNAgDpQol2r/mQI0TcPk6Fjm5anXPXpATIx6bxOWuX7BmE1g1ZSw1AXhJZ1DnJYa1yGIFQ9czncHM8gprmD1/nSOZxfz8vdHmDuuJ4m2SuoQP0/ySio5cLaAUbERhqs5yEGoOoahw/ybf5/whtA9wp+JcVF0jfDHYru/o2Pp72U5z1n9paDnWErxjnshwlIQBEGoH70d0IwZ8NvfAnA4oxCALP9QKgKC8CoqgCNHeP2MJ11Wfc3Qs4co9/blL1fez/BRAxh3u2qFc+KDnQDkl1aSll9GB0fHstKWFxkbq4Tlvn2wU43P9QsmrkMQh9IKyS2u4JwuLGNCz1tu1wh/7r1C9XS8IjaC29/cxofbU/D39kDTYGCnYGJC/fhmXxr7z+TTo10AGQXlmE3Qr0OQMY+jYAt3QX4lgNls4j/Thzkdc3QsY8L8MJlaX/A2C7pjKcU7boWEwgVBEIT6OWdrSq63FgKVIwlgMpHZsZt6ffAgh4+e4dEN7wCQOns+GYERHLftXANwPNv++lB6gSr+8fBQ7mRxscq77NYNOtmc0H37AMjxDSLc35teUYGAis6H+3vZK6Lr4LLYCK7oGUFltcYrG44BMDEuin4dlYDcf7aAPTa3sldUoFERDmrXGx1XFO7UheMOQJ1+ofmVIFs6uisiLAVBEIT60YWlnvcIHMksMl6faGcr4Fm/nvsWP0RUUS7lnbtiXjBfnc8uxmrVsFo1TjoIy4NphaoQSK8sB+jcWR2LiVHvK5QzmesXTKCPBwMdciAHx4Q0yK17ZFIfp/cT+0XT31aks/9MPnuM/MoQp3HeHhb8vFS4NtwF2znWhaNj2fmXml8JYLFJEMmxdCtEWAqCIAj1o+c+htrDznooHGBfUHv14rXXGHl8F6Ue3mivvUZMhzA8zCZKK6vJKCwjo7DMaEIOcDCtQL3Qw+GgwuBgF5Y2lLD0ZICDsKyZX1kX/TsGc80gtX1k13A/ekYG0L+jmudEdjE/HVN9Mh3zK3X0PEtXhcJrI9yhOv1Cjm1bxtgrXBxLt0KEpSAIglA/umNpE5YVVVZO5tibkCf62UPk+6N6MOv3r+MzOQFPi9lw1I5nFTuFxMFBWOoFPGAXlnoo3EaOXzBBPh4M6OgoLM/Pr6yLP13Vhwl9I/nj5D6YTCbC/L3oGKJEme5YOrqhOnq7n7YUCve0mI38z19qqyHAIcdShKU7IcU7giAIQv3ojqUtFH4iu5hqq0agtwcmE/zYeRAF11zP8YgYbg4fx2/620Vhtwh/jmcXczyrSOVPAgM6qq0ZT2QXU1ZZjY+jY9mjh3qu4Vjm+AXTw9uD3tGBRAR4U1ltrVUI1kX7YF/enDHc6Vi/DkGcyVON0b09zPSODjzvOl1Qhrugh2V9XBYbwQ+HsxjcQNe2TWKRvcLdERGWgiAIbZ1Zs1TF9DvvGOKsNanIzsELWJ5SyjXYw+CxUQGYgJ1lVWz62xJ+OppD5fYU+jgItO7t/Fl3SBXtmFBrH9EtjDN5auecIxlFDGiAY6nnWHp7WPjqd5dh1TQCL7KvZP+OwaxJUn0t+3UIwtNyfhDvrsu64mkxc2Vc1EXdq7l5+bZLqKi24u1hcfVSmowmVeFuiYTCBUEQ2jJFRaqH5LJlcPasa9aQp3alWZaUh9WqGYU7vSID6dEuAICjmUWG4OzlJCzV+eNZxZzILrId86dvezXmYFqB2tpRdyj791fPfn4QHm7Mk+MbbAjJmDA/uoT7X/TH6t/R3lqotvxKgPF9o3j7ruFEBvpc9P2aE5PJ9IsWlQCY9KpwzbXrEJoVEZaCIAhtmYIC++uTJ1v+fpqmxKxOZSVeZSqf8kilJ9tO5BqthnpGBRAb6SAs09Xx3lF2YdktQgnA49lFnLBVhHePCKBPtBJ1B9MLlAu7fDn897/Qs6f93g6upe5YNif9HfI1a1aEC62ALozFsXQrRFgKgiC0ZQrt1detIiwffVQV6fz8s3qvF+4Ahd5+fL3rjOFM9oyyO5ZbjuVQWF6Fh9lkiElQ7iTA6XOlpJ4rNY7p4XKnAp5rr3Vei83FLPb2o8LD86JD3zWJDPShV1QA3h5mhncLa9a5hQZgkuIdd0RyLAVBENoyrS0sN2yAqirYvBmGDTMKdwq8/LCaLazcn0ZJhXKYekYGUFGlREFOseo32aNdAF4eds+iXYA3gd4eFJZXUa1p+HlZiAz0pm975VgeSi9UWzvWljtqE5bnbPuEN7djCfD+PSMpKKsyKsSFVkSKd9wScSwFQRDaMq0tLNPSnJ9tjmWBj3IeC8uqjIrw9sE+dAr1xcuh6KVXjcpqk8lEt3Z2B7NbhD8mk4nYyAAsZhN5JZVkFJTXvhZbKDzbT4nQoGZ2LAEig3yMcL7Qytj2PhfH0r0QYSkIgnAh9HY7rcFPP0F2tv29Y75jSwtLTYP0dPVaF5a6Y+kTwGU97MU0sVEBmEwmPCxmp9B376jzRVr3CGdhCeDjaTGOG+HwmgwYAMCx0I5AyziWgguxqH9PEZbuhQhLQRCE+njhBZVz+PXXLX+vbdvg8svhrrvsx1rTsczNNbZQNASmzbHM9wngrsu6GkN7Orh8PSLtwrFX1Pm9IPXK8Jqv9XD4R9tTqKpW4uJ4VhE3vvITj329D6ZMoWDFKp4Yfx8AASIs3QvZK9wtEWEpCIJQHxs3qucffmj5e+3bp56PHbMfcxSWp0617C9h3aV0eF2RpbY7LPD257LYCKPi21FAxjqIRb3a2xGnYh6H19NGdsbTYmJNUgbzP93DT0ezueGVzexMyeOj7alUaXDu0ssp8AnA19NSa59J4ReMRfpYuiPyf6kgCEJ9nD6tnlNTW+9eDpXYTsKystJZ/DU3tQjLUpuwLPQNxN/Lwt9v7M9NQztx8zD7zjg9bO6lr6eFTqHnF8F0b3d+KBxgZPdwXpk2FE+LieV7zjLtzW3kl1YCUG3VSMsvo7CsCpAwuFti7BUufSzdCRGWgiAI9aELSl30tST6PRxzOh2FJTQqHH4ko5DSika4QY7CMisLKisNx7IiIAiTycTQLmE8e/Mggn3thTTDu4bh42nmN73aYTafX93dPUK19PG0OBfyAFwZF8XLtw/Bw3bdlIHtjf3FU3NLKChTQlOEpRticyzN4li6FfJ/qiAIQl1UVECG2vKvVR3L8nIoKwMfn/OF5YkTMGrUBaf6/lAGdy/9mdtGxLDoxoEXHA+c74ZmZFCVq9zTquC69+XuEOLLz49diZ9n7TvB+HpZeGvGcKqs1loruxP6RfPZ/fGknivl6gHtmbl0Bym5JZw+V0qwnxrf3D0shTaA4VhKjqU7IcJSEAShLhy3UExLU/0dPVrwa9PRFT13Dtq3b7Jj+dUutfYfj2Y7n3jnHUhOhkWLzt93vKawTEvDmpsLgDUopN77BXjX/3O5vGdEvecv6RzKJZ1DAYxweuq5Esxm5XCKY+mG2PpYSlW4eyGhcEEQhLpwFHrV1Xb3sjXup4fDdWEZZtsZpgHCsqLKyobkTABSc0sptIWTsVrRHngA/vlPe6GQIzWFZXo6Jn0doaEN+gjNQYxDKFxfe0v0sBRci8kiO++4IyIsBUEQ6qJmXmVLhsMLCyE/3/5eL+DRhaWtp2NDhOX2E7lG0QtAsm0Pb86exVSs9usuTko+/0LdodWdzLQ0zPl5AJjDWlFYhtqE5blSKd5xZ8SxdEtEWAqCINRFTSHZkgU8Z844v6/pWDZCWH530NlZ1RuQ5+89YBw7vTPp/AttjmVh11jjvVeBErueEa23l3ZMmC0U7uBYirB0Q4w+llK8406IsBQEQaiL1nQsa95Ldyz1nXd0YZmSUu/eypqmsTZJCUu952RSmhKnp3fYw9/Fhw7XvNAQlt/5qp1uSEvDu1hd6x0RTmvRyeZYZhaWk12kGrZL8Y77YdIdS00cS3dChKUgCEJd6GLPz8/5fWMpL69XDAJoNUSrtWYovHdvVTh0gV6WB9MKOZNXio+nmXuu6AbAoXTlWBbuO2iMM5844XxhYSGUlABwIKq7Onb2LL42YekX1XrCMtTPE38vJToO2cL44li6IWa93ZAIS3dChKUgCEJd6EJyxAjn940hPR1iYuDqq+sdln3ouNP7kkzVP9IQliEh0Lmzel1POFwPg18e244hnUMAlWNptWqYHHb0CUo/TbVjY2o9DO7ly6nQDurYkSOYbW6SX1T9Vd3NiclkMgp4jmbqwlIcS7fDQ3Is3RERloIgCHWhC8lLL1XPTQmFf/yxaja+atX5eZQOZB486vS+NMPWJkgXloGB0LWret0AYXllXCRdw/3x9jBTUlHNsawiQs/Yr+t4Lp1jGQX2C23CMjMgjEx/W6HOUbWmcosHIWHnb9XYkujh8MpqJX7FsXQ/TGapCndHRFgKgiDUhmPIOT5ePTfCsSyrrFaFJ599Zj+4Zk3d40+kAJASHAVARU6uCp/bwtNOwvLECVi3Dn73O6dem5mFZew9nY/JBOP6ROFhMdM7WuVZfpmYSudz9hC6d3Ulh3Y6VIbbPmuWfyhZATZhWaUqsgt8Agj1927wZ28O9AIeHRGWbohF/ZuKsHQvRFgKgtAqvLvlJJNe2MSxrCJXL6VhpKWpghYPDxgyRB07e/aCuZI6t/xnK7f+5TPYvNl+cPXqWsfml1Tim6mE3dlOPQCozsm1F+6As7B85RWYMAFefRX+/GdjSOJJlZfZJzqIdoHettdKWG5cvxufqgqqLBby27VX99ptrwzXbAI1MyCMLH/n1kIF3gGE+Xs16HM3F7pjqSN9LN0Pu2MpVeHuhAhLQRBahS93nuFQeiGP//cAmqZd+AJXkJ2tdqUBuzvZsaPaAcfDQ4nK9PQLTpNbXMGe1DxG7tygDoTbCl/WriUrrwSr1fnzbzqSRXShyqks6xMHgJaXZw+De3iAt7ddWDqu4YsvoLQUgJ0pSljquZUAfdurEHbw6VMAFEV3orK7aidUmGSvDM8/rhzTnMAw+nVtR66vPfRd4BuAn1ft2zW2FDGh4li6PXqD9Lb6fSA0CRGWgiC0CgWlqh/hj0ez+e5gpotXUwc33KDa+uzaZReWMTGqkXMHW0FLA8LhhzOUILwq+Ud14E9/guBgyM3l3gdf4/1tp5zG/7g3hbBSle9Y1lsJS0t+nnN+pckEw4ap91FRsGKFWlthIXzzDQC7UvIAjK0ROXCAuDDlNHY5pxxJa2wsfn16AuB16iTF5SrcnX9c5Y96dOxAvw5B9jxLoNQ/EFPN7R9bGL14R0eKd9wPvd2QWRxLt0KEpSAIrUK+TVgC/O2bJMqr2tgvE6sVtm9XuZWvvGIIyMKIKBb+dz8V7W3CsgEFPIczCokuyGbYGdXex3rzb2H8eABGn9iphPWePfDss1jLKziUeAiAaj9/PHuoFkFehQWGsCzx8uXj7SnQt68qqDlyBKZMgdtvVzf84AMqqqzsPaOamQ/pHALffQf9+3PJEw8B0NUmLAP69cGvt3IsO+VnsPe0uqbitCosCo3tQs/IADID7A3RywNat3AH7PuF64hj6X4YfSyt4li6EyIsBUFocTRNM4Slv5eFUzklLP3ppGsXVZP0dKhQzbj56CM4oHap2VkdwLtbTnHcxya0GuhYTj78EwA7OsaRZA4kZ9RYAEaf2EXFzl1oo0fDQw+R/eIrRn6lOaYTQe0jAVT/SJuwPF3twR+/3MfpcyXQo4dyMAGmTVPPK1eSfDCFiiorIX6edIvwN/I5vb74jH6WUrrZhKVXn97QXfWpjMlLZ3dqHgCemaqavFPf7vSMCrQX8ACVQcEN/zk2E4E+noT4KZfSx9OMp0V+Xbkbxl7hWhv7I1O4KOT/VEEQWpySimqqbK7Ew5P6APDS90cpKq+q77LWxbGFT3ExvP8+AKn+SlCe8AlR5xrkWBYx5ZAKg3/T53I2Hs7im/Zq55xLzh7ixWV/wlSgQt+WDz8w8itNnToR2lEJS//SIrCNKfZU7t2247lO96nu1x/694eKCvI/+FjNHxOiwtaJiWpQVRV3Hv+JLnpFeM+eDsIyg20ncjiTV0povmpv1GNQT+VY+tsdS2tQyAU/c0ug7xkuYXA3xSyOpTsiwlIQhBZHdys9zCbuuLQLnUJ9KSqvYvuJHBevzIGavSFt7uVJXyWwDlpsrt0FHEtN0zhxJpfBZ1UR0Nqel7LxcBYfZZo5FtYJD81KVFEuJV17gNlM+J6fGZG6X13cqRPtOql2Q2Y0KlPVvYq8lLDcctz+8/pkRwq9H/uWpHHXABC5/AsAhnQOVdXsO3caY69OXEXXfFvBT2wsdFPh9vZFOWw5cIYJf/uW4PJiAAK7daZdoDcFIfaG6FpoSL2fuaXQWw5JGNw9MXlIjqU74lJhuWjRIoYPH05gYCCRkZFcf/31JCcnO40pKytjzpw5hIeHExAQwNSpU8nIyHAak5KSwpQpU/Dz8yMyMpKHHnqIqipnJ2TDhg0MGTIEb29vYmNjWbp0aUt/PEEQbBSUKWEZ7OuJxWziip7tAPjxSBsUltddBz4+xuGj3iEAHPZSzxcSltlFFYScOYWHZqU6MIgzQe34+WQuB9MK2NR9KABnAyP4dNHbMG4cAFP3f68u7tSJoNAAyjxUwU3JMbWmYm8lsLY6CMulm09RZdV42FMV+8QeTCS6IJshXULh+HHIzwcvL/D2JuDIIXwqy1V1edeuEB6OZgunx5XnEF6kqsmrPL0gRDmenh3bG/eyhDq3H2otxLF0b4x2Q1IV7la4VFhu3LiROXPmsHXrVtauXUtlZSUTJ06kuLjYGDN//nyWL1/OZ599xsaNGzl79iw33nijcb66upopU6ZQUVHB5s2bWbZsGUuXLmXhwoXGmBMnTjBlyhTGjh3L7t27mTdvHvfeey+r6+gpJwhC85JfYheWAJfHKjfsp6PZLlvTeejCcvBguPVW43Cyh3Iq0wNtDl5qKhVVVh74aBcvfHeYmhzJKCQ2R4XLLX370L1dAHqkb9e0/8euO2Zz221/Z3u1P9bbVPGNd7WtsKlTJ0wmE8W+AQBUnVDV48U2x/L0uVJSc0tIzS3hYJoKk+/3COVk70GYNY1xx3cwsFOwPQw+cKCqdNfp2lWJS5MJky0c/uXkDnx2bVcAzB06qOpzwL9rjHGZR4Q9LN6a6JXh+n83gnth8lBOtDiW7oVLheWqVau466676NevH4MGDWLp0qWkpKSQaPtSzM/P56233uL5559n3LhxDB06lHfeeYfNmzezdetWANasWUNSUhLvv/8+gwcPZvLkyTz11FMsWbKEClso67XXXqNbt24899xz9O3bl7lz53LTTTexePFil312Qfg1oYfCg2wCIb5HOCYTJGcUkllY5sql2dGFZdeucP/9AGh+fmT4KmF5VheWZ8/yY3I6y/ec5cV1RziTV+o0zeGMQnrk2lzNvn0Z3audcW70FQMoefxJToV24MDZAo6Pnmi4kwB06gRAqb+tCvu0EqhFXr5YzErwbTuRy+oDKqytNy3/LHIgAFPO7lHuni4shw6FmTPt8/fsaX9tC4ebTpygfWkeAOYOdpcyLLaL8dpL78PZykwZ0J5rBnVg1hXdXHJ/oWURx9I9aVM5lvn5qu1FWJj66zgxMZHKykomTJhgjOnTpw+dO3dmy5YtAGzZsoUBAwYQFRVljElISKCgoIADtqrOLVu2OM2hj9HnqEl5eTkFBQXGo1DvJScIQpPQhaXuPIX5e9GvgxJPm482Phx++lwJn/2cel6j8YvilK23ZNeuMHIkLF3KmVffxmorMMj2D6HaYoHqanZtV+2BNA2+2ukcGj+cWWQ4lvTpw296K2HpZTEzsV8UcbaG5adySvgps5LvYkfaL7YJy8pANcbrrJq72MuXCX1VUc/W4zmsOaDSgR4YF8vIbmFs6qZ2Bhp6dJfKDdXzK4cOVW2ObPMSG2u/l82x5Phx1b4IVCN4G+372MWcb5Q937I1CfX34qXbLjFSJwQ3Q3Is3ZI2IyytVivz5s1j1KhR9O/fH4D09HS8vLwICQlxGhsVFUW6beeJ9PR0J1Gpn9fP1TemoKCA0lJntwFU7mdwcLDxiIuLa5bPKAi/VmoKS4BRtnD4j7Zw+PpDmSQs3uSUR1gXcz7YyUOf7+W7gxm1nq+stvLzyVyqGyo8rVZnYQkwYwYn48fbh5gtZAUq567Pa8/jW6Gc1s8TTzvtJHQko5AeOXbH8orYCO4e1Y0nr+tHkI8nof5edAhWOZyf7Ejl67gx9nXYBGB1cAgAflnq85V4+3HzUBWaXn8okx2nVHX4xH7RPDYljv3RPcjxDcKntBi2bHF2LC0W1aDdYoFrr7XfSxeWH39s3xZy8GDjdI/u0WQEhFFtMuPXtXPDfo6C0AjMUhXulrQZYTlnzhz279/Pxx9/7Oql8Oijj5Kfn288kpKSLnyRIAh1UlCLsHTMszyTV8rvP95FckYhH25LqXeuwxmF7LE19T6SWfu+48s2n+Sm17bw1IoG/r+bkQHl5Up86e4ekFWkxKPurv5nyHUATNn+Dd8ufZCRWUc5mVNC4ilV/KJpGkfSCpxC4R4WMwuviePWEXZxFtdBhdeT0grY2H0IOYOHQ0KCfetH2x/TFpuT4xUcRHyPcCxmEznFFWgaDOgYTMcQXwZ0CubBCb3Z1/9Sde1rr8G5c+DpqVoRAcyerT6fY+TGFgrn7FklrKdPh//7P+N0dJAPf5jxNDNveZLIniIshebHqAqXPpZuRZsQlnPnzmXFihWsX7+eTg5f6tHR0VRUVJCXl+c0PiMjg+joaGNMzSpx/f2FxgQFBeHr67y7A4C3tzdBQUHGI1BvRiwIQpOw51ja28YM7xqGl4eZtPwy7np7OwVlqpPDrtRzxhhrtZXv5y5k82v2Pzi/3HnGeJ2SU1Lr/X4+qeZ4b+spjmXVLj6d0PMrO3VSxS02sgtVnnZsZAAdQ3x5e/h1vPLoK5wNjKDrubO888lCgsqK+MIWDs8qLMc/Mw2/ynI0T0+7K1iDuA72nWwqLZ7krfkeVq0yCmc8wp2LZfzDQ/D39lCFOTYS+tmjMPOv7MWYB+5Qbz79VD0PGKCqwnUsNfb67tdP3c/DA156CZYuBYfvQ5PJxF8X3sGDi2YTHuBd+89NEC4Co0G6OJZuhUuFpaZpzJ07l6+++orvv/+ebt2cE7SHDh2Kp6cn69atM44lJyeTkpJCfHw8APHx8ezbt4/MTPvew2vXriUoKMgIYcfHxzvNoY/R5xAEoWWpLRTu42lhWBfVxuZIZhH+XhZMJkjNLSW7qByAvWs2M27JU/R4aA4HzuZTbdX4apc9pzH1XO3C8kimyouutmo8s+qQ80mrFZKTYflyyMpSx3Rh2aWL09As2zraBXgbYvBZrTOT7n6ZnK498Ssu4HdbPmXFnjRKK6o5nFFED1t+palnTyeR6kg/B2EZ4O1Bt3B/p/NeNaqwgyLV+/ju9iKaSf2jnSedONH++UCFweujSxe17ePOnTB3riFqHYmNDGBYV9dUhAvuj8mi/v+waFandBLhl41LheWcOXN4//33+fDDDwkMDCQ9PZ309HQj7zE4OJh77rmHBQsWsH79ehITE5k5cybx8fFceqkK+0ycOJG4uDjuvPNO9uzZw+rVq3nssceYM2cO3t7qr+z777+f48eP8/DDD3Po0CFeeeUVPv30U+bPn++yzy4IvyZqE5Zgz7MEeOK6/vSMVG12dqfkAXAyUYWyo4pyefqDLfxwJIuMgnLjmtqEZUWVlZMOTubqAxn8fDJXVdpMnw7BwdCnj8o3vMPm8jlWhDuQXajuFRHobYhBqwYFPgHk//VvAMzcuZzArLO8u+Uke8/kEeuQX1kXjsKyX4cgzGZnUecX6VyFHRat3v/GVmHeOyqQ2MgakZSoKBgyxP7+QsISVB/NAQMuPE4QWgDdsTRrVsS0dB9cKixfffVV8vPzGTNmDO3btzcen3zyiTFm8eLFXH311UydOpXRo0cTHR3Nl19+aZy3WCysWLECi8VCfHw8d9xxB9OnT+fJJ580xnTr1o1vvvmGtWvXMmjQIJ577jnefPNNEhISWvXzCsKvFT3MXVNYXjOwA2H+Xtw+sjNTh3RkcEwIYA+HZySfMMYW7T/IQ5/vBWBCXxUGPptXRlW11WnOkznFVFs1Ar09uHW4Knj5+8qDaElJ8N57UFRkb4C+bh0bth5m9TeqfVlVjHMuoaNj2a+DPQwd6udJlztugt/8Bu+qShb88AGLvj3EM6uS6ZFrrwivi44hvsbPwjG8reMf7VwFHdlBCfCR3cN5+65hvDF9WO0TT5pkf90QYSkIrsSWY2nStIYX2gltHpeHwmt73HXXXcYYHx8flixZQm5uLsXFxXz55ZdG7qROly5dWLlyJSUlJWRlZfHss8/iUSMENWbMGHbt2kV5eTnHjh1zuocgCC1LzT6WOp3D/Uh8bAJ/v2EAJpOJSzqr0PiulDzKKqspTbGHvbvlniHL5iDOHtMdLw8z1VaNtHznPphHMlROZWxUAAuu7IWvp4WdKXkkr7DtbhMfD4WFylGsrubbF97Dx9bWZ9HBUj5PtN8zqxbHEuDynu2wWMzwzDMA3Hjgey4vPavu2wDH0mQyMbyr+qwjup3fI9KzRo5ldCe70BzXJ4rO4X61T6z/sexYuCMIbRSzrY+lRbNilVC429AmincEQXBv6gqFgxJZOpd0DgFgT2oeO0+dI7ww1zh3abV63T3CnyGdQ+kUqgpNUnOdw+GHM1R+Zc/IACKDfLhpqCoIzNmw2TbRpSr38aqrABh2YCtdClWuZZJXOP/32R6SzqpdbbIdHMv2wT6E+qn16yFpRoyAm2/GrGm8d/Rrvv/DbxhSYtuTux5hCfD0DQN4Y/owoz+lEzW2UAyPbmAfycsvh0cfhSVLnLalFIS2iLHzjmZFdKX7IMJSEIQWpz5h6UjPyED8vSwUV1Tz/rZTRBbZheUU3yJGdAvjkcl9MJlMxj7SKTWE5VFbC6KethzEybYil8D9e9SA4cMBKBx7JQBjTvxMTKEq/ovop3am2X4ih2qrRm6xqgqPCPTCZDKx4MpeTIyLMuYE4O9/Bw8PTCtX0v2HNXjm2Lap7N273s8aFeTDlXFRTsLaoEbvXnNw0PljasNsVuuZNath4wXBhZidcixFWboLIiwFQWhRyiqrqahSeZAXEpYWs4lBtjzLb/enE1lkbz0UlHqST/9fPAn9lKiLCbM5ljUKePSK8NgoVQg0olsYEd4meqUfUwOGqfzEJVXRFHn50q44D0t5OZjN9BqqXMbdqXnkFJdj1cBsgnB/VQh4Z3xX/jN9GP7eDqk2sbHw//6fen3ffeq5c2fwd670bhQ1HEsCApo+lyC0Uex9LLU2IyxfffVVBg4caLQbjI+P59tvvzXOl5WVMWfOHMLDwwkICGDq1KnntTNMSUlhypQp+Pn5ERkZyUMPPURVVZXTmA0bNjBkyBC8vb2JjY1l6dKl561lyZIldO3aFR8fH0aOHMn27dudzjdkLa5AhKUgCC2K7laaTeDvVXv7HUf0cLimQbtiu7Dk8GEc42W6Y5maa989q7LayonsYgCjwtzDYub2gCJ8qioo9QuEHj04k1fK2zvO8lOXQfb5O3ZkYA8Vlt5zOt/oYRnm72Xs010nf/mLEn85tl2D6incaRAOjmW1hyd4Sx9Jwf0w2XbesVitRpcsV9OpUyf+8Y9/kJiYyM8//8y4ceO47rrrjC2i58+fz/Lly/nss8/YuHEjZ8+e5cYbbzSur66uZsqUKVRUVLB582aWLVvG0qVLWbhwoTHmxIkTTJkyhbFjx7J7927mzZvHvffey+rVq40xn3zyCQsWLODxxx9n586dDBo0iISEBKfWihdai6sQYSkIQoviWLhTs61ObVwSY3PrNI1IR2GZnw/Z2cbbzmE2YengWJ7KKaGyWsPfy0LHEHuz78llqqBmf3QsVRos+GS3aks04jf2+bt2ZZCtQvtEdjFHbY3VIxrSHDwqymnXmgvlV16QIHvo2xwkGzQI7onJojuWLR8KLywspKCgwHiUl5fXOu6aa67hqquuomfPnvTq1Yunn36agIAAtm7dSn5+Pm+99RbPP/8848aNY+jQobzzzjts3ryZrVtVZ4k1a9aQlJTE+++/z+DBg5k8eTJPPfUUS5YsoaJC/bH62muv0a1bN5577jn69u3L3Llzuemmm1i8eLGxjueff55Zs2Yxc+ZM4uLieO211/Dz8+Ptt98GaNBaXIUIS0EQWpSG5lfqDLY5lsFlRXhVq2uJsu0yc+SIMS5GF5YOOZZH9TB4ZIBT7mLP1GQAfm7Xg99/vJttJ3Lx87Jw1cN322/ctSshfl50tVVcr7PtQ94usIFu4R/+YF+nbXOGJmOxqH6bgEl2/hLcFLNH6wnLuLg4goODjceiRYsueE11dTUff/wxxcXFxMfHk5iYSGVlJRMctkbt06cPnTt3ZsuWLQBs2bKFAQMGEBVl3xkrISGBgoICw/XcsmWL0xz6GH2OiooKEhMTncaYzWYmTJhgjGnIWlzFheNSgiAIF0F+SeOEZUSAN53D/PBOPqUOhIWp7QczMpSwvOwywB4Kzy6qoKSiCj8vD3uroRrNwz12JgKwNzqWb/elAfDXa/sRMzAGBg6EvXuN5uiDY0I4mVPC+kMq5NSuodsZBgSo7RQ/+ABuv71h19RHSIhyaUVYCm6K3bHUWrxBelJSEh07djTee9eTXrJv3z7i4+MpKysjICCAr776iri4OHbv3o2XlxchNYrroqKiSE9X3SDS09OdRKV+Xj9X35iCggJKS0s5d+4c1dXVtY45dOiQMceF1uIqxLEUBKFFKShrnLAEmH9lTxLCbL9p2reHnqpa29GxDPbzJMhH/W18+pzKszyiV4RHORS7lJcr4Qjsa6/mmTKgPTfb2hDx8MPQsSPYcpP04iG9qXtEQx1LgNGj4fXXm6fYRi/gEWEpuCsOofCW3tIxMDDQKMgJCgqqV1j27t2b3bt3s23bNmbPns2MGTNISkpq0fW5EyIsBUFoUepqjl4fN1zSif8baMszjI62C8vDh53G6eHwFNsWjo49LA327YPKSrTwcKpjOtMtwt9oyA7AtGlw+jQMHgzYhaVOgx3L5kZ3IkRYCu6KuW1u6ejl5UVsbCxDhw5l0aJFDBo0iBdffJHo6GgqKirIy8tzGp+RkWFs3BIdHX1eZbb+/kJjgoKC8PX1JSIiAovFUusYxzkutBZXIcJSEIQWpbE5lgZ6OKd9e+jVS712cCzBoTL8XAlV1VaO2yrCe0U5iLEdOwAwDR/OhofH8u3vryDYr+61xLUPwtNiz8+MCPRq3LqbCxGWgrtjsVeFV7eRdkO1YbVaKS8vZ+jQoXh6erJu3TrjXHJyMikpKcTHxwMQHx/Pvn37nKq3165dS1BQEHG23Ov4+HinOfQx+hxeXl4MHTrUaYzVamXdunXGmIasxVVIjqUgCC1Kk4VlmsqFPC8Urmlw/DisW0eX9irfMjW3lNRzpVRUWfHxNDtVhPPzz+p52DC8bcUC9eHjaaFv+yD2ns4HoF2Ai3awkVC44O4YjqWGtY1Ylo8++iiTJ0+mc+fOFBYW8uGHH7JhwwZWr15NcHAw99xzDwsWLCAsLIygoCAeeOAB4uPjufTSSwGYOHEicXFx3HnnnTzzzDOkp6fz2GOPMWfOHCP8fv/99/Pyyy/z8MMPc/fdd/P999/z6aef8s033xjrWLBgATNmzGDYsGGMGDGCF154geLiYmbOnAnQoLW4ChGWgiC0KEYo3OcihGX37uqXUHExJCfDxImQmsplf3uZ1+lKSm4J/9l0HIDeUYHObY1++EE923bcaQiDOoUYwtJljqUIS8HdccixrGobupLMzEymT59OWloawcHBDBw4kNWrV3PllWqnrsWLF2M2m5k6dSrl5eUkJCTwyiuvGNdbLBZWrFjB7NmziY+Px9/fnxkzZvDkk08aY7p168Y333zD/PnzefHFF+nUqRNvvvkmCQkJxphbbrmFrKwsFi5cSHp6OoMHD2bVqlVOBT0XWourEGEpCEKLUtAcjqWXl6raPn4cbrkFUlMB6HbsAER2ZdPhLCqqrZhMMO/KXvY5jh1TLqeHB4wZ0+BbD44J4b2tqirdZTmW06bBzp3qWRDcEXPb29Lxrbfeqve8j48PS5YsYcmSJXWO6dKlCytXrqx3njFjxrBr1656x8ydO5e5c+de1FpcgeRYCoLQojRLKBzs4XBbhTdAxFFVqVlRrbbteDihD2N7R9rnWLVKPY8a5dR0/ELovTS9PMyE+rnIsRwyBNavh5EjXXN/QWhp9BzLNiQshYtHhKUgCC1KswtLUKFwwHf/XkyoX0jXDOrA/b/p7jyHLiwnT27UrXu0C+BPV/Xh7zcMaNBuQYIgNAGbY2lqQ3uFCxePhMIFQWhRCkpVP8hGCcviYihUrYPQW2cMHKie+/aFzz6DiAhM587xp35+HPAMZdGNA51226GsDL7/Xr2eNKnR675vdI9GXyMIQiNwqApvI7U7QjMgwlIQhBalSY6l3mrIz89evDJ9OlRUwHXXqbB2v36wezezAgvg+nHnz/Hjj1BSohxPXZQKgtB2aIM5lsLFI6FwQRBajIoqK6WV1UAjhaVjGFx3Ib29Yc4c6GTbMeeSS9RzXQnw336rnidNss8hCELbwaEq3Gp18VqEZkOEpSAILYbuVppMEOjTiABJzfzK2rDtlHNBYdnI/EpBEFoJxz6W4li6DSIsBUFoMXRhGeDt0bgimIYIy/ocy1On4OBB9YtrwoSG31cQhNbDoSpcdKX7IMJSEIQWo9kqwmtj0CD1fPo0ZGc7n1u9Wj3Hx9sbjQuC0LYwqsLb9paOQuMQYSkIQovRLM3R6yIoCGJj1euaruXWrep5XC1FPYIgtA2cqsJFWLoLIiwFQWgxmuxYnlK73hithuqirnC43kRdz8MUBKHt4ZBjqYmwdBtEWAqC0GI0SVj+97+wYYN6rQvHuqhNWFZVwYED6rW0GRKEtosuLNGwSiNLt0GEpSAILUZOcQUAYf4N3BbxzBm4+271+g9/sOdR1kVtwvLoUdUc3c8Punev/TpBEFyPLRQOYK2qduFChOZEhKUgCC1GdlE5ABEB3hceXF0Nd9wBubkwdCj8/e8XvkYXlocPQ1GReq2Hwfv3NxwRQRDaIA7/f1qrRVi6C0361h03bhx5eXnnHS8oKGCcJMsLgmAjxxCWDXAs335bhcD9/eGjj8CrAddERUHnzqBpsHmzOqYLSwmDC0LbxsGx1ERYug1NEpYbNmygoqLivONlZWX88MMPF70oQRDcg+wi9T3RIMdSF4bz50PPng2/ydix6nn9evUswlIQfhk4OJYiLF1HXl4eb775Jo8++ii5ubkA7Ny5kzNnzjRpvkbtFb5X/8IGkpKSSNf38wWqq6tZtWoVHTt2bNJCBEFwP/RQeHhDhKVeCd67d+NuMnYsLFsmwlIQfmk4OpZVsqejK9i7dy8TJkwgODiYkydPMmvWLMLCwvjyyy9JSUnh3XffbfScjRKWgwcPxmQyYTKZag15+/r68tJLLzV6EYIguCc5hmPZgLC2Liy7dGncTXTH8uefVbN0fZ4BAxo3jyAIrYtjjqVVHEtXsGDBAu666y6eeeYZAgMDjeNXXXUVt99+e5PmbJSwPHHiBJqm0b17d7Zv3067du2Mc15eXkRGRmJx+AtEEIRfL2WV1RSVVwEQEXgBx9JqhdRU9bqxwrJzZ1X9ffw4vPaaOtapE4SFNXLFgiC0Ko6OZWWVCxfy62XHjh28/vrr5x3v2LGjU1S6MTRKWHaxfeFbrWJZC4JQP3oY3MtiJtD7Al81aWlQWal+0XTo0PibjRmjhKX+BSlhcEFo+zjmWIqucAne3t4UFBScd/zw4cNO5mFjaJSwdOTIkSOsX7+ezMzM84TmwoULmzqtIAhuQrZDGNxkMtU/WA9fd+oEHk34Who7VlWV63uGi7AUhLaPo7CUPpYu4dprr+XJJ5/k008/BcBkMpGSksIjjzzC1KlTmzRnk4TlG2+8wezZs4mIiCA6Otrpl4bJZBJhKQiC0WqoQYU7J0+q565dm3YzPc9SR/IrBeEXgdVkxqxZQXIsXcJzzz3HTTfdRGRkJKWlpfzmN78hPT2d+Ph4nn766SbN2SRh+be//Y2nn36aRx55pEk3FQTB/cluTA/Lphbu6HTsqFoUHTmi3otjKQi/CKxmM+ZqqziWLiI4OJi1a9fy008/sWfPHoqKihgyZAgTJkxo8pxNEpbnzp3j5ptvbvJNBUFwfxrVw/JihSUo1/LIEfD0bHzLIkEQXILVpMLhUrvR+lRWVuLr68vu3bsZNWoUo0aNapZ5m9Qg/eabb2bNmjXNsgBBENyTJvWwvBhhqf+FPWiQEpeCILR5rLY8S61KqsJbG09PTzp37kx1Mzenb5JjGRsby1/+8he2bt3KgAED8KzxJf7ggw82y+IEQfjlkt0aPSwdmToVXnwRrrii6XMIgtCqaDbHUqsWx9IV/PnPf+ZPf