llama2-finetune.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "45c3ef31",
   "metadata": {},
   "source": [
    "# Our process for fine-tuning Llama 2 with custom data\n",
    "\n",
    "We successfully trained LLaMA-2-7b-hf on a custom dataset to perform a single, routine task (keyword extraction). We thought this would be easy, but it was WAY HARDER than we expected. We went down a lot of wrong turns, and many of the notebooks/code examples either didn't work or were improperly documented. \n",
    "\n",
    "**Much of the code in this notebook is borrowed from other walkthroughs. There are a few key changes that took us a while to figure out, and so we were inspired to share.**"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b3ad7352",
   "metadata": {},
   "source": [
    "# What you'll need...\n",
    "\n",
    "1. About **10K inputs and outputs** you'd like to train the model on. We generated these using some python scripts and openai's gpt3.5. For us, the average total token count of each input/output pair was ~300, and **no pair was longer than 800 tokens**.\n",
    "2. A **runpod account**, and about **~$25**. That's what it cost us, once we had everything figured out. \n",
    "3. **ChatGPT** open in another tab. This notebook works for us as of **September 2023** but there's no gaurantee it'll work for you. With any luck, you (and ChatGPT) should be able to overcome any obstacles that arise. \n",
    "4. About **2 hours** of setup and the patience to wait **5-8 hours** for results."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0bb70ea5",
   "metadata": {},
   "source": [
    "# Getting set up in Runpod"
   ]
  },
  {
   "attachments": {
    "instanceselect.png": {