/oT7733HmHN1KKtScLyP//5DwEBAWzcuJGNGzc6nTOZTCIsBUFw2Cf8Ao6lpjWPsDSbQb57BOEXhe5YIls6uoSXX36Zo0eP0qFDB7p06YK/v7/T+Z07dzZ6ziYJyxMnTjTlMkEQfkVkN1RYZmdDSYl6HRPTwqsSBKEtoemhcMmxdAnXX399s8/Z5D6WgiAI9WGEwgMvEArX3cr27cHHp4VXJQhCW8IIhUuOpUt4/PHHm33OJgnLu+++u97zb7/9dpMWIwiCe1BVbeVciRKW4f4XcCybIwwuCMIvEsOxlBxLl5KYmMjBgwcB6NevH5dcckmT52pyuyFHKisr2b9/P3l5eYwbN67JixEEwT3ILalA08BkgjD/BjqWIiwF4VeHURUuOZYuITMzk1tvvZUNGzYQEhICQF5eHmPHjuXjjz9u0raOTRKWX3311XnHrFYrs2fPpkePHk2ZUhAENyLHFgYP8/PCYm7gdo4iLAXhV4ceCkdyLF3CAw88QGFhIQcOHKBv374AJCUlMWPGDB588EE++uijRs/ZpD6WtU5kNrNgwQIWL17cXFMKgvALpcGFOyDCUhB+xWj6H56SY+kSVq1axSuvvGKISoC4uDiWLFnCt99+26Q5m01YAhw7dowq+Y9DEH715DS0cAfs+4SLsBSEXx1WswWQqnBXYbVaz+tFDqp5elN3Q2pSKHzBggVO7zVNIy0tjW+++YYZM2Y0aSGCILgPxq47FyrcAXEsBeFXjFEVLsLSJYwbN47f//73fPTRR3To0AGAM2fOMH/+fMaPH9+kOZskLHft2uX03mw2065dO5577rkLVowLguD+ZDU0FF5QAHl56rUIS0H41WFUhVdJ8Y4rePnll7n22mvp2rUrMbY+wqmpqfTv35/333+/SXM2SViuX7++STeryaZNm/jXv/5FYmIiaWlpfPXVV07NOu+66y6WLVvmdE1CQgKrVq0y3ufm5vLAAw+wfPlyzGYzU6dO5cUXXyQgIMAYs3fvXubMmcOOHTto164dDzzwAA8//HCzfAZBEM5HD4WHX2g7R92tDAuDwMAWXpUgCG0NXVhK8Y5riImJYefOnXz33XccOnQIgL59+zJhwoQmz3lRDdKzsrJITk4GoHfv3o0uSy8uLmbQoEHcfffd3HjjjbWOmTRpEu+8847x3tvb2QGZNm0aaWlprF27lsrKSmbOnMl9993Hhx9+CEBBQQETJ05kwoQJvPbaa+zbt4+7776bkJAQ7rvvvkatVxCEhqGHwtvV51iWlsJTT6nX3bq1wqoEQWhrGMKyWuozXIXJZOLKK6/kyiuvbJb5miQsi4uLeeCBB3j33XeN5E6LxcL06dN56aWX8PPza9A8kydPZvLkyfWO8fb2Jjo6utZzBw8eZNWqVezYsYNhw4YB8NJLL3HVVVfx7LPP0qFDBz744AMqKip4++238fLyol+/fuzevZvnn39ehKUgtBAXLN5JT4frr4dt28DDA/70p9ZbnCAIbQYjx7Jac/FKfp08+OCDxMbG8uCDDzod1/cQf+GFFxo9Z5OqwhcsWMDGjRtZvnw5eXl55OXl8d///peNGzfyhz/8oSlT1smGDRuIjIykd+/ezJ49m5ycHOPcli1bCAkJMUQlwIQJEzCbzWzbts0YM3r0aLy87L/gEhISSE5OPq/Ru055eTkFBQXGo7CwsFk/kyC4O/UW76xaBUOHKlEZGgpr1kAdEQtBENwbzVYVLo6la/jiiy8YNWrUeccvu+wyPv/88ybN2SRh+cUXX/DWW28xefJkgoKCCAoK4qqrruKNN95o8kJqY9KkSbz77rusW7eOf/7zn2zcuJHJkydTbevQn56eTmRkpNM1Hh4ehIWFkZ6eboyJiopyGqO/18fUZNGiRQQHBxuPuLi4ZvtMguDuaJrm4Fg6CMvCQrjvPpg8Gc6ehT59lLgcO9ZFKxUEwdXofSxlS0fXkJOTQ3Bw8HnHg4KCyM7ObtKcTRKWJSUl54k1gMjISEpKSpq0kNq49dZbufbaaxkwYADXX389K1asYMeOHWzYsKHZ7lEbjz76KPn5+cYjKSmpRe8nCO5EQVkVFbZfEuGO2zk+8AC88Yba53HePEhMhJ49XbNIQRDaBIZjaZWqcFcQGxvrVBCt8+2339K9e/cmzdmkHMv4+Hgef/xx3n33XXx8fAAoLS3liSeeID4+vkkLaQjdu3cnIiKCo0ePMn78eKKjo8nMzHQaU1VVRW5urpGXGR0dTUZGhtMY/X1duZve3t5ORUIFBQXN+TEEwS1ZsfcsK/akcTKnGIBAbw98PC32Ad9/r54//BBuvdUFKxQEoa1h7LwjjqVLWLBgAXPnziUrK4tx48YBsG7dOp599llefPHFJs3ZJGH5wgsvMGnSJDp16sSgQYMA2LNnD97e3qxZs6ZJC2kIp0+fJicnh/bt2wNK4Obl5ZGYmMjQoUMB+P7777FarYwcOdIY8+c//5nKykqju/zatWvp3bs3oaGhLbZWQfg1UVpRzYJP91BRZf/lMKJbmH1AQQGkpqrXEye28uoEQWiriGPpWu6++27Ky8t5+umnecrWpaNbt2689tprTJ8+vUlzNklYDhgwgCNHjvDBBx8YfY9uu+02pk2bhq+vb4PnKSoq4ujRo8b7EydOsHv3bsLCwggLC+OJJ55g6tSpREdHc+zYMR5++GFiY2NJSEgAVK+lSZMmMWvWLF577TUqKyuZO3cut956q9FB/vbbb+eJJ57gnnvu4ZFHHmH//v28+OKLsqe5IDQju1LOUVFlpV2gN4tuGEBMmB+xkfZesti+J4iOVj0rBUEQAIx2Q+JYuoLS0lJmzJjB7NmzycrKIiMjg7Vr19aa7thQmiQsFy1aRFRUFLNmzXI6/vbbb5OVlcUjjzzSoHl+/vlnxjok7utbRc6YMYNXX32VvXv3smzZMvLy8ujQoQMTJ07kqaeecgpTf/DBB8ydO5fx48cbDdL//e9/G+eDg4NZs2YNc+bMYejQoURERLBw4UJpNSQIzci2E7kAXNYjnAlxtXwhHTyonqUQThAEB8SxdC3XXXcdN954I/fffz+enp5MmDABT09PsrOzef7555k9e3aj52ySsHz99deNBuSO9OvXj1tvvbXBwnLMmDFoWt29q1avXn3BOcLCwmpdiyMDBw7khx9+aNCaBEGon6pqK58lnmZcn0iiglSO9bYTqg3YyG7htV+kF8CJsBQEwQGpCnctO3fuNCK4n3/+OVFRUezatYsvvviChQsXNklYNqkqPD093chzdKRdu3akpaU1ZUpBEH4hvLf1FI9+uY/5n+wGoLyqml0peUCNvEpHdGHZt2/LL1AQhF8M4li6lpKSEgJt2+muWbOGG2+8EbPZzKWXXsopfcvdRtIkYRkTE8NPP/103vGffvrJyG0UBME9WblP/fG4+VgORzOL2JOaT3mVlYgAb3q086/9InEsBUGoDcmxdCmxsbF8/fXXpKamsnr1aibaiiszMzMJCgpq0pxNCoXPmjWLefPmUVlZ6VSe/vDDDzf7zjuCILQdsgrL+fmUfceqj7anEOKrui2M7BaGyWQ6/6KSEjhxQr0WYSkIggOaRd95RxxLV7Bw4UJuv/125s+fz/jx442WkWvWrOGSSy5p0pxNEpYPPfQQOTk5/O53v6OiQu2w4ePjwyOPPMKjjz7apIUIgtD2WZuUgaZBgLcHReVVfJ54ml5Rqvp7ZPc6wuDJyaBpEB4O7dq14moFQWjz2PYKxyqOpSu46aabuPzyy0lLSzPaRwKMHz+eG264oUlzNklYmkwm/vnPf/KXv/yFgwcP4uvrS8+ePZ2qtQVBcD9WH1DboP6/0d35eEcqZ/JK2XFSOZgXzK+Mi1O77giCINjQLLqwFMfSVURHR5+3YcyIESOaPF+ThKVOQEAAw4cPv5gpBEFowyxaeZDvDmbwyrShtA/xYfMxtXfs5AHtMZtN/Gt1MgAhfp70igysfRJpNSQIQh1o4li6HU0q3hEEoeWpqrby391nyC+tdNkavtx1hmNZxdzx1jaW/XSSymqNHu38iY0M4OZhnfCwtQoZ0TUMs7kON1IKdwRBqAtb8Y5JcizdBhGWgtBG+Xr3WX7/8W5+//Eul9y/2qqRU1QOqKKd59YeBiChnwqZRAb6MGWgajs2pndk3RNJqyFBEOpCL94Rx9JtuKhQuCAILceRzEIANiRnkZxeSO/oOkLNLUROUTlWDcwm6Bzmx8mcEsAuLAEW3TiA6wZ3YEyvOoRleTno27aKYykIQg00o92QOJbugjiWgtBGSc8vM16/8cPxVr9/ZqFyKyMCvHn/3pH0aOfPiG5hDOwUbIzx8/JgXJ+ousPgR46oXxhBQSA9bgVBqIlZHEt3QxxLQWijpOXZheV/d5/h/yb2JjrYp9Xun1Gg7h8Z5E2nUD++W/Cb2vtU1odUhAuCUB/6H6UiLN0GcSwFoY2SVlAKqIrrymqNpZtPtur9dccyMlCJ2UaLSoBDh9Sz5FcKglALeoN0kwhLt0GEpSC0QaxWjYx8Jezmje8JwAfbTlFUXtVqa8gs0IXlRfSnTVbtiOjduxlWJAiC22GWdkPuhghLQWiD5JZUUFFtxWSC20Z2pnuEP4VlVSz96USrrSGzUA+FX0T4/bCqJKdXr2ZYkSAIbofuWErxjtsgwlIQWpnMgjIKyurvTakX7kQEeOPtYeFBm2v5yoZjTkU9LUnGxTqWmmZ3LEVYCoJQG+JYuh0iLAWhFUnLL2XssxuY8NxGUmzte2ofp8Rje1uxznWDOzCkcwglFdX849uDrbLWLN2xbKqwzMiAwkJVtBMb24wrEwTBXdDEsXQ7RFgKQivyyY5UiiuqySwsZ/rb28iyFcjUJD1fFe5EB9kLZ564tj8mk2qc/vPJ3BZfq1G809RQuO5Wdu0K3heRpykIgvuib+moiWPpLoiwFIRWotqq8emOVAB8PS2czCnhrne21xoW1x3LDiG+xrEBnYK5ZVgMAH9dfgBN01psrVarZojeJjuWen6lFO4IglAXHrpj2TaE5aJFixg+fDiBgYFERkZy/fXXk6z/kWyjrKyMOXPmEB4eTkBAAFOnTiUjI8NpTEpKClOmTMHPz4/IyEgeeughqqqciy83bNjAkCFD8Pb2JjY2lqVLl563niVLltC1a1d8fHwYOXIk27dvb/RaWhsRloLQSmw6ksXZ/DJC/Dz5es4owv29OHC2gLH/2sAbm45TWmEPBel5lDX7Vv5fQm98PS3sP1PAwbTCFltrbkkFVVYlXNs1VVhKfqUgCBfAZG5bjuXGjRuZM2cOW7duZe3atVRWVjJx4kSKi4uNMfPnz2f58uV89tlnbNy4kbNnz3LjjTca56urq5kyZQoVFRVs3ryZZcuWsXTpUhYuXGiMOXHiBFOmTGHs2LHs3r2befPmce+997J69WpjzCeffMKCBQt4/PHH2blzJ4MGDSIhIYHMzMwGr8UlaMIFSU1N1QAtNTXV1UsRfsHMWrZD6/LICu2J/x3QNE3T9p3O037zzPdal0dWaF0eWaGNfPo7LTW3WNM0Tbv19S1al0dWaF/vOn3ePNPe2Kp1eWSF9u7mE826vlX707Stx7I1TdO0A2fy1Vqvm69pt9+uaQUFjZ/wmms0DTRtyZJmXacgCO7DodtnaRpo30y+s0Xm139/JyUlafn5+cajrKysQddnZmZqgLZx40ZN0zQtLy9P8/T01D777DNjzMGDBzVA27Jli6ZpmrZy5UrNbDZr6enpxphXX31VCwoK0srLyzVN07SHH35Y69evn9O9brnlFi0hIcF4P2LECG3OnDnG++rqaq1Dhw7aokWLGrwWVyCOpSC0ApkFZaw7pP7KvG2ECmf37xjMdwt+wzM3DaRDsA/pBWV8tD0FgHTbrjddzqVBRYXTXMO6hgKw4+S5Zltfcnoh/++9RO5euoOyymqj1dDvNrwHH34Izz/fhEnFsRQE4QJYlAwxtbBjGRcXR3BwsPFYtGhRg67Lz88HICwsDIDExEQqKyuZMGGCMaZPnz507tyZLVu2ALBlyxYGDBhAVFSUMSYhIYGCggIOHDhgjHGcQx+jz1FRUUFiYqLTGLPZzIQJE4wxDVmLKxBhKQitwGeJp6m2agzrEkrPqEDjuIfFzG+HxfCHiSoP8ccj2WiaRlp+KRMPb2HwuOEqR3HpUrDl5wzvavuCO9V8wnL5nrMAFFdUsyslj8zCcnwqy4jIz1YDnn8echtRMFRZCcdt+5tLjqUgCHWh7xXewlXhSUlJ5OfnG49HH330gtdYrVbmzZvHqFGj6N+/PwDp6el4eXkREhLiNDYqKor09HRjjKOo1M/r5+obU1BQQGlpKdnZ2VRXV9c6xnGOC63FFYiwFIQWxmrV+HiHciJvHdG51jGX94wAYO+ZfE7llFBWaWVE6n518uRJmDkTBgyAkycZHBOCxWziTF4pZ/JKG7SGvJIK7l22g//ZBKQjmqaxfK/9+JbjOWQWlBGT55AAXlAAzz3XoHsBcOKEEsK+vtCxY8OvEwTh14Utx9JkbbliRIDAwECCgoKMh3cDOlXMmTOH/fv38/HHH7fo2twNEZaC0FSSkmDZMtUIvB42H8shNbeUQB8PpgxoX+uYqCAfekUFoGnwxc7TAMTlnVEnJ02C0FC17/ann+Lv7UFc+yCABrcd+nLnGb47mMk/Vh48r5p8/5kCTuWUMD1xOVMO/sDWYzlkFpbTJc/2F6+PrYDoxRchK6v2G1itcO21MGyY6l3puOOOWb5mBEGoA70q3Nq2+ljOnTuXFStWsH79ejp16mQcj46OpqKigry8PKfxGRkZREdHG2NqVmbr7y80JigoCF9fXyIiIrBYLLWOcZzjQmtxBfKNLwhNZfZsuOsuePdd+zFNg/XrITvbOKTnTd5wSUd8vSx1TndFz3YAfJ6ohGVstrqOxx6DBx9Ur48eBex5lj83MM9y6/EcAM7ml5Gc4VxNvmLvWXpnneTJ717n+W+e48CJDFJyS+isC8urr4ahQ6G4GP75z9pv8PnnsHw5JCbCyy9LfqUgCA3C1MZ23tE0jblz5/LVV1/x/fff061bN6fzQ4cOxdPTk3Xr1hnHkpOTSUlJIT4+HoD4+Hj27dvnVL29du1agoKCiIuLM8Y4zqGP0efw8vJi6NChTmOsVivr1q0zxjRkLa5AhKUgNBU9h/Df/7a7lm+8AePGKdEJ5BSVsyZJCbRbh9ceBtfRw+Fp+WX4l5cQmWf7Uurb175zjU1Y6nmWPzcgz9Jq1dju4GyuP2R3HTVNY8XeNEam7APAu7qK7ukn2Xw0h855aWpQjx7w1FPq9QsvwFtvOd+gqgoc2mjw7LNKYILkVwqCUD82YWluI8Jyzpw5vP/++3z44YcEBgaSnp5Oeno6paUq7Sg4OJh77rmHBQsWsH79ehITE5k5cybx8fFceumlAEycOJG4uDjuvPNO9uzZw+rVq3nssceYM2eOEYK///77OX78OA8//DCHDh3ilVde4dNPP2X+/PnGWhYsWMAbb7zBsmXLOHjwILNnz6a4uJiZM2c2eC2uQISlIDQFTQP9r9GdO2HrVigupuovSmCVrluPZrXyxc7TVFZrDIoJIa5DUL1TjuwWhpetQrJHrnItiY6GsLDzhOWwLsqxPJRecMF9x5MzCskrsY9Zn2z/K3pnSh5n8koZdeaAcaxfxjEqqq12x7JHDxWOv+celWB/773w5JN2Mf3++8qh1NeZmwt6TpI4loIg1IdtS8e24li++uqr5OfnM2bMGNq3b288PvnkE2PM4sWLufrqq5k6dSqjR48mOjqaL7/80jhvsVhYsWIFFouF+Ph47rjjDqZPn86TTz5pjOnWrRvffPMNa9euZdCgQTz33HO8+eabJCQkGGNuueUWnn32WRYuXMjgwYPZvXs3q1atciroudBaXIGHS+8uCL9U8vOd2wC99BJVffvikanyYXzP5fD4G+v4odATgNuGx1xwSj8vD4Z0CWHr8Vx6ZqsdeujbVz336KGeU1OhtJTIIF+6hPtxKqeEnafOMaZ3ZJ3zbrOFwWMjAziaWUTiqXPkl1QS7OepqsE1jcvOJhnj+2UqJ7aLo7A0mZQbGx0NTz8Njz8O69bB3LnwxBNq3B//CB06wB132EWnOJaCINSHUbzTNnIsa+ag14aPjw9LlixhyZIldY7p0qULK1eurHeeMWPGsGvXrnrHzJ07l7lz517UWlobcSwFoSnobqWeH/TZZ1QvUvmHlbb2Gae++4nj2cX4e1m4ZlCHBk2r51n2zLHlV9rycYiIgCCb43niBABDba7lmqQMPt2Ryt9XHmTzsezzvhi3Hldh8Bsu6UhsZADVVo0fjmaRVVjOV7vO0CP3NIH59lB5v4xjmDQrnfJtwrJ7d/VsMsHf/gavvgoeHrBpE/z2t6pqPToa5syBW291dinFsRQEoR5MFr14p2WrwoXWQ4SlIDQFvVKve3eIj4eqKrxLi0mK7EbupGsAuCTnFADXDu6Iv7ctOFBVBffdp1y/WrjClmdpFO7owtJkqjPP8sNtKTz8xV7+s+k4t7+xjetf2czaJLU+Pb8yJi+d+265nL/uUCHq7w9l8vj/9pNfWsl1BcfUvDHKVY3LOknHgiy8q6uUgIyp4bbef7/KL/3Tn5TgBSU4/fxUWEvPt+zQAWr0VxMEQXDEZNFzLNuGYylcPCIsBaEp6I5lZCRl9//OOLz7d48QNXYUAPcE5PHAuFj+b6KDa/fjjyqk/NhjcOTIedP26xBM+2AfeuXYQuG6sITzhOVverXDz8uCxWxiSOcQrh/cAW8PM3tS85j17s+8+cNxjmQWkVtcwZRj2/DMyuCyL96mXdE5VuxJY+W+dCxmE9MrlABmxgzw98enspwxx23FN1272nOgHImJUeI4NVV9jnvusZ+77TZ46SV4770G/zgFQfiVYrH90d1G9goXLh7JsRSEpuAgLNf3v4KqPldQFh7BTX+6FzauByDg4H5jRx2D776zv/7Pf+Bf/3I6bTGb+PiOQcT82eaI6jmWcJ6w7BDiy9Y/jccEBPqoXM7Hisp5+fujLN18kqdXHuTKvirEfXmREqrmygpm7lvFM/G3AXD/6G6EvLtZzTt2LHz/PWzezKTkn9QxPbezLnx87OvSMZtV7qUgCMKF0HMsq0VYugviWApCU9BD4VFRrD16jgeue4TkP/4NL08LDB6szh09qnasccSxb9k770BZ2XlTd8k5jUnTVJV1pENRji7gjh0zDgX5eBqiEiAiwJvHr4njthExaJrKvwTom3bUGHPH7m/xrqqgRzt/HuzmAWfOgKcnXHopXHIJAJedtu36o+dXCoIgtABGjqU4lm6DCEtBaAo2x9Larh0bklVfyPF9bS0gIiJA36lh7177Nfn5sH27eh0eDjk58MUX58+dZKvQjotTuZU6NRzLujCZTDx5XX/iu4cD4FNZRljqceO+QQW5/LlwD0umDcF784/q+PDhKkfSJizN+r69F3IsBUEQLgZbjqUIS/dBhKUgNAWbY5nqGUhucQVBPh7GbjiA3bV0bCWxcaPq1dazJ8ybp4699tr5czsKS0d0YXnypHOro1rwtJh59Y4h9O8YxKTqTExWK0RFwUMPATB929f0CfeF1avVBaNHq+chQ5wnEsdSEIQWRHcs20qDdOHiEWEpCE3B5ljuKvcCYEzvSDwtDv872Zw/du+2H9PzKydMUMUuFosq5vnwQ/jmG/VISrK7nI75laBa+vj5KXF66tQFlxji58X/5lzO4l6afU2zZqk59uyB4GDQm/7qwrJfPxUW1xHHUhCElsToYynC0l0QYSkITcEmLH/IU6Hq8X1rNCivzbHU8yvHj4f27eH669X7adPUftxXX62E3fLl6nhNx7KWlkNO5OXZG5PbMJtNmPQ1DBmi8jbvvlu9Ly1V7YCmTVNiF8DLS61Bp8Y+uYIgCM2JyUPVEItj6T6IsBSEpmALhe+u8MZiNjGmVw1hqTuWBw6osPXZs8qNNJlU9TWofo+XXKKE3NChSowGBqpzISEwbNj5961LWC5bBqGhanvFmujCUl/Tv/4FX34J+/apPM/333d2KfVxkZH29QiCILQAJrMtj1xyLN0GaTckCI2lvFwV4gDZfiEM6xJKsJ+n85iuXdVOOQUFcPCgEnGgBGSYamzOwIFqn3FH9D3I/f0hIOD8e9clLN9+Wz2vWAF33mk/Xllpv7eeP+njAzfcUPfnu+QSVbEu+ZWCILQ04li6HSIsBaGx2MLgVWYL+T4BTNCrwR0xmZQDuWkTPPKIClODCoPXh8mkimzqojZhWVAAm229KA8edB6flKQc0+Dghoe1b70VVq1ybnouCILQApgkx9LtEGEpCI3FJiyz/YLx9fJgysD2tY8bPVoJS73yGi4sLC9EbcJy/Xq1VSRAcrJ6bXMBjDD44MHOrYvqo107VUgkCILQwpgs4li6GyIsBQHILionwNsDH8/zty8sqaji+TWH6RLhz+0jOnPueCoRQI5fCA8l9KZDiG/tkz72mBJ0+/crJzEqCsaNu7iF6sLyxAm7gFy1yn6+okLt493Lto2kHmqv2UZIEAShLSB9LN0OEZbCr54T2cVMXLyRsb0j+c9054IZq1Vj3se7jR1svtp5miu3/8xsoCI8ghmXda17Ym9vmDpVPZqLjh3tuZvLl6vKct0RtVigulqJWF1Y1izcEQRBaEMYO++IY+k2SFW48Kvn55O5VFZrrD2YQWah8xaL/1qTzJqkDLwsZgK8PdiZkkfeqTMA9OjfA4u5geHl5sJshgcfVK//9CcV+j5xQlV1X321Oq43WK+utvfRFMdSEIQ2iMnmWJrFsXQbRFgKv3pSc0sAVZC95kCGcfzzxNO8ukHty/3MTQNZPX80v+nVjojiPACCunRs9bUCavec8HA4dAjuuksdGzUKRoxQr/UCnt27oahIOZy9e7tipYIgCPVilqpwt0OEpfCrJzW7iPk/fMClKXtZfSAdgNyicvym384HH/+J+Ze25/pLOtIxxJdld49gRqyfujAysp5ZW5CgIPjzn9XrbdvUc0KCvaG67lhu2KCeR4+2F/MIgiC0JcySY+luiLAUfvUEb/2B32/+iBeWP8vWo1nklVTw7RcbuCrpB0ad2suDX/3babxXdpZ64SphCTB7NnTubH+fkGDfAvLgQbXtoy4sx4xp7dUJgiA0CCMULo6l2yDCUhBOq5zJ6KJc+p49wqr96aR/vtw4bXp3mfOONrZ2Q/X2m2xpfHzgiSfU6/btYdAgta+3pyeUlKi8y02b1HkRloIgtFHMtnZD4li6Dy4Vlps2beKaa66hQ4cOmEwmvv76a6fzmqaxcOFC2rdvj6+vLxMmTODIkSNOY3Jzc5k2bRpBQUGEhIRwzz33UFRU5DRm7969XHHFFfj4+BATE8MzzzzT0h9N+IVQXF6FT2628f7KI9v4x6pDDDy0AwCr3t5n9mzQ/9vThaUrHUuA6dPh9dfh889VOMnDw14N/uGHqnI8ONi+b7kgCEIbQxxL98OlwrK4uJhBgwaxZMmSWs8/88wz/Pvf/+a1115j27Zt+Pv7k5CQQFmZvXJ32rRpHDhwgLVr17JixQo2bdrEfffdZ5wvKChg4sSJdOnShcTERP71r3/x17/+lf/85z8t/vmEtk/quRLCS/KM91ce2UpRYSnxKWobRPOHH6ocxaIitRON1do2HEtQYvK+++Cyy+zH9DzL119Xz6NHqzZEgiAIbRHb95NUhbsPLs3onzx5MpMnT671nKZpvPDCCzz22GNcd911ALz77rtERUXx9ddfc+utt3Lw4EFWrVrFjh07GDZM9R986aWXuOqqq3j22Wfp0KEDH3zwARUVFbz99tt4eXnRr18/du/ezfPPP+8kQB0pLy+nvLzceF9YWNjMn1xoK5zKKTGqvAH6ZJ/i2oMbCagoxRoegXnoUHjvPdWY/IcfYO1a1cYH1A41bQ1dWJ5R4X0JgwuC0JYxe4iwdDfabI7liRMnSE9PZ8KECcax4OBgRo4cyZYtWwDYsmULISEhhqgEmDBhAmazmW22atktW7YwevRovLy8jDEJCQkkJydz7ty5Wu+9aNEigoODjUec/stacDtSc0sIL8lXb2xbHj6ycRkA5gnjlSvYuTP89rdqzF/+op5DQsDhv6k2g17AoyPCUhCENoy+paPJqrl4JUJz0WaFZXq6avsSVSPcGBUVZZxLT08nskaem4eHB2FhYU5japvD8R41efTRR8nPzzceSXr7FsHtSMktoV2x7Q+MSZMAiCrKVe8nTrQP1JuS71C5ly4Pg9eF4x9BISGqqEcQBKGNYraoP+gt4li6DW1WWLoSb29vgoKCjEdgYKCrlyS0EKdyHBzLWbOcT155pf31iBEwcqT9vasLd+qiVy+jL5zkVwqC0OaRqnC3o80Ky+joaAAyMjKcjmdkZBjnoqOjydQLKWxUVVWRm5vrNKa2ORzvIfx6OZ1dRJguLEeOhIED1evevSEmxnnwAw/YX7dVx9LbW7UdAgmDC4LQ5jHbqsItViuaJuFwd6DNCstu3boRHR3NunXrjGMFBQVs27aN+Ph4AOLj48nLyyMxMdEY8/3332O1Whlpc5fi4+PZtGkTlZWVxpi1a9fSu3dvQkNDW+nTCG2RaqtGUVomHvpfyu3awa23qte2gjEnbr4Z9D9G2qpjCTBvHlx6Kdx+u6tXIgiCUC9mT92x1JA0S/fApcKyqKiI3bt3s3v3bkAV7OzevZuUlBRMJhPz5s3jb3/7G//73//Yt28f06dPp0OHDlx//fUA9O3bl0mTJjFr1iy2b9/OTz/9xNy5c7n11lvp0KEDALfffjteXl7cc889HDhwgE8++YQXX3yRBQsWuOhTCy3Jqv3p/GfTsQb95ZtRUEZQgcqn1MLCVHPxhx6Cb76xNx93xMsL/vQn9fryy5tz2c3L734HW7a0XVdVEARBx2yvCreKY+kWuLTd0M8//8zYsWON97rYmzFjBkuXLuXhhx+muLiY++67j7y8PC6//HJWrVqFj4+Pcc0HH3zA3LlzGT9+PGazmalTp/Lvf9u34AsODmbNmjXMmTOHoUOHEhERwcKFC+tsNST8cqmosjL/k92UVlbTOzqI3/Sqvx3QqZwSImxhcJPuQHp4wFVX1X3RAw/AtGkgbrcgCMJFY/awhcJFWLoNLhWWY8aMqddZMplMPPnkkzz55JN1jgkLC+PDDz+s9z4DBw7khx9+aPI6hV8GB87mU1qpekx++nPqBYVlam4JEXpFeGPcvbCwpi5REARBcMBkKzA0aRqiK92DNptjKQiN5eeT9r6kaw9kcK64ot7xKY49LNtyzqQgCIKbYvZQ/pZFs1ItSZZugQhLwW3YcTLXeF1RbeWrXWfqHX8qt4QIfTtHyUcUBEFodUy29miSY+k+iLAU3AJN00g8pRzLGy7pCKhweH2pFim5JYTr2zmKYykIgtDq6FXhZqtVqsLdBBGWgltwIruYnOIKvD3M/Omqvnh5mDmUXsje0/l1XpMqjqUgCIJLMdtyLM1o0sfSTRBhKbRNDh6EO+6A1NQGDdfzKwd1CqFdoDeT+6t+ky99f5TEU7kUlFU6jV+x9yy5xRVGVbg4loIgCK2P2UMJS4s4lm6DS6vCBaFO/vY3+PBD1ZD82WcvOFzPrxzWVbUBumV4DP/dfZbvDmbw3cEMzCZ17M9T4jibV8rDn+8FoFt1kZpAHEtBEIRWR68KN0vxjtsgwlJom+zcqZ6TkpwOl1VWYzaZ8PJwNtv1/MrhXVUroPju4fz5qr78eDSbwxmFpOWX8dH2VDYdzsbDYqKkoprLeoQTXGirJBfHUhAEofVxEJYSCncPRFgKbY+iIkhOVq8dhGVOUTkJL/xAhxAfvvrdKCxmEwDZReUczy4GYEhn5ViaTCZmje7OrNHdAdh6PIeHPt9Dam4pAB2CfXjpmp6Y7itRk4tjKQiC0PoYVeGypaO7IDmWQttjzx6MTrmnTimhCXy9+yzZReXsPZ3P6gPpxnA9v7J3VCDBfp61Tnlp93C+/f1opsd3oXdUIK/dOdTew9LXF/z9W+7zCIIgCLVjdth5x2p18WKE5kAcS6HtsWuX8/tDh2DYMD5PPG0cen3jMSb3j8ZkMrH9hHN+ZV0EeHvw5HX97Qe2HFTPUVFgMjXL0gVBEIRGYAuFAyIs3QQRlkLbQ8+v1ElK4kCHnhxMK8DLYsZkgj2n89l6PJdQf08+2HYKgFGxEY27T2amepb8SkEQBNdgtgdOtapqFy5EaC4kFC60PXRhGROjng8c4ItEtYvOhLhIfjtMHX/p+yPM/XAX5VVWxvT+/+3deXzNV/4/8NdNbnIjsiIrkaASgoqlIioUIVpjKEpRVG01UjTVGjMqdNrqF9PSX0P3qK2WGTWKoZZgSiyJLYsEpY0tSYkslmz3nt8fH/fGlWgk+dzFJ6/n45GHez+fc8/nnNM0eef9Oed8PNC/jbe0RVHnzkBBQdXXyc6W/uX8SiIiy3ggY6llYKkIDCzJuhQVAamp0uvRowEAutRUbDklBZbDOjXBxPBmsFEBh3+5iQs5t+HprMGSl9rDJv0ssHYtkJQEfP991dfSZywZWBIRWYZRxrLMgg0huTCwJMv773+BL76QFuykpABlZUDDhkC/fgCAotMp0mbmThr0aOkB/4b18Xw7HwDS1MilI0LQyEkDrFlTXmdcXNXX1WcseSuciMgyOMdScTjHkiyroAAYMkTKVKrVgFa6FVIW0gH/vuuMEQAcrmbCobQIQzo2h/rXS4CzM96MCMS5rEK83KUpuj3VCNDppGyl3tGj0tN7Wrd+9LWZsSQisqwHMpZgxlIRmLEk0ystBUpKKj+3ebMUVALAm28CW7cCAP7n5IfZB7OQW88FNkKgdcF1vOKpBdq2BVq1wlNXzmF3dE9M6N5M+uzPPwOZmYCzsyHTWWXWkhlLIiLLejBjqWXGUgkYWJJpFRRIi3D69Cnfm/JB69ZJ/9rbA4WFwPbtAIAfVNIWQHeatwQArA1zRtONq6Qg9NYtqb7ExPJ69LfBhw0Dpk6VXq9e/cd/ATNjSURkWQ9kLHVcvKMIDCzJtE6fljKDP/8MHDtmfC4rC9i7V3q9ZQvg4GA4dcajGTr5u8Ove2cAgGPKaeCbb6STTZsCeXlARIQUiBYVARs3SufGjAEGDAA8PKT6d+58dNuYsSQisqwHA0stA0slYGBJpvXrr+WvH5wDCQAbNkhzI7t2BZ5/HvjgAwBAgYMTfnP3waTw5kBwsFT2yy+B3FzA318KVrt3B/LzgT/9CWjZUnrdpAnQsydgZydtO6T/3MOZ0rQ0YMoUqT6AgSURkaWoVNDpH1DBjKUiMLCkSv2UmoWLv9+ufUWXLpW/3rDB+Na0PtC8v60QZsxAYtTfMONPb6FpIyf0DfYqDyzvSM8Cx+uvA25u0krymTOlRzFeuf9EnlGjyv/6nTBB+vfHH4G//10KLvPzgXHjgDZtpIATACIjgUbV3FidiIhko1NJP7d1OgaWSsBV4VRBfEYOJq9OQmO3eoif9Rzs1bX4++PBjGVODrBnD9C/P3D+PHD8OIStLV672wyJMbvg61YP2R49kVe/FO91bwZbG1V5YAkAGk15wOjkBHzyCfDuu8CKFVIWctas8rJt2gDLlgEzZgALFwLXrgHx8dICH5UKGDxYCkzDw/k4RyIiC5ICSy0Eb4UrAgNLquD7o5kAgKt597Dl5FUMf8av5pXpM5be3tKcx7Vrgf79UbTiCzgAONA0BPG5AFCGjOxCAIC7ox2GdWoifc7HB3B1lbKNI0ZIcycf1KCBlJGszPTpUgbzjTeA776TjrVoIS3qCQureZ+IiEg2OhsbQAuIMq4KVwIGlmTk98Ji7EvPMbxfceAXDO3URMoe1oQ+YzlrlvT1ww8Qc+fC4ZN/AgD+1a4PXurUBOO6BeDmnRJcy7uHED83ONrf/9ZUqaQM59at0nZE1RUVJe2POXu2FJh+/LGU7SQiIqsg7t81ErwVrggMLMnI5hNXUKYTaOPrgqt593Dpxh3sSL6Oge19q19ZWRlw+bL0+uWXgeXLgYsXobq/SOfzbsMxftk76BTQ8I/rWb1a2oqoQYPqtwGQ5mVOmcJb3kREVkhrI+1lqSvlBulKwMU7ZCCEwIZEKRAc09Uf47tJm4/Hxl+AqGwPyqpcuSI9SUejkW5pjxoFANCpVJgXMQV35/+j6qASkFZ51zSo1GNQSURklcozlrwVrgTMWJJB4m+3cPH3O6hnZ4s/tfdFmVaHLw/+gvSsQuxLz0Gf1tXcSPz+/Eqdvz+KynRwnDkT6Ylnsah+G6R2DEd8z+Ym6AURET1JdPczltxuSBkYWBIAoFSrw/L4CwCAAU/7wEkjfWu80tUfXxy8iM/iL6B3K0+oqpP5uz+/8qjWGaNidqFpA0fkdJ6Ie6VafNy/Vfk8SiIiqrOYsVQW3gonFJVqMXVNEuIzfofaRoVxYQGGcxPCm8FebYOTmXk4cjG3ehXfz1hedPaAEMBvN+