     "image/png": "iVBORw0KGgoAAAANSUhEUgAAATsAAADiCAYAAAAxmmCUAAABamlDQ1BrQ0dDb2xvclNwYWNlSVRVUl8yMDIwAAAokWNgYNJJLCjIYWFgYMjNKykKcndSiIiMUmB/wMDIwMogwGDJoJ6YXFzgGBDgA1TCAKNRwbdrQNVAcFkXZFY4t3TZsQUBq+bnBPfu07X7gKkeBXClpBYnA+k/QJyfXFBUwsDAmANkK5eXFIDYE4BskSKgo4DsJSB2OoS9A8ROgrDPgNWEBDkD2Q+AbIXkjMQUBgYmkPk6SUji6UhsqL0gIBqUmqyn4BkSqhuk4BSiZ2RgZKBrSMDdJIOS1IoSEO2cX1BZlJmeUaLgCAytVAXPvGQ9HQUjA0MzBgZQ2ENUfw4Ehymj2BmE2JpMBgb3/P///zcixLSAYmseMzCwr0WIKTAzMHCvZWA4/a8gsSgRLAQUYWCyeMLA8AIYH+ICDAwgtsie4jRjI7A8I48TAwPrvf//P6sBzZrMwPB3wv//vxf9//93MVDzHQaGA3kAWuRoHHs2YjEAAABWZVhJZk1NACoAAAAIAAGHaQAEAAAAAQAAABoAAAAAAAOShgAHAAAAEgAAAESgAgAEAAAAAQAAATugAwAEAAAAAQAAAOIAAAAAQVNDSUkAAABTY3JlZW5zaG90nEThkwAAAdZpVFh0WE1MOmNvbS5hZG9iZS54bXAAAAAAADx4OnhtcG1ldGEgeG1sbnM6eD0iYWRvYmU6bnM6bWV0YS8iIHg6eG1wdGs9IlhNUCBDb3JlIDYuMC4wIj4KICAgPHJkZjpSREYgeG1sbnM6cmRmPSJodHRwOi8vd3d3LnczLm9yZy8xOTk5LzAyLzIyLXJkZi1zeW50YXgtbnMjIj4KICAgICAgPHJkZjpEZXNjcmlwdGlvbiByZGY6YWJvdXQ9IiIKICAgICAgICAgICAgeG1sbnM6ZXhpZj0iaHR0cDovL25zLmFkb2JlLmNvbS9leGlmLzEuMC8iPgogICAgICAgICA8ZXhpZjpQaXhlbFlEaW1lbnNpb24+MjI2PC9leGlmOlBpeGVsWURpbWVuc2lvbj4KICAgICAgICAgPGV4aWY6UGl4ZWxYRGltZW5zaW9uPjMxNTwvZXhpZjpQaXhlbFhEaW1lbnNpb24+CiAgICAgICAgIDxleGlmOlVzZXJDb21tZW50PlNjcmVlbnNob3Q8L2V4aWY6VXNlckNvbW1lbnQ+CiAgICAgIDwvcmRmOkRlc2NyaXB0aW9uPgogICA8L3JkZjpSREY+CjwveDp4bXBtZXRhPgr1FOGXAAA8CElEQVR4Ae1dB3gUVdc+aRBqKCn03nvoRaT3JkUQRUVF5bPhZ+8/2Bt2rKigYkMBsVCkd+lNegskJJAQUoBUkv++Z787mV12U8hkkyzn8ISZvXPn3jvv7Lxzzrl3z/GqXr1DZmxsDIkIAoKAIOCJCFSqFEiZmenkVapU3UxPvEC5JkFAEBAENALBwVXIW3+QrSAgCAgCnopAXNx5ITtPvblyXYKAIJCFQGpqipBdFhyyJwgIAp6MgJixnnx35doEAUHAQEDIzoBCdgQBQcCTERCy8+S7K9cmCAgCBgJCdgYUsiMICAKejICQnSffXbk2QUAQMBAQsjOgkB1BQBDwZASE7Dz57sq1CQKCgIGAkJ0BhewIAoKAJyMgZOfJd1euTRAQBAwEhOwMKGRHEBAEPBkB34K6uPL+ValMiSAq6VuOfL39C6obaVcQEASKOQLpGcmUkp5IF1OjKSE5ssCuxvIQTyC5SqXrqcFfoAspZyg5PZ7SLydTpvonIggIAoKAGQEv8iJfH3/y9w2gsiVDlHJUlmIvHSsQ0rOU7ILKNuZBx1w8TElp583XJPuCgCAgCOSIQCm/ihRYpiErSdEXDuZYPy8VLCM7EB3M1aiE3aLF5eUOSF1BQBCwQwDaXpXyrQjmrZWEZ8kEBUxXqKFCdHb3TD4IAoLAVSAAlxe4BJwCbrFKLCE7+OhguopfzqrbIu0IAtc2AuAScAq4xSrJN9mBeTEZIT46q26JtCMICAJAAJwCbrFKu8s32WF5CWZdRQQBQUAQsBoBcEuZEoGWNJtvssM6OiwvEREEBAFBwGoEwC0lfctb0my+yQ4zsFhHJyIICAKCgNUIgFt8vUta0qwlv6CwamIiIKAc3XzzKEpPT6cvvpiTrwsMCqqs2hpJBw4coSVLVl3