/iXqkW7Zu4YnBIYxl7QERETyrDPpZazrFUAqaM6ricgiK88f1JHL2UC43aBite6Yh2TVwN5z2dHTCisx9WH/kNsfEXENbi0XMif/n9NlYn/IZJPZqjsVs9aC9egi2Ay67e+PDFdgho5IgruffwXJAHbGq6ypyIiBRF3H+whdAyY6kEzFjWUacu52H69yfR7aN9OHopF04aNVa91gW9W1WcRzmlZ3OobVT4+cINnLqc98g6Ny5ehUGvD8WHH65HUakWuakZAIA7Po0xvHMTdGvRCMOf8YOni8Mj6yAiorpF3M9YghlLRWBgWQf9lJqFF5cfwtbT11CmE+js7471k7sitHnl2cgm7o4YdP/Wdez9eZgPu5x7F102r0SH6xkY8J+vseDHVMMcyw49O0Jty281IiKqSKfPWHKDdEXgb/s6JrugCLP/fQZCAH2DvbDtje7419RuaNvY9Q8/N/W5FlCpgN1p2Ziw8jg2n7iC28Xlf11uSryMDtekDGWfX47jp/hkNMy7AQCI6B9qug4REdETTadfvMNHOioCA8s6RKcTeGvjady6W4o2vi6IHdWxyoBS7ylPJ8O+lnvTcxC98TT6fXwAOYVF0OoEjuw6ggb3CgAAGm0pxidthQ0ESjT14OJfg83ViYioThD3txuylifvHDx4EAMHDoSvry9UKhW2bNlidF4IgXnz5sHHxwf16tVDREQEzp8/b1QmNzcXo0ePhouLC9zc3DBhwgTcvn3bqMyZM2cQHh4OBwcH+Pn5YdGiRRXasmnTJrRq1QoODg5o164dduzYUe22mBsDyzrk20OX8POFG3Cws8GylzvAXl29//zzBgZj95s9ML1PS3i7OOBafhGmf38S8ek58Es/bVR2wqnt0osAf25OTkREj6SfY2ktt8Lv3LmD9u3bIzY2ttLzixYtwqefforPP/8cR48eRf369REZGYmioiJDmdGjRyM1NRW7d+/Gtm3bcPDgQUyePNlwvqCgAP369YO/vz+SkpKwePFizJ8/H19++aWhzOHDhzFy5EhMmDABJ0+exODBgzF48GCkpKRUqy1mJ6hKly9fFgDE5cuXLd2UGrtTXCpav/tf4T97m1h75Lc/LqzTCfH110KcOPHIIhdyCkXw/fraxuwUa9v3FwIQYtgw6V/914ABMveEiIiU5LJPgBCAOPz1JvnrruXvbwDihx9+MLzX6XTC29tbLF682HAsLy9PaDQa8f333wshhEhLSxMAxPHjxw1l/vvf/wqVSiWuXr0qhBBi+fLlwt3dXRQXFxvKzJ49WwQFBRneDx8+XAx46HdoaGiomDJlymO3xRKYsawj4tN/x90SLfwa1MPILn5/XHjPHmDiRCAyErhzp9IiLTyc8NHQpwEAhUVl6HAtXToxahQQFlZeMCBAhtYTEZFSGTKWJpxjWVhYiIKCAsNXcXFxjeq5dOkSsrKyEBERYTjm6uqK0NBQJCQkAAASEhLg5uaGzp07G8pERETAxsYGR48eNZTp0aMH7O3tDWUiIyORkZGBW7duGco8eB19Gf11HqctlsDAso7YkXwdAPBCO5+qn56ze7f07++/AytWPLLYwPa+GNPVH07FdxF04zfpYFgYMHp0eaFmzWrTbCIiUjj9qnCY8Mk7wcHBcHV1NXwtXLiwRvVkZWUBALy8jLfm8/LyMpzLysqCp6en0Xm1Wo0GDRoYlamsjgev8agyD56vqi2WwMCyDrhXosW+9BwAwIB2PlV/ID6+/PWiRY/MWgLSvMsVLctgI4QURHp7Ay+9BNjef/YrM5ZERPQHyudYmm4fy7S0NOTn5xu+5syZY7Jr1XUMLOuA+Iwc3CvVool7PbSrahV4fj5w4oT02sdHylp+/vkji9vZ2iD8xv0VaPpb4J6eQFQU0KIF0LOnDD0gIiKlKn/yjjDZNZydneHi4mL40mg0NarH29sbAJCdnW10PDs723DO29sbOTk5RufLysqQm5trVKayOh68xqPKPHi+qrZYAgPLOmD7/dvgAx7nNvjBg9LtiJYtgQ8+kI5VkbWEfi7Hg3Mrly4FLlwAGjWqecOJiEjxygNL63/yTrNmzeDt7Y29e/cajhUUFODo0aMIu/87MCwsDHl5eUhKSjKU2bdvH3Q6HUJDQw1lDh48iNLSUkOZ3bt3IygoCO7u7oYyD15HX0Z/ncdpiyUwsFS4eyVa7Dsr/eX0QnVug/fqBbzyinR7OydHykDevVuxvE4HHDkive7WTaZWExFRXSHMMMeyOm7fvo1Tp07h1KlTAKRFMqdOnUJmZiZUKhVmzpyJ999/H1u3bkVycjLGjh0LX19fDB48GADQunVr9O/fH5MmTcKxY8dw6NAhREVF4eWXX4avr7Sv86hRo2Bvb48JEyYgNTUVGzZswLJlyxAdHW1ox4wZM7Bz507885//RHp6OubPn4/ExERERUUBwGO1xSIsth79CfIkbze048w14T97m3j2o71Cp9NV/YGQEGmbIP1WBevXl28dFBgoxI8/CrFlixCffirEP/4hxJtvSuccHYUoLTVtZ4iISHHOP/W0EIA4+NHnstddk9/f8fHxAkCFr3HjxgkhpG1+3n33XeHl5SU0Go3o06ePyMjIMKrj5s2bYuTIkcLJyUm4uLiI8ePHi8LCQqMyp0+fFt27dxcajUY0btxYfPTRRxXasnHjRhEYGCjs7e1FmzZtxPbt243OP05bzE0lhDDdpAaFuHLlCvz8/HD58mU0adLE0s15bMVlWrzy9VEc//UWJvdojr+90PqPP5CbK926FgK4fl1aiAMAu3YBr70GXLv26M/27g08lLInIiKqyvlWHdEy4yQOfrgcPeZMlbXuJ/X395NMbekGkGmUanV4Y91JHP/1FhztbTHimSr2rgSk+ZVCAK1blweVgLSfZUoKEB0tBY/e3oCfH9CwIWBnB9SrJwWeRERE1ST0c//5rHBFYGCpQDqdwNubTuOntGzYq23w9djOaOHhVPUHH5xf+TB3dyAuTt6GEhFRnad/Vji01jHHkmqHi3cU6Kv/XcSWU9egtlFhxeiO6PbUY6zMvnMHWL9eet2nj2kbSEREdJ9hVbiOGUslYGCpMJdz7+KTPecAAO8Naos+rb2q+MR9K1ZIq7+bNwcGDjRhC4mIiMoZVoUzY6kIVh1Yzp8/HyqVyuirVatWhvNFRUWYNm0aGjZsCCcnJwwdOrTCRqGZmZkYMGAAHB0d4enpibfffhtlJtzd35KEEIjZmoqiUh1CmzWo+pngenfuSHtVAsDcudK8SSIiIjMo326IGUslsPo5lm3atMGePXsM79Xq8ia/+eab2L59OzZt2gRXV1dERUVhyJAhOHToEABAq9ViwIAB8Pb2xuHDh3H9+nWMHTsWdnZ2+PDDD83eF1PblZqFfek5sLNV4YMX21a9GbpebKz0hJ0WLYAxY0zbSCIiogeUZywZWCqB1QeWarW60kcT5efn45tvvsG6devQu3dvAEBcXBxat26NI0eOoGvXrvjpp5+QlpaGPXv2wMvLCyEhIfjHP/6B2bNnY/78+bC3tzd3d0wi/24p/nP6Kj7dKz1acUqPFnjK0/nxPlxYCCxeLL1+911AbfXfEkREpCD6xTuCt8IVweqjiPPnz8PX1xcODg4ICwvDwoUL0bRpUyQlJaG0tBQRERGGsq1atULTpk2RkJCArl27IiEhAe3atYOXV/k8w8jISEydOhWpqano0KFDpdcsLi5GcXGx4X1hYaHpOlgDd0vKsCs1C2evF+Ls9QIcvZSLkjLpf8gWHvUR1fupP65Ap5Oe/71lC3D4sHQrvGVLYPRo0zeeiIjoQdxuSFGsOrAMDQ3FypUrERQUhOvXr2PBggUIDw9HSkoKsrKyYG9vDzc3N6PPeHl5ISsrCwCQlZVlFFTqz+vPPcrChQuxYMECeTsjk8u5dzHhu+M4l33b6Hgrb2cM7+yHoZ2awMHO9o8r+ewzYMaM8veensAXXzBbSUREZids7//OspJHOlLtWHUk8fzzzxteP/300wgNDYW/vz82btyIevXqmey6c+bMMXpe59WrVxEcHGyy6z2uE5m3MHlVIm7cLoGHswYD2vkg0MsZIX5uaO3j/HhzKm/cAGJipNfR0cD48UBwMGBj1eu4iIhIqTjHUlGsOrB8mJubGwIDA3HhwgX07dsXJSUlyMvLM8paZmdnG+Zkent749ixY0Z16FeNVzZvU0+j0UCj0RjeFxQUyNiLmjl04QbGrzyOkjIdgn1c8M2rneHjWoPget48IC8PaN9eWgluW0V2k4iIyITKV4UzY6kET1Sa6vbt2/jll1/g4+ODTp06wc7ODnsfeD51RkYGMjMzERYWBgAICwtDcnIycnJyDGV2794NFxcXq8hAPq4LObfx+poklJTp0LuVJza9HlazoPLMGemWNwAsW8agkoiILI6rwpXFqjOWs2bNwsCBA+Hv749r164hJiYGtra2GDlyJFxdXTFhwgRER0ejQYMGcHFxwRtvvIGwsDB07doVANCvXz8EBwdjzJgxWLRoEbKysjB37lxMmzbNKCNpzXLvlGDCd8dRWFSGTv7uWD66Y9VzKB8lOlr6i/Cll4CePeVtKBERUU3YcI6lklh1YHnlyhWMHDkSN2/ehIeHB7p3744jR47Aw8MDAPDJJ5/AxsYGQ4cORXFxMSIjI7F8+XLD521tbbFt2zZMnToVYWFhqF+/PsaNG4f33nvPUl2qluIyLV5fnYTfbt5FE/d6+GJMp5oHlefOAXv3SllK/WboREREFsaMpbJYdWC5Xv/s6kdwcHBAbGwsYmNjH1nG398fO3bskLtpJieEwN82p+DYr7lw1qjx7avPoJFTLbKsa9dK//brBwQEyNJGIiKiWjOsCheWbQfJ4omaY1mXrDjwC/594gpsVMBnozsi0OsxNzyvjBDAunXSa+5VSUREVkSfsVTxkY6KwMDSCu1MuY5FOzMAAPP/3AY9Az1qV+Hx48CFC4CjIzBokAwtJCIikgefFa4sDCytzJkreZi54RQA4NVuARgbFlD7SvXZykGDACen2tdHREQkF8McSy7eUQIGllbkev49TPwuEUWlOvQM9MDcAa1rX2lZGaCfqzpqVO3rIyIikhOfvKMoVr14p64QQuDMlXzM2ZyMnMJiBHo54f+N6gC1rQxxf3w8kJ0NNGwIREbWvj4iIiI56edYclW4IjCwtKC8uyX4V9IVbEq8gozsQgBAw/r2+GbcM3BxsKtd5TodsGULMHeu9H74cMCulnUSERHJTHAfS0VhYGlB2QXFeH/7WQCAvdoGz7f1xhu9n4JfA8eaVZibC+zfD/zvf8COHdLelQDg7AxERcnTaCIiIjlxVbiiMLC0oCBvZ7zUqQme9nPDn5/2hatjLTKKV64AzzwDZGWVH3N1lQLKGTMAj1quLCciIjIFzrFUFAaWFrb4pfbyVLR0qRRU+vpKq7/Dw4EBAwAXF3nqJyIiMgXDdkPcIF0JGFgqQX4+8OWX0uuvvgJeeMGy7SEiInpc9zOWXLyjDNxuSAm++gooLASCg4H+/S3dGiIiosenz1gK3gpXAgaWT7qSEuk2OAC89Vb5/6BERERPAm43pCiMQp50GzcCV68CXl58DjgRET1xxP09m1XMWCoCA8sn2a1bwPvvS6+nTwc0Gsu2h4iIqJpU+n0s+UhHRWBg+aTKywP69QMyMgBPT+D11y3dIiIiourTL95hxlIRGFg+ifLzpcczJiYCjRoBe/YADRpYulVERETVZ6MCwDmWSsHA8kmTmAh07gwcOyY9/3vPHqBdO0u3ioiIqGZs7+98yA3SFYGB5ZNACOnJOh99BHTrBly4APj5SUFle5k2WCciIrIEwyMdGVgqATdIt2ZaLTBrFrB+vfGjGocOlfaudHe3XNuIiIjkoJ9jycBSERhYWrMlS8r3qLS1Bdq0Ad54A5gwAVCpLNo0IiIiOaj02w3pOMdSCRhYWqukJGDuXOn10qXApEmAo6NFm0RERCQ7rgpXFAaW1ujOHWDUKKCsDBg2TNqjkhlKIiJSIJXhyTsMLJWAi3es0cyZwLlzQOPGwBdfMKgkIiLlYsZSURhYWptVq4Cvv5aCyVWruD8lEREpG1eFKwoDS2uSnFz+BJ2YGKB3b8u2h4iIyNS4KlxRGFhai+xs4KWXgHv3pEc16hfuEBERKZjKEFhyVbgSMLC0Blu2AG3bSs/9btIEWLvW8BccERGRkpVvNyQs3BKSAwNLSyooAF57DXjxReDGDekpOrt3S8//JiIiqgtspESKDTOWisDA0pLS0oDvvpMW6vz1r8DRo0CrVpZuFRERkfmo9avCmbFUAu5jaUlduwIffwx07AiEh1u6NURERGZn2MeSGUtFYGBpaTNmWLoFREREFlO+eIcZSyXgrXAiIiKyGBU3SFcUBpZERERkMfpV4Tbcx1IRGFgSERGR5TBjqSgMLImIiMhi9It3uN2QMjCwJCIiIstRS+uIud2QMjCwJCIiIouxsWXGUkkYWBIREZHFlK8KZ8ZSCRhYEhERkeXY6h/pyMU7SsDAkoiIiCyG+1gqCwNLIiIishgb7mOpKAwsiYiIyGJUtvpV4QwslYCBJREREVkMn7yjLAwsiYiIyGJ8GtQHALg7qC3cEpID/ysSERGRxdTzawy88w7sXVws3RSSAQNLIiIishwfH+D//s/SrSCZ8FY4EREREcmCgSURERERyaJOBZaxsbEICAiAg4MDQkNDcezYMUs3iYiIiKwM44WaqzOB5YYNGxAdHY2YmBicOHEC7du3R2RkJHJycizdNCIiIrISjBdqp84Elh9//DEmTZqE8ePHIzg4GJ9//jkcHR3x7bffWrppREREZCUYL9ROnVgVXlJSgqSkJMyZM8dwzMbGBhEREUhISKhQvri4GMXFxYb3+fn5AIDr16+bvrFEREQkC/3v7fz8fLg8sJ2RRqOBRqOpUL668QJVVCcCyxs3bkCr1cLLy8vouJeXF9LT0yuUX7hwIRYsWFDheJcuXUzWRiIiIjKNtm3bGr2PiYnB/PnzK5SrbrxAFdWJwLK65syZg+joaMP7srIynD17Fn5+frCxkXf2QGFhIYKDg5GWlgZnZ2dZ635ScAw4BnocB44BwDEAOAaAPGOg0+mQmZmJ4OBgqNXlIU9l2UqSR50ILBs1agRbW1tkZ2cbHc/Ozoa3t3eF8pWlyJ999lmTtK2goAAA0LhxY6M0fV3CMeAY6HEcOAYAxwDgGADyjUHTpk0fu2x14wWqqE4s3rG3t0enTp2wd+9ewzGdToe9e/ciLCzMgi0jIiIia8F4ofbqRMYSAKKjozFu3Dh07twZXbp0wdKlS3Hnzh2MHz/e0k0jIiIiK8F4oXbqTGA5YsQI/P7775g3bx6ysrIQEhKCnTt3Vpiga24ajQYxMTF1er4Hx4BjoMdx4BgAHAOAYwBYbgysNV54UqiEEMLSjSAiIiKiJ1+dmGNJRERERKbHwJKIiIiIZMHAkoiIiIhkwcCSiIiIiGTBwLKWFi5ciGeeeQbOzs7w9PTE4MGDkZGRYVSmqKgI06ZNQ8OGDeHk5IShQ4dW2Hx1+vTp6NSpEzQaDUJCQiq91pkzZxAeHg4HBwf4+flh0aJFpupWtZlrHPbv349BgwbBx8cH9evXR0hICNauXWvKrj02c34v6F24cAHOzs5wc3OTuTc1Y84xEEJgyZIlCAwMhEajQePGjfHBBx+YqmuPzZxjsGvXLnTt2hXOzs7w8PDA0KFD8euvv5qoZ49PjjE4ffo0Ro4cCT8/P9SrVw+tW7fGsmXLKlxr//796NixIzQaDZ566imsXLnS1N17bOYah82bN6Nv377w8PCAi4sLwsLCsGvXLrP0sSrm/F7QO3ToENRqdZU/P8k0GFjW0oEDBzBt2jQcOXIEu3fvRmlpKfr164c7d+4Yyrz55pv48ccfsWnTJhw4cADXrl3DkCFDKtT12muvYcSIEZVep6CgAP369YO/vz+SkpKwePFizJ8/H19++aXJ+lYd5hqHw4cP4+mnn8a///1vnDlzBuPHj8fYsWOxbds2k/XtcZlrDPRKS0sxcuRIhIeHy96XmjLnGMyYMQNff/01lixZgvT0dGzduhVdunQxSb+qw1xjcOnSJQwaNAi9e/fGqVOnsGvXLty4caPSesxNjjFISkqCp6cn1qxZg9TUVPz973/HnDlz8NlnnxnKXLp0CQMGDECvXr1w6tQpzJw5ExMnTrSaoMpc43Dw4EH07dsXO3bsQFJSEnr16oWBAwfi5MmTZu1vZcw1Bnp5eXkYO3Ys+vTpY5b+USUEySonJ0cAEAcOHBBCCJGXlyfs7OzEpk2bDGXOnj0rAIiEhIQKn4+JiRHt27evcHz58uXC3d1dFBcXG47Nnj1bBAUFyd8JGZhqHCrzwgsviPHjx8vSbjmZegzeeecd8corr4i4uDjh6uoqd/NlYaoxSEtLE2q1WqSnp5us7XIx1Rhs2rRJqNVqodVqDce2bt0qVCqVKCkpkb8jtVDbMdD7y1/+Inr16mV4/84774g2bdoYlRkxYoSIjIyUuQfyMNU4VCY4OFgsWLBAnobLyNRjMGLECDF37txq/Q4heTFjKbP8/HwAQIMGDQBIf2mVlpYiIiLCUKZVq1Zo2rQpEhISHrvehIQE9OjRA/b29oZjkZGRyMjIwK1bt2RqvXxMNQ6Pupb+OtbElGOwb98+bNq0CbGxsfI12ARMNQY//vgjmjdvjm3btqFZs2YICAjAxIkTkZubK28HZGCqMejUqRNsbGwQFxcHrVaL/Px8rF69GhEREbCzs5O3E7Uk1xg8/P96QkKCUR2A9HOxtj9TTMVU4/AwnU6HwsJCRf9crGwM4uLicPHiRcTExJig5fS4GFjKSKfTYebMmXj22WfRtm1bAEBWVhbs7e0rzIHz8vJCVlbWY9edlZVVYdd//fvq1GMOphyHh23cuBHHjx+3ukdtmXIMbt68iVdffRUrV66Ei4uLnM2WlSnH4OLFi/jtt9+wadMmrFq1CitXrkRSUhKGDRsmZxdqzZRj0KxZM/z000/429/+Bo1GAzc3N1y5cgUbN26Uswu1JtcYHD58GBs2bMDkyZMNxx71c7GgoAD37t2TtyO1ZMpxeNiSJUtw+/ZtDB8+XLb2y8GUY3D+/Hn89a9/xZo1a6BW15mHCloljr6Mpk2bhpSUFPz888+WbopFmWsc4uPjMX78eHz11Vdo06aNSa9VXaYcg0mTJmHUqFHo0aOH7HXLyZRjoNPpUFxcjFWrViEwMBAA8M0336BTp07IyMhAUFCQ7NesCVOOQVZWFiZNmoRx48Zh5MiRKCwsxLx58zBs2DDs3r0bKpVK9mvWhBxjkJKSgkGDBiEmJgb9+vWTsXXmY65xWLduHRYsWID//Oc/8PT0rPG1TMFUY6DVajFq1CgsWLDA8POALIcZS5lERUVh27ZtiI+PR5MmTQzHvb29UVJSgry8PKPy2dnZ8Pb2fuz6vb29K6wa1b+vTj2mZupx0Dtw4AAGDhyITz75BGPHjq1ts2Vl6jHYt28flixZArVaDbVajQkTJiA/Px9qtRrffvutXN2oFVOPgY+PD9RqtdEvkdatWwMAMjMza9d4mZh6DGJjY+Hq6opFixahQ4cO6NGjB9asWYO9e/fi6NGjcnWjVuQYg7S0NPTp0weTJ0/G3Llzjc496ueii4sL6tWrJ29nasHU46C3fv16TJw4ERs3bqwwRcDSTDkGhYWFSExMRFRUlOHn4nvvvYfTp09DrVZj3759Ju0bPcTSkzyfdDqdTkybNk34+vqKc+fOVTivn5j8r3/9y3AsPT29xot3HpyUP2fOHKtZvGOucRBCiPj4eFG/fn3x2WefydZ+OZhrDNLS0kRycrLh6/333xfOzs4iOTlZ5Obmytqn6jLXGOzatUsAEBcuXDAcO3XqlAAgMjIy5OlMDZlrDKKjo0WXLl2Mjl27dk0AEIcOHap9R2pBrjFISUkRnp6e4u233670Ou+8845o27at0bGRI0dazeIdc42DEEKsW7dOODg4iC1btsjbiVoyxxhotVqjn4nJycli6tSpIigoSCQnJ4vbt2+bpnNUKQaWtTR16lTh6uoq9u/fL65fv274unv3rqHM66+/Lpo2bSr27dsnEhMTRVhYmAgLCzOq5/z58+LkyZNiypQpIjAwUJw8eVKcPHnSsAo8Ly9PeHl5iTFjxoiUlBSxfv164ejoKL744guz9vdRzDUO+/btE46OjmLOnDlG17l586ZZ+1sZc43Bw6xpVbi5xkCr1YqOHTuKHj16iBMnTojExEQRGhoq+vbta9b+VsZcY7B3716hUqnEggULxLlz50RSUpKIjIwU/v7+RteyBDnGIDk5WXh4eIhXXnnFqI6cnBxDmYsXLwpHR0fx9ttvi7Nnz4rY2Fhha2srdu7cadb+Poq5xmHt2rVCrVaL2NhYozJ5eXlm7W9lzDUGD+OqcMthYFlLACr9iouLM5S5d++e+Mtf/iLc3d2Fo6OjePHFF8X169eN6unZs2el9Vy6dMlQ5vTp06J79+5Co9GIxo0bi48++shMvayaucZh3LhxlZ7v2bOn+Tr7COb8XniQNQWW5hyDq1eviiFDhggnJyfh5eUlXn31Vav4A8OcY/D999+LDh06iPr16wsPDw/x5z//WZw9e9ZMPX00OcYgJiam0jr8/f2NrhUfHy9CQkKEvb29aN68udE1LM1c4/Co75Vx48aZr7OPYM7vhQcxsLQclRBCVHqPnIiIiIioGrh4h4iIiIhkwcCSiIiIiGTBwJKIiIiIZMHAkoiIiIhkwcCSiIiIiGTBwJKIiIiIZMHAkoiIiIhkwcCSiIiIiGTBwJKIFGv//v1QqVTIy8uzdFOIiOoEPnmHiBTjueeeQ0hICJYuXQoAKCkpQW5uLry8vKBSqSzbOCKiOkBt6QYQEZmKvb09vL29Ld0MIqI6g7fCiUgRXn31VRw4cADLli2DSqWCSqXCypUrjW6Fr1y5Em5ubti2bRuCgoLg6OiIYcOG4e7du/juu+8QEBAAd3d3TJ8+HVqt1lB3cXExZs2ahcaNG6N+/foIDQ3F/v37LdNRIiIrxowlESnCsmXLcO7cObRt2xbvvfceACA1NbVCubt37+LTTz/F+vXrUVhYiCFDhuDFF1+Em5sbduzYgYsXL2Lo0KF49tlnMWLECABAVFQU0tLSsH79evj6+uKHH35A//79kZycjJYtW5q1n0RE1oyBJREpgqurK+zt7eHo6Gi4/Z2enl6hXGlpKVasWIEWLVoAAIYNG4bVq1cjOzsbTk5OCA4ORq9evRAfH48RI0YgMzMTcXFxyMzMhK+vLwBg1qxZ2LlzJ+Li4vDhhx+ar5NERFaOgSUR1SmOjo6GoBIAvLy8EBAQACcnJ6NjOTk5AIDk5GRotVoEBgYa1VNcXIyGDRuap9FERE8IBpZEVKfY2dkZvVepVJUe0+l0AIDbt2/D1tYWSUlJsLW1NSr3YDBKREQMLIlIQezt7Y0W3cihQ4cO0Gq1yMnJQXh4uKx1ExEpDVeFE5FiBAQE4OjRo/j1119x48YNQ9axNgIDAzF69GiMHTsWmzdvxqVLl3Ds2DEsXLgQ27dvl6HVRETKwcCSiBRj1qxZsLW1RXBwMDw8PJCZmSlLvXFxcRg7dizeeustBAUFYfDgwTh+/DiaNm0qS/1ERErBJ+8QERERkSyYsSQiIiIiWTCwJCIiIiJZMLAkIiIiIlkwsCQiIiIiWTCwJCIiIiJZMLAkIiIiIlkwsCQiIiIiWTCwJCIiIiJZMLAkIiIiIlkwsCQiIiIiWTCwJCIiIiJZ/H/hzrp7zryHPwAAAABJRU5ErkJggg==",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import seaborn as sns\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "scores = (\n",
    "    stories.sort(by=\"time\")\n",
    "    .group_by_dynamic(\"time\", every=\"1mo\", period=\"1mo\")\n",
    "    .agg(\n",
    "        [\n",
    "            pl.count(),\n",
    "            pl.sum(\"score\"),\n",
    "        ]\n",
    "    )\n",
    ")\n",
    "\n",
    "ax1 = sns.lineplot(data=scores, x=\"time\", y=\"count\", label=\"Posts\")\n",
    "\n",
    "ax2 = ax1.twinx()\n",
    "\n",
    "sns.lineplot(data=scores, x=\"time\", y=\"score\", ax=ax2, color=\"r\", label=\"Upvotes\")\n",
    "\n",
    "ax1.set_ylim(0, None)\n",
    "ax2.set_ylim(0, None)\n",
    "\n",
    "# Handle lines and their labels\n",
    "lines, labels = ax1.get_legend_handles_labels()\n",
    "lines2, labels2 = ax2.get_legend_handles_labels()\n",
    "\n",
    "# Combine them\n",
    "ax1.legend().remove()\n",
    "ax2.legend(lines + lines2, labels + labels2, loc=0)\n",
    "\n",
    "# increase dpi when displaying in jupyter\n",
    "plt.rcParams[\"figure.dpi\"] = 300\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(285603, 15)"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "id_to_text = hn[[\"id\", \"text\"]].to_pandas().set_index(\"id\", drop=True)\n",
    "\n",
    "stories = stories.with_columns(\n",
    "    pl.col(\"kids\")\n",
    "    .map_elements(\n",
    "        lambda kids: None if kids.is_empty() else id_to_text.iloc[kids[0]][\"text\"]\n",
    "    )\n",
    "    .alias(\"top_comment\")\n",
    ")\n",
    "\n",
    "stories.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "stories.write_parquet(\"data/stories.parquet\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div><style>\n",
       ".dataframe > thead > tr > th,\n",
       ".dataframe > tbody > tr > td {\n",
       "  text-align: right;\n",
       "  white-space: pre-wrap;\n",
       "}\n",
       "</style>\n",
       "<small>shape: (5, 15)</small><table border=\"1\" class=\"dataframe\"><thead><tr><th>id</th><th>type</th><th>by</th><th>time</th><th>title</th><th>text</th><th>url</th><th>score</th><th>parent</th><th>top_level_parent</th><th>descendants</th><th>kids</th><th>deleted</th><th>dead</th><th>top_comment</th></tr><tr><td>i64</td><td>str</td><td>str</td><td>datetime[μs]</td><td>str</td><td>str</td><td>str</td><td>f64</td><td>f64</td><td>i64</td><td>f64</td><td>list[i64]</td><td>bool</td><td>bool</td><td>str</td></tr></thead><tbody><tr><td>11369756</td><td>&quot;story&quot;</td><td>&quot;based2&quot;</td><td>2016-03-27 12:30:43</td><td>&quot;Dimple: An obj…</td><td>null</td><td>&quot;http://dimplej…</td><td>150.0</td><td>null</td><td>11369756</td><td>31.0</td><td>[11372346, 11372145, … 11374740]</td><td>null</td><td>null</td><td>&quot;&lt;a href=&quot;https…</td></tr><tr><td>15392159</td><td>&quot;story&quot;</td><td>&quot;pritambarhate&quot;</td><td>2017-10-03 13:17:38</td><td>&quot;Matcha – A fra…</td><td>null</td><td>&quot;https://github…</td><td>320.0</td><td>null</td><td>15392159</td><td>125.0</td><td>[15392855, 15393525, … 15392780]</td><td>null</td><td>null</td><td>&quot;I think it&amp;#x2…</td></tr><tr><td>15147497</td><td>&quot;story&quot;</td><td>&quot;AlexeyBrin&quot;</td><td>2017-09-01 11:21:56</td><td>&quot;Linux desktop …</td><td>null</td><td>&quot;https://www.ne…</td><td>146.0</td><td>null</td><td>15147497</td><td>157.0</td><td>[15148786, 15148659, … 15150049]</td><td>null</td><td>null</td><td>&quot;Been running L…</td></tr><tr><td>28218892</td><td>&quot;story&quot;</td><td>&quot;gamontserrat&quot;</td><td>2021-08-18 08:39:24</td><td>&quot;Environmental …</td><td>null</td><td>&quot;https://home.c…</td><td>41.0</td><td>null</td><td>28218892</td><td>6.0</td><td>[28219635, 28219619]</td><td>null</td><td>null</td><td>&quot;I miss the sol…</td></tr><tr><td>21961416</td><td>&quot;story&quot;</td><td>&quot;Decabytes&quot;</td><td>2020-01-05 14:44:26</td><td>&quot;Torvalds&#x27; Comm…</td><td>null</td><td>&quot;https://www.ph…</td><td>34.0</td><td>null</td><td>21961416</td><td>8.0</td><td>[21962017, 21962153, … 21962130]</td><td>null</td><td>null</td><td>&quot;Active discuss…</td></tr></tbody></table></div>"
      ],
      "text/plain": [
       "shape: (5, 15)\n",
       "┌──────────┬───────┬──────────────┬──────────────┬───┬──────────────┬─────────┬──────┬─────────────┐\n",
       "│ id       ┆ type  ┆ by           ┆ time         ┆ … ┆ kids         ┆ deleted ┆ dead ┆ top_comment │\n",
       "│ ---      ┆ ---   ┆ ---          ┆ ---          ┆   ┆ ---          ┆ ---     ┆ ---  ┆ ---         │\n",
       "│ i64      ┆ str   ┆ str          ┆ datetime[μs] ┆   ┆ list[i64]    ┆ bool    ┆ bool ┆ str         │\n",
       "╞══════════╪═══════╪══════════════╪══════════════╪═══╪══════════════╪═════════╪══════╪═════════════╡\n",
       "│ 11369756 ┆ story ┆ based2       ┆ 2016-03-27   ┆ … ┆ [11372346,   ┆ null    ┆ null ┆ <a href=\"ht │\n",
       "│          ┆       ┆              ┆ 12:30:43     ┆   ┆ 11372145, …  ┆         ┆      ┆ tps:&#x2F;& │\n",
       "│          ┆       ┆              ┆              ┆   ┆ 11374740]    ┆         ┆      ┆ #x2F;dc-js… │\n",
       "│ 15392159 ┆ story ┆ pritambarhat ┆ 2017-10-03   ┆ … ┆ [15392855,   ┆ null    ┆ null ┆ I think     │\n",
       "│          ┆       ┆ e            ┆ 13:17:38     ┆   ┆ 15393525, …  ┆         ┆      ┆ it&#x27;s   │\n",
       "│          ┆       ┆              ┆              ┆   ┆ 15392780]    ┆         ┆      ┆ really      │\n",
       "│          ┆       ┆              ┆              ┆   ┆              ┆         ┆      ┆ great t…    │\n",
       "│ 15147497 ┆ story ┆ AlexeyBrin   ┆ 2017-09-01   ┆ … ┆ [15148786,   ┆ null    ┆ null ┆ Been        │\n",
       "│          ┆       ┆              ┆ 11:21:56     ┆   ┆ 15148659, …  ┆         ┆      ┆ running     │\n",
       "│          ┆       ┆              ┆              ┆   ┆ 15150049]    ┆         ┆      ┆ Linux since │\n",
       "│          ┆       ┆              ┆              ┆   ┆              ┆         ┆      ┆ 2002.<p…    │\n",
       "│ 28218892 ┆ story ┆ gamontserrat ┆ 2021-08-18   ┆ … ┆ [28219635,   ┆ null    ┆ null ┆ I miss the  │\n",
       "│          ┆       ┆              ┆ 08:39:24     ┆   ┆ 28219619]    ┆         ┆      ┆ solar       │\n",
       "│          ┆       ┆              ┆              ┆   ┆              ┆         ┆      ┆ powered     │\n",
       "│          ┆       ┆              ┆              ┆   ┆              ┆         ┆      ┆ bins wi…    │\n",
       "│ 21961416 ┆ story ┆ Decabytes    ┆ 2020-01-05   ┆ … ┆ [21962017,   ┆ null    ┆ null ┆ Active disc │\n",
       "│          ┆       ┆              ┆ 14:44:26     ┆   ┆ 21962153, …  ┆         ┆      ┆ ussion:<p>< │\n",
       "│          ┆       ┆              ┆              ┆   ┆ 21962130]    ┆         ┆      ┆ a href=\"ht… │\n",
       "└──────────┴───────┴──────────────┴──────────────┴───┴──────────────┴─────────┴──────┴─────────────┘"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "training_stories = stories.sample(3000, seed=42)\n",
    "\n",
    "training_stories.head(5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<OpenAIObject chat.completion id=chatcmpl-8GVUWw34KnXCbP02rpghPdEIu3UPa at 0x7f411659fc90> JSON: {\n",
       "  \"id\": \"chatcmpl-8GVUWw34KnXCbP02rpghPdEIu3UPa\",\n",
       "  \"object\": \"chat.completion\",\n",
       "  \"created\": 1698944992,\n",
       "  \"model\": \"gpt-4-0613\",\n",
       "  \"choices\": [\n",
       "    {\n",
       "      \"index\": 0,\n",
       "      \"message\": {\n",
       "        \"role\": \"assistant\",\n",
       "        \"content\": null,\n",
       "        \"function_call\": {\n",
       "          \"name\": \"classify\",\n",
       "          \"arguments\": \"{\\n\\\"ai_ml\\\": false,\\n\\\"crypto\\\": false,\\n\\\"remote_work\\\": false,\\n\\\"rust\\\": false\\n}\"\n",
       "        }\n",
       "      },\n",
       "      \"finish_reason\": \"stop\"\n",
       "    }\n",
       "  ],\n",
       "  \"usage\": {\n",
       "    \"prompt_tokens\": 465,\n",
       "    \"completion_tokens\": 23,\n",
       "    \"total_tokens\": 488\n",
       "  },\n",
       "  \"openpipe\": {\n",
       "    \"cache_status\": \"SKIP\"\n",
       "  }\n",
       "}"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from joblib import Memory, Parallel, delayed\n",
    "import openpipe\n",
    "import os\n",
    "from dotenv import load_dotenv\n",
    "import json\n",
    "\n",
    "load_dotenv()\n",
    "\n",
    "openpipe.configure_openpipe(api_key=os.getenv(\"OPENPIPE_API_KEY\"))\n",
    "\n",
    "openpipe.openai.api_key = os.getenv(\"OPENAI_API_KEY\")\n",
    "\n",
    "memory = Memory(\"/workspace/cache\", verbose=0)\n",
    "\n",
    "\n",
    "@memory.cache\n",
    "def classify_story(row):\n",
    "    resp = openpipe.openai.ChatCompletion.create(\n",
    "        model=\"gpt-4\",\n",
    "        messages=[\n",
    "            {\n",
    "                \"role\": \"system\",\n",
    "                \"content\": \"Determine whether the HN story is likely related to any of the given subjects. Multiple subjects may apply.\",\n",
    "            },\n",
    "            {\n",
    "                \"role\": \"user\",\n",
    "                \"content\": f\"Headline: {row['title']}\\nURL: {row['url']}\\nTop comment: {row['top_comment']}\",\n",
    "            },\n",
    "        ],\n",
    "        functions=[\n",
    "            {\n",
    "                \"name\": \"classify\",\n",
    "                \"description\": \"Important: only return true for a label if the story is likely related to the subject. If it isn't related, or if you aren't sure, return false.\",\n",
    "                \"parameters\": {\n",
    "                    \"type\": \"object\",\n",
    "                    \"properties\": {\n",
    "                        \"ai_ml\": {\n",
    "                            \"type\": \"boolean\",\n",
    "                            \"description\": \"Related to AI, ML, AGI, etc.\",\n",
    "                        },\n",
    "                        \"crypto\": {\n",
    "                            \"type\": \"boolean\",\n",
    "                            \"description\": \"Related to crypto, blockchain, etc.\",\n",
    "                        },\n",
    "                        \"remote_work\": {\n",
    "                            \"type\": \"boolean\",\n",
    "                            \"description\": \"Related to remote work, WFH, etc.\",\n",
    "                        },\n",
    "                        \"rust\": {\n",
    "                            \"type\": \"boolean\",\n",
    "                            \"description\": \"Related to the Rust language.\",\n",
    "                        },\n",
    "                    },\n",
    "                    \"required\": [\"ai_ml\", \"crypto\", \"remote_work\", \"rust\"],\n",
    "                },\n",
    "            }\n",
    "        ],\n",
    "        openpipe={\"tags\": {\"prompt_id\": \"classify_hn_story_v4\"}},\n",
    "        function_call={\"name\": \"classify\"},\n",
    "    )\n",
    "    json.loads(resp.choices[0].message.function_call.arguments)\n",
    "    return resp\n",
    "\n",
    "\n",
    "classify_story(training_stories[2].to_dicts()[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "  0%|          | 0/3000 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "  0%|          | 1/3000 [00:00<15:07,  3.31it/s]Exception ignored in: <function _releaseLock at 0x7f42521681f0>\n",
      "Traceback (most recent call last):\n",
      "  File \"/usr/lib/python3.10/logging/__init__.py\", line 228, in _releaseLock\n",
      "    def _releaseLock():\n",
      "KeyboardInterrupt: \n",
      "  2%|▏         | 60/3000 [00:09<06:25,  7.63it/s]"
     ]
    }
   ],
   "source": [
    "from joblib import Parallel, delayed\n",
    "import tqdm\n",
    "\n",
    "\n",
    "def process_input(row):\n",
    "    try:\n",
    "        output = classify_story(row)\n",
    "    except Exception as e:\n",
    "        print(e)\n",
    "    row[\"tags\"] = json.loads(output.choices[0].message[\"function_call\"][\"arguments\"])\n",
    "\n",
    "    row[\"prompt_tokens\"] = output.usage.prompt_tokens\n",
    "    row[\"completion_tokens\"] = output.usage.completion_tokens\n",
    "    return row\n",
    "\n",
    "\n",
    "# Parallelize using joblib\n",
    "results = Parallel(n_jobs=30)(\n",
    "    delayed(process_input)(row) for row in tqdm.tqdm(training_stories.rows(named=True))\n",
    ")\n",
    "\n",
    "results = pl.DataFrame(results).unnest(\"tags\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div><style>\n",
       ".dataframe > thead > tr > th,\n",
       ".dataframe > tbody > tr > td {\n",
       "  text-align: right;\n",
       "  white-space: pre-wrap;\n",
       "}\n",