Rlqu+vLy8qFmzhtSuXWuKi0ugzZt3UFTUWbvz/fz81PGW1KJFEzp27CRt2bKTEhMv2NVx/FCvXm1q27YlVa5ckcLCwmnNmk106VKSUS03/aJy3bq1qE2b5hQSEkgREWdo+/bdahtltFOnTk0aPnyA8Tk1NVXV2cN/wFVEEChOCNi4xcuSIVtCdlaMZMiQPjR16mNMBjExsfkmu9GjB9OUKZMoNjaOli1bQ5cvZxjDdNUXCOeTT16n/v17GHWxM3XqdPrmm7lcBrL69deZVKtWdaNOUlIyTZjwAO3YsdcoM++gv3femUZ+fllwHz0aRjfdNJnOnTtPuekX7Y0ffwO9+OIT5OOTpZBnZGTSK6+8T19//SN3CTJ8+OFJ5u55//TpMzRx4hQ6cuTEFcekQBC4FhDIemoK8WpHjx5CH374itLoLlNqapolIxk5cjC3U6lSBbruuk5Gm9n11bNnVya6PXv2U7duI2jSpMdYu3vyyfupfPly3Mbdd9/CRPfdd/OoQ4dB9NJL75G/f0l65pkpRh+OOy+88Ii6rlRFNg8rjXAgzZr1M9WvX1sR5Giumpt+O3YMpZdffpJOn46iRx+dRqGh/Wjy5CcpOvocPf/8wzRqlO16dd9z5/6hNNv76NZbH6Q335xBVasGE8YhIghcqwgUCbIrW7YM/fzzQurT50Z+ePXNqFmzGu3bt5oWL/7e0GaGDeuvTMdN9N57L+pqV2xhXjZsWNcoHzNmqLHvqi9UGDq0L9d77rk3KTLyDK1YsY4++uhrKlXK39D2QJYJCReUhvUOa2XQqJYtW8tmLcbrKNDCYFJv27abTdfz5+Pos8++4Wq6fm76hZaKlwE0yPnzF1F8fCItXbqatcO0tDS6554Jdl2Hh5+mTZu20/r1W+jTT7+lw4ePU6NG9e3qyAdBwGoEOndu67LJ7I65PMnCA0WC7ObO/Z2eeupVOx8WrvHUqdP0/vsz1UNaT2lF45h0nn12Cp0/H0/Tpk13CcPIkYP42Lhxk2nnzn+ZqEqXLsVlrvrCwVatmnHb0Oy0bNy4lXdbtWpKgYGV2Mxev34z+xV1nU2btvFuy5ZNdZGxhfm8Z88BRYatlCbYRpG2D9177218HAQIyalfjB1fFPgPgYlZ4P8bNWqSMrXfNhfb7Y8YMUBpknXUSyLMrlw+CAJWIoDv6Pfff8zuI8d28bLGscIkvCwnkuPo3PjZ0VEPH5aWmTPnEMjr0UcnU+PGDSg4uLIC83kmJV3HvIUmBe0rMvIsbd26i377bQk79IcN60c//bTQjlDRj7mvChXKE/yFZtGfK1QIIByHwM9mFnMdc7nef/zxl5Q29qXq/1NdRKtXb+LxoCCnfqtXr8LjPH78pHE+tEWYwmYx+/L++997CH9aoI2++uoH+qNsBQHLEYAlAeUExAbBPgSf8YfPqFNYUiTILruLh2YEH9XChbNozJghtG7dZvr9979dntKjR1flXytLGzduo06dQg2zGD48kF1OkplpH4rK29tGvObyjIysyQ605+WlFWT7c3EMBPTyy0+wX2/lyg0Ewurduxv16NGZ/Wy//PIHqpG5fXw295uWpmdRs14C8PO98cazqGpIy5a9jX1ogJil9fb25tllYAKtGH48x76Mk2RHEMgnAmaC001potPHdLm7t0We7ADImTPRamnHRSaxffsOZYvRqFE2E3bAgB6EPy0dO7ahGjWqUni46+CA0NCqVAlmLUoTQmBgZW4CY9AaHeqYJSioEn+Mioo2F/N+t24dqX371jRjxiyaPt2m2b3xxgxaufIXNWt6N4HscuoXS19AeGY/5KpVG5i40Mm0aY8TlpxgVlgL2v3ww6/4o6+vr/Lbva5I9jplMjelXbv26WqyFQQsR0CTmlnD02WWd5aHBrVKkodT3F912rTHmOjQ81133UzNmzdyOgjMmPbv35NOnoxglRkA4w8OfYj25Tk9WRVCxcb6u+7ds2ZvYf5CsJYOvsKDB48StCo9OwszeOjQfry0ZceOPVzX/B/MTQjGpAUTCjCzg4NtP4PJqd/k5BRavnwta6rXX9+Zm8EsLM47efK0IroatHv3fnLUOHV/WF+3davNPwhzXEQQKGgE9LOntwXdX27aL/KaHYhn0KDebL5imcdff32nNKSpNGTIrYpgLttdI0jH19eHfvxxAc9A6oMlS5agwYP7KDN4qKHt6GPmLfyD48YNp7fffoFJEoQ2adItvDAZM64QaEsfffQKzZkzgz7//DtFfF2odetmvA4PZOgomFSAYPkKyCgiIpLHgkXJmCmF5Kbfl19+n7p0aa/WH76ltME/2efXoUNrNoWxyHnGjK+5Lf1fjRrV2Bns7e3DM8V33HGT8lcmu1wLqM+TrSBgFQIguqIk+Q7e2SCwDx2JWW7ZNa1du4BATh07DibMQv7990+sRfXtO5bN2SeeuF+tL7tVEdKn9PHHs+z6nTv3C36wsUYOS0fMgqUqw4f3V5MXk4wH3tyXrovJjddee4ZJE2X4FcW4cffy0g1d5/XXn6GxY4frj7R//2FFpHfbmZHGQbXzyCP3qhnYW+0WFZ85E8OLfKEpQnLTL5bUzJw5nSdp+CT1HzS8t976hM1hlPXq1Y2+/PLKmeqUlFS6/fYpPKOrz5WtIFAcELCKY4oc2YWGtmCHP34GhQXBWHYCXxnWiUFAhKiDh9fxFwuY1nZWjvNCQoLUT61qqp9qRRhEaO4LdbTAt9enT3e1ni6R19A5+ykYtDks9MUvIVav3mD3Cw3djnlrWz7Sjn1rmCXG8hbtF9T1ctMv/G9du7anpk0bqr5PsIYHs1hLxYoBatbafj0d+tm//whfj64nW0GguCBQpMjuaMwKCdxZXL45Mk5BoBghgBDt9QN7KetxRb5Hne8JCsSJR3YgEUFAEBAErEYA3JKekWJJs/kmO+R7RKx4EUFAEBAErEYA3JKSnmBJs/kmOyS2Rb5HEUFAEBAErEYA3HIxNcaSZvNNdsjgjcS2yPcoIggIAoKAVQiAU8At4BgrJN9kh0EggzcS28KZKCIICAKCQH4RAJeAU8AtVoklZAfmRax4JLYVwrPq1kg7gsC1iQA4BFwCTrFKqwOSlpAdGkLmbszM1qjQQUxaACIiCAgCeUYApis4BFwCTrFS8r2o2HEwyPGIxLbI94g0aGBnJM2wKk+FY3/yWRAQBIovAtDisLwEs66YjICPDqarlRqdRsdystMNg/SQUxapFpGBTEQQEAQEAWcIQIvDEjas7CgIktN9FhjZ6Q5kKwgIAoJAUUDAMp9dUbgYGYMgIAgIAq4QELJzhYyUCwKCgEchIGTnUbdTLkYQEARcISBk5woZKRcEBAGPQkDIzqNup1yMICAIuEJAyM4VMlIuCAgCHoWAkJ1H3U65GEFAEHCFgJCdK2SkXBAQBDwKASE7j7qdcjGCgCDgCgEhO1fISLkgIAh4FAJCdh51O+ViBAFBwBUCQnaukJFyQUAQ8CgEfD3qauRiBAFBINcI+Hj5UcOgPlSxdB1KTouno+dWq6gjp3N9fnGrKJpdcbtjMl5BwCIEmlcdocKvlaRdET/T6YRd1KHWRPLzKWVR60WvGR8/v4pT3TmsSpUqUNWqIZSUlELp6elOu65QIYAqVqxAFy5cdHrcWWFAQDmqWbM6+fr60qVLSVdUqVIlmMqWLcN/pUr5U3JyCmVmZl5RTxcEBlbi3bS0K8dYpkxpQn/oB21WqlTRaPvy5csurwsN4tr8/Uty/7ovvcUYMaa0tDRdZLctV66swiWALl68ZFcuHwSBq0Ggachg2hu1kC6lxXIsueCyjdX2HCWlnbdrrlzJKlS5TD0qraII16rYiby8vFgTrF2pMwWXbaIC9CZQ2mXbd9Lby4eqV2hHtVS04fL+1elC6lm6nJFG5VRgzuoBoRSXdJLbLu9fTYVeb0HxSeF2fRXkB7eZsf7+/nTLLSMJD3R8fCKBTJYuXUUbNmw1rq9EiRI0ceJYPgbSuHw5g+bMmUeRkWeMOo47IJ2xY4dTnTo1ud3SpUvRuXPnaf78vyg83JaVyNvbmx588E5FsMmqzctMiD4+PrRjxx5auHCpU9Lr1asbdzV37u+OXdLw4QMUWSWrPhZRly7tqHv3Ttw2KpYqVYrOn4+jX3/9k06ejLA7F1+Se++doMjwMr3zzqeq36zDeoxHjpygr7/+MeuAaW/ChFEUEhJMU6e+bSqVXUHg6hCIvnCYagS0pSMxKxUxVVXpFCo4JZ8yJQOpcXB/pf3tVuQVTa2qjVH1Iijm4mG6nJlOHWvdSSuPvKUGkUktqo6kEj5lKCJ+u4pYXpfa1biVNp741DgP5Bej+m1R9QY6HL386gZ+lWe5jey6d+9IPj6+9MYbM1jzqVYthO67byIdOHCEYmPjePiDBvVi0nj99Y+YgPr06U5jxgyljz+exSTleI2+vj40adItFBMTS2+//YkimXgqUcKPBgzoRbffPpZmzPia4uKyEuzOnv0znTp1Wr2ZiGrUqEYTJoym06fP0JYtOx2bVkS4l26+eST5+fnZaVrQypo0qa8I6SfjnKNHwwhtQ1C/c+e2irTH0ZtvzrDT4OrXr00gNX9/X6pVqwaFhV35VkOdypUrMmEbHagd4IUxO9M0zfVkXxDILQLHlI+uU+1JVK9yd/VM+NCeyPkq90OK09MRSfjQ2aV8LEBpZRmZl+lE7Ab+XFVpaCBL+PsOnFnE+SNw/EzifurX+Hk2ldHu3sgFTJRhSkNELlirc0w4Hbip0G0+u5CQIDp8+Jhh4oFkQHLQ9CAlS5akjh1DacmSVZSRkcFkt2rVeoKm1rBhXdOQs3Y7dWrLdX/8cQETHY6kpqbR778vZe0K7TkTaFQgvX37DikCqeqsCh09eoJSUlKpadMGdsebNWtMCQkXFFGdsivXH2CCrl37D5MzzGqzhIa2pF279tHevQcoNLSF+ZCxHxUVzTgYBf/bwbVERp51LJbPgsBVIeDt5as0sruUdrWMlhycRmuOvkt1K11HQWUbOW0PZKcFOWXMn1OVCeun8khAkEOiWZVh1LXOf+i6eg+oEi/y8S7Bx+KTI+h8Uhg1CR5I+6P+4DJ3/uc2sjty5LgijoaK1GwXDrOzfPlydPy4zYYPDKxIiYkX6ezZGOP6Ye7BhNX+M+PA/3bq16/DmiFIyVE+/ni2MpNXOxYbn+G3q1u3FkVEOE/AC9/Zrl3/UuvWzY1zsNO6dTPaufNfOxPUroL6AM0RGhxMZi0w0Zs1a0S7d+9jwmvZsimb0/q43m7atI3atm1pdwwvAowDx0QEASsQgA8tkzKUublTfZcvs6YVHreVQso2vermoR12qn23SphznLacmqUI9D3lr8t6NjH7W1kl47qkfIKVy9grEVfdaR5OdJsZu2mTsuGVI/+ppx5UmlEiE91XX/1g+LpAaM4mJGCauiI7mHsgDy3wiWHyQAvIBn46LYMG9WFfG/x11atXUeSxnTZvvtKE1fVhyt5//0Tlh/PndjBBUK9eLeXnW6Kr8BbaJyYqIOXLlyVonNAwoT1qad4cGmEi+xFBhpgggda4Z88BXYW3x46F8QQEyBA+RQi0wOjoGMMHyYXynyCQDwRgRiIZls38jCRoetDqohL/vepWfVQbXl7eyjw9pCYsklSS6wZ2s7sNgnrTuYtH6Xjsevbznbt4hKAVukvcRnYggDZtmtPy5WuVHy2eGjWqr3xio+jTT2ezXw1anJ+f8+HgmDNJSUmxO6dcuTL05JNQnW2CCYLPPvtWf1Sm6XHlC4tTWpcXRUWdpa5d27MmaSZMo7LaOXMmWv3FUMuWTZgUQUCnTkVe4U+DKfz00w8ap0LT/Oijr+x8fSAsmLAQmNHoE2atI9nh+D//7FCE2cYgO5iw69ZtxiERQcASBOBD2xnxI4VWv1kRlJciOz+KSthDp+K2XHX7aPP4uXXUo/4jPEN7IeWsIjPbigqQavWANrT22AdMhOHx26hpyBDadXruVfeX1xOds0teW8mhPky6gQN7EXxrBw8e5dp79x6k8eNHUo8eXei335YwgVSoUJ5NQPMsJTSlY8dspq5jN5htrVatiirexYfgS3v22dd5f9Cg3qxlmc85dOiYnbYF/9iAAT3stENzfexDu2vVqjmTHUzYbdtsfZnrwb+nJyj69r2ezXXzxEhAQHnCxAP+evbswqdCu4RAE3VcSoI++/fvwUt0YPYDgz179qmJC9tyGD5R/hME8okANLDVF6ZzqtPLiqhc5XaOStiriHCv0duBs4uNfexsDvvS+HwkZoXS3NYSTFaz1gYf34rDbxj13D0Ti47d4rPDAwutTc+66is+dy6WYBpCsI83TN26tfVh9lvVq1dbaV/RRpl5B7OZ8INh7ZlZMEvbsGE95Q90Pomg68bGnuf+QcauBH67WrWqKfO1tppMCXKqiWGJjP5bs2YTm7QdOrQ2moRGi+Uwb775MU2f/hn/YR9LcBx9gjgJy1qg+XXqFMqTFdu375FZWANN2bEaAeRtdUV0V9MX1tWZie5q2iiIc1w/5Rb2Br9ZWFiE0qJ68mJckAsmB9q1a0X79x/mnrCkYsWK9TR0aF/WaOAnu+GGgcpPdVppdmFORwOz8PDh47zMBD4xaEm1alVX6/lGc33MepoFa+DgW8PECOpjiQrWtWH215VA60KdsWOH8WSI2Qfo7BwcX716I/XufZ1aBmObjIEJi8mF+PgEuz+UuZqV3bx5hzL7W6hxNlJa5Q5nXUmZICAI5AEBt5ixGM+cOb/SqFGD6eGH72EtDzOva9b8o8zC3cZw16zZyGR45503qTp+vPxj3rxF2c58YvFwv349COZjcHBl/lUDzN4vv/z+il9S3H77jdwXZlqxNg+a4V9/LTf6d7WDiYLGjeuzSeuqjrl848ZtarFxe15sfPDgEV43t3v3fnMV3of2BnMVy3Kio8/ZHY+IiFL+wrO8/AVjFREEBIH8IeBVqlRd5S53n2ByAL+mcPaTLj0K1MHPvjCjmRex/QxMqeRuvaK8jFDqCgKCQGEh4HayK6wLlX4FAUHg2kbALT67axtiuXpBQBAoCggI2RWFuyBjEAQEgQJHQMiuwCGWDgQBQaAoICBkVxTugoxBEBAEChwBIbsCh1g6EAQEgaKAgJBdUbgLMgZBQBAocASE7AocYulAEBAEigICQnZF4S7IGAQBQaDAERCyK3CIpQNBQBAoCggI2RWFuyBjEAQEgQJHQMiuwCGWDgQBQaAoIOD2vLEIAoAwTIg8gtDkjoLwT8ikhajDCK+U2x/1Z5c3FqGWgoIqG7ldEVEFMeOyE8dzfHy8OQKJq3MQmh25bh0DcaI+cuXq0O7Ozke8P4wPEY6zCzfleK4tpJXK9KRCVDnLw+FYP7efMRYkQkIgBscctrgWBCNFqCxcM6JFZ2TYR17Auc6uBZnZgFF2QSByO0apJwjkFQG3hXjCwBAGHRGLEcQTeSUQj27evL+MxDR4SJA3Fg8ZSA/BPJFDNbsQR7nJG1uzZjVC2Cid4wIJbPAQL168ymnkYYzV2Tl4sFesWMdh01HHLMOG9VPh25uqwJyfGpnO9HGkg0QYpzfe+MhpJJeePbvS9dd3ViHqv7GLpKzPd9wih8XAgb051wVCQyEXB1JSIletOcmP43k5fUain759uysSK61ybiRxfEDk6Vi8eIVBaLgW3Cfgh3uE1JUnTpxSUah/4zwdKEOOXmfX0qBBXRoypC/jkNNY5LggYDUCbiM7RPq9/vou9MEHXzJ5IYAm8sa2b99akcd2vq6bbhpB69dv4VSEKECst1GjhtDnn2flkTADkJe8sQgO+tprH/LpOA8pERGQEykRXZGp+RyEnapfv64KJX8D10euWC0gz6ZNG6mYdLEceXjVqg36kLGFVtOqVTPautU+rDvGgiCmeZF69epwBrL33/+CiRUa1uTJt3GwT2dh