       "</style>\n",
       "<small>shape: (2, 2)</small><table border=\"1\" class=\"dataframe\"><thead><tr><th>rust</th><th>counts</th></tr><tr><td>bool</td><td>u32</td></tr></thead><tbody><tr><td>false</td><td>588</td></tr><tr><td>true</td><td>12</td></tr></tbody></table></div>"
      ],
      "text/plain": [
       "shape: (2, 2)\n",
       "┌───────┬────────┐\n",
       "│ rust  ┆ counts │\n",
       "│ ---   ┆ ---    │\n",
       "│ bool  ┆ u32    │\n",
       "╞═══════╪════════╡\n",
       "│ false ┆ 588    │\n",
       "│ true  ┆ 12     │\n",
       "└───────┴────────┘"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "results[\"rust\"].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\n",
      "  \"id\": \"adcc61d4-4d91-41c0-9885-f2bc9f60d382\",\n",
      "  \"object\": \"chat.completion\",\n",
      "  \"created\": 1697603891961,\n",
      "  \"model\": \"openpipe:classify-hn-headlines\",\n",
      "  \"choices\": [\n",
      "    {\n",
      "      \"finish_reason\": \"stop\",\n",
      "      \"index\": 0,\n",
      "      \"message\": {\n",
      "        \"role\": \"assistant\",\n",
      "        \"content\": null,\n",
      "        \"function_call\": {\n",
      "          \"name\": \"classify\",\n",
      "          \"arguments\": \"{\\n\\\"rust\\\": true\\n}\"\n",
      "        }\n",
      "      }\n",
      "    }\n",
      "  ],\n",
      "  \"usage\": {\n",
      "    \"prompt_tokens\": 110,\n",
      "    \"completion_tokens\": 16,\n",
      "    \"total_tokens\": 126\n",
      "  }\n",
      "}\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<OpenAIObject chat.completion id=adcc61d4-4d91-41c0-9885-f2bc9f60d382 at 0x7f51ef189ee0> JSON: {\n",
       "  \"id\": \"adcc61d4-4d91-41c0-9885-f2bc9f60d382\",\n",
       "  \"object\": \"chat.completion\",\n",
       "  \"created\": 1697603891961,\n",
       "  \"model\": \"openpipe:classify-hn-headlines\",\n",
       "  \"choices\": [\n",
       "    {\n",
       "      \"finish_reason\": \"stop\",\n",
       "      \"index\": 0,\n",
       "      \"message\": {\n",
       "        \"role\": \"assistant\",\n",
       "        \"content\": null,\n",
       "        \"function_call\": {\n",
       "          \"name\": \"classify\",\n",
       "          \"arguments\": \"{\\n\\\"rust\\\": true\\n}\"\n",
       "        }\n",
       "      }\n",
       "    }\n",
       "  ],\n",
       "  \"usage\": {\n",
       "    \"prompt_tokens\": 110,\n",
       "    \"completion_tokens\": 16,\n",
       "    \"total_tokens\": 126\n",
       "  },\n",
       "  \"openpipe\": {\n",
       "    \"cache_status\": \"SKIP\"\n",
       "  }\n",
       "}"
      ]
     },
     "execution_count": 1,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import openpipe\n",
    "from dotenv import load_dotenv\n",
    "import os\n",
    "\n",
    "load_dotenv()\n",
    "\n",
    "# openpipe.configure_openpipe(\n",
    "#     api_key=os.getenv(\"OPENPIPE_API_KEY\"), base_url=\"http://localhost:3000/api/v1\"\n",
    "# )\n",
    "openpipe.configure_openpipe(api_key=os.getenv(\"OPENPIPE_API_KEY\"))\n",
    "\n",
    "await openpipe.openai.ChatCompletion.acreate(\n",
    "    model=\"openpipe:classify-hn-stories\",\n",
    "    messages=[\n",
    "        {\n",
    "            \"role\": \"system\",\n",
    "            \"content\": \"This is an HN headline and top comment. Your job is to determine whether the story is likely related to any of the given subjects. Multiple subjects may apply.\",\n",
    "        },\n",
    "        {\n",
    "            \"role\": \"user\",\n",
    "            \"content\": \"Headline: A new Rust compiler is released\\nURL: https://rust-lang.org\\nTop comment: This is a top comment about Rust.\",\n",
    "        },\n",
    "    ],\n",
    "    function_call={\"name\": \"classify\"},\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "ename": "NameError",
     "evalue": "name 'training_stories' is not defined",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
      "\u001b[1;32m/workspace/opipe-experiments/hn-analysis/generate-story-classification-training-data.ipynb Cell 13\u001b[0m line \u001b[0;36m6\n\u001b[1;32m     <a href='vscode-notebook-cell://ssh-remote%2Bdev1/workspace/opipe-experiments/hn-analysis/generate-story-classification-training-data.ipynb#X15sdnNjb2RlLXJlbW90ZQ%3D%3D?line=55'>56</a>\u001b[0m     json\u001b[39m.\u001b[39mloads(resp\u001b[39m.\u001b[39mchoices[\u001b[39m0\u001b[39m]\u001b[39m.\u001b[39mmessage\u001b[39m.\u001b[39mfunction_call\u001b[39m.\u001b[39marguments)\n\u001b[1;32m     <a href='vscode-notebook-cell://ssh-remote%2Bdev1/workspace/opipe-experiments/hn-analysis/generate-story-classification-training-data.ipynb#X15sdnNjb2RlLXJlbW90ZQ%3D%3D?line=56'>57</a>\u001b[0m     \u001b[39mreturn\u001b[39;00m resp\n\u001b[0;32m---> <a href='vscode-notebook-cell://ssh-remote%2Bdev1/workspace/opipe-experiments/hn-analysis/generate-story-classification-training-data.ipynb#X15sdnNjb2RlLXJlbW90ZQ%3D%3D?line=59'>60</a>\u001b[0m classify_story_with_mistral(training_stories[\u001b[39m0\u001b[39m]\u001b[39m.\u001b[39mto_dicts()[\u001b[39m0\u001b[39m])\n",
      "\u001b[0;31mNameError\u001b[0m: name 'training_stories' is not defined"
     ]
    }
   ],
   "source": [
    "import os\n",
    "from dotenv import load_dotenv\n",
    "import json\n",
    "\n",
    "load_dotenv()\n",
    "\n",
    "# configure_openpipe(api_key=os.getenv(\"OPENPIPE_API_KEY\"))\n",
    "\n",
    "# openai.api_key = os.getenv(\"OPENAI_API_KEY\")\n",
    "\n",
    "\n",
    "# @memory.cache\n",
    "def classify_story_with_mistral(row):\n",
    "    resp = openpipe.openai.ChatCompletion.create(\n",
    "        model=\"openpipe:classify-hn-headlines\",\n",
    "        messages=[\n",
    "            {\n",
    "                \"role\": \"system\",\n",
    "                \"content\": \"This is an HN headline and top comment. Your job is to determine whether the story is likely related to any of the given subjects. Multiple subjects may apply.\",\n",
    "            },\n",
    "            {\n",
    "                \"role\": \"user\",\n",
    "                \"content\": f\"Headline: {row['title']}\\nURL: {row['url']}\\nTop comment: {row['top_comment']}\",\n",
    "            },\n",
    "        ],\n",
    "        functions=[\n",
    "            {\n",
    "                \"name\": \"classify\",\n",
    "                \"description\": \"Important: only return true for a label if the story is likely related to the subject. If it isn't related, or if you aren't sure, return false.\",\n",
    "                \"parameters\": {\n",
    "                    \"type\": \"object\",\n",
    "                    \"properties\": {\n",
    "                        \"ai_ml\": {\n",
    "                            \"type\": \"boolean\",\n",
    "                            \"description\": \"Related to AI, ML, AGI, etc.\",\n",
    "                        },\n",
    "                        \"crypto\": {\n",
    "                            \"type\": \"boolean\",\n",
    "                            \"description\": \"Related to crypto, blockchain, etc.\",\n",
    "                        },\n",
    "                        \"remote_work\": {\n",
    "                            \"type\": \"boolean\",\n",
    "                            \"description\": \"Related to remote work, WFH, etc.\",\n",
    "                        },\n",
    "                        \"rust\": {\n",
    "                            \"type\": \"boolean\",\n",
    "                            \"description\": \"Related to the Rust language.\",\n",
    "                        },\n",
    "                    },\n",
    "                    \"requred\": [\"ai_ml\", \"crypto\", \"remote_work\", \"rust\"],\n",
    "                },\n",
    "            }\n",
    "        ],\n",
    "        function_call={\"name\": \"classify\"},\n",
    "    )\n",
    "    json.loads(resp.choices[0].message.function_call.arguments)\n",
    "    return resp\n",
    "\n",
    "\n",
    "classify_story_with_mistral(training_stories[0].to_dicts()[0])"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "hn-analysis-9rBFKxKV-py3.10",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}