43PTNuL1Ifn3vHl/qhSThznlZdWqISoQ6ijWHBcvXmk0g+tDRGYINLzRo4fQuHEjaNasrFy6RmXZEQSKCAJu89l16dJOkdgmg1iQaevbb+caqROhIUBL2bIliwyQsrBGjSqs4TnD62rzxiKBDwJnguSqV3eeN9axP5hqyHuLyMjQUMzSokUT1dY5WrlyncvIw8j5ijDrjoKIyTDn8xK7D1oi2kPmNQjMQmhXCBvvTPCiQU4OswBvaGk6pP3IkYNVINNlCpf9Rm5fpLGE5o2EPzD9nQm0ZaSsRNY1xCAUEQSKKgJuI7vatWtw4hzkZhg5cpCK5NuOs4rpvLF42OfMmWfnS6tZszrnboB/z5nUr3/1eWNhilWoEMB5aZ217aoMmphjtjOdOWz//iOc0wImsKMgjwQS5jgm5e7YsS2HXc+L+YkEP/BrwrcGQR4PYAEidibI79GtWwc2d/XxJk0acl4NJAZCoqOyZUsz0enjeouQ+C+++M4Vvjt9HFv4DDlDldJ+RQSBooqAW17FyKQFU2vIkD5MFCdPhrN/C5rZzJnfG740M0jQOJD5a8EC+0xG5jp5yRsLkrrtNltYdpiUIIqff15oaJbmdvU+kgTpfLA4B6YeSGXlyvW6CmtGtWtXp7lzFzIhIKcGyM+cMxaV4e9D8h5cc3j4n3w+NDTkr0XIevjtcitI8Th//iL2jcH/CaygXR06ZMvc5tgOEvvA7IYZrcfeqlVTlex7L08UQbtFO84mjBzbcvyMe4sMcbhe7Wt1rCOfBYGigIBbyA4mE978SHUIgoHA73PnneN50gIPqlkwc3nbbWM5P8W+fYfMh+z2QSDmXLPZ5Y2F9qHzQPj5+bBWg4Q+MGXPno2xa9f8wZwPFuWzZ8+1S1aNpDgnToRzpjAch+kNX+Cffy6/YrIA+WDhW0PeCyTmgXmIpEB5nZ0EQSKfB3JdIDcuSBMJfpDBDBMVzgR5NEBKIDvg26hRPVqyZBVXhW/SlZnqrK0BKnESiBN+TLxw0O+sWT87qyplgkCRQcAtZIcHG0sRQARa4AP7999D1KJFY13EWzx00MCQxHrp0lV2xxw/5CVvLPqDJqMFvkE41rt378SzmLrcvIVpOW3adC7C5MSECaPYN2eug9lLyIsvPs5bPYvcpEkDdX0HuUz/h2s6ffoMJ8fGRAVSLF6NUx8JhrZs2almSW2TBugHhImZbldkh5cGJiBgvkOTgz9OJ/mJiIjk3LTQYvWMtR4zric4OJDralMbbSHjGdwLSCKemHhBV+cypJV05r/Di0m3YZwgO4KAmxBwm88uOjqG15uZrwvmlzmZNB6sm28eyebUL7/8nuMau/zmjYVGgjVjrgQEqfPBwkQMC4vgGWJdH+sFIR9++JWRD/attz5hbc11isTtSqNro4iumZpgiLvC3NVtZ7eFfy27HLzOzoWJCVKE+Yo/JOLWgrWBID749RylZcsmdMcd45jE9DGYrPAPIsWkmehwHASIl1CdOjV1dWOLMmiiIoJAYSDgNrJbvXoT9elzHZuPeOtjFrJDhzYq6bQtxSDWjkHTwpKUhQuXKlPLluMV2gYW9DqTvOSNRftoC38wvbDkpXPntjzD6qxtZ2VLlqxUmmgTY5IBhAYtBxqbOSfsunWb2UyEn9JRYLaivH//nk7X6znWd/b5wIGjbP5jaYg2JWHG6hy8zs5BGQgOy1wwWaRNel33hx8WsFmtUztijKGhLWnEiIG0fPk6noTQdXPabt26U/kgu6i+WvO1YgIFGjQ02e3bs1Jn5tSOHBcErETALWYsBgxSwOwn1m3hFwPQ6P74Y5lhdlWpEsIPA+o+9thkbAz57LNvXWoEuc0bC4LV/jf4qOCng89r7dp/jH5y2oHGAu0IC3pnzfqRJ1kWLFh0xWnQfHB98GshEbZZMJOLXLnw12HC4mpkxYq1ykz04YXS8L/BRYBx/f770mybO348jJeVQLty9BNi0uPbb3/hxc3IeQs/a1xcPN+jvBIUZp4h8BGOHDmQ94H3nDnzWRvkAvlPEHAzAm5PpYiJCjygjg+bFdeNdvEzMBcrVazoosi1gZ+MXbqU+5/V5eYCMMMKsoOvNb8CDRGTQ1cz05vfvuV8QcCMgNvJzty57AsCgoAg4C4EnDvD3NW79CMICAKCgJsQELJzE9DSjSAgCBQuAkJ2hYu/9C4ICAJuQkDIzk1ASzeCgCBQuAgI2RUu/tK7ICAIuAkBITs3AS3dCAKCQOEiIGRXuPhL74KAIOAmBITs3AS0dCMICAKFi4CQXeHiL70LAoKAmxAQsnMT0NKNICAIFC4CQnaFi7/0LggIAm5CQMjOTUBLN4KAIFC4CAjZFS7+0rsgIAi4CQEhOzcBLd0IAoJA4SIgZFe4+EvvgoAg4CYEhOzcBLR0IwgIAoWLgJBd4eIvvQsCgoCbEBCycxPQ0o0gIAgULgJCdoWLv/QuCAgCbkJAyM5NQEs3goCVCHTr1pHTVFrZpqe3VShkV6lSBU4ziAxWVgrarVevNv/VrVuTM2RZ2b6728I1IO2kiCBgRgC5j5966gFq3bq5yqnsYz5k7CNtqU7ibhQ67CDTX926tTiHs6tnsVy5MtS2bUuqUiXY4eysj8gJPGhQby5AInR//5JZB4vQnluzizVsWI+eeOI+AiDx8YmENIDIq/r66x9y7tP84nLnneNp7NjhdPHiRZU82ptTNh48eJTmzfuLVq/emN/m3X7+okXf08MPv0AHDx5xe9/SYdFDAN/pu+66mZPJ+/nZUj6fPx9PL774jkp6vo8HDAIbPrw/3X33BEpMvEjjx9vnYNZXhUTxb7zxHFWsWIFTXaK9559/00haj3pTptxNQ4b0oYiIKKpevQotWbKK3n338ysSpr/22jOcSH3ZsjW0YMHX9H//95bKiWwbj+6vKGzdptlVqRJEr776NP3zz3Z1sybRmDGT6N57H6fg4EB65pkphJtkhezcuZfbHznyTpow4QFaunS1Itj7qVu3DlY0L20IAoWGQK9e3ahXr640ceJDtGHDVrrttgdp0aLliuwe56TpGNjLLz+pCKov/fbbkmzH+fzz/+Vk7aNH36UUhHto/vxFiuz+a2iKI0YM5GfmjjseJvyNH/8fpUk2o8GDbRqcbhzJ55EMPq+J1PX57tzaXg9u6HHSpAmK7f+lmTO/N3pDZvpnn31dlU2nrl070Pr1mykoqLK6iTfSN9/MpRtuGERVqwYzkH/8scw4L7c7MTGx9OefyygtLY0Jdfjw2+ny5ct8erVqIeqL040aNapPJ06col9++Z3fhDjYrl0rNoUPHz5OAwf2ooSERJo793fWFm+4YSAFBJQnvMV05nucU6FCeR5v/fp1lGZ5ib+E+u2m2/v334M0YEAvZV770t9/r7X7gvj6+lC/fj2477i4BPrhh/loVkQQMBBo3rwxbd68g6KiorksPf0yzZ49l8LCIpTp6E8XLlzk4/jOX399F+rd+zrjXPMOXD01a1anxx9/iYszVVb5OXPmsQZXurQ/PwedO7dVhLmYy1Dp3Lnz6ju5gL/j5mcRY4qMPEOxsXFGFzCtx40bTk2aNFTnR9JPPy1UbV7g4yNHDqbTp6P4ue7UqS29/fYn3LZxcgHuuE2za9KkAa1Yse6KS4mLi2fSaNq0AR+DPwIP/SOP3EsxMecUkGeVBngbAaSrlWXL1jJRwQcGCQkJovfee4mJFWZ0/fq1acaM1wz/GHwdMAVGjRpEe/fup8aN67NWCj8JiC829rwyAZ5n1R7tIev9Bx+8wr6PjRu38s177bVnCWY7BO0NG9afbr11DB06dFRpsd7cHtrVApPhxhuH0b59h9SX9gJNm/Y4j1kfl60gsG7dZoJ/rEWLJgYY6enp/OIF0UGg0YEEs5OWLZuw2QtFYvLk2+k//7ldaWdNuR2YvhD4ilNSUu2aSUpKNr7z+kD79q1o69bd+iNvYWqXLl1a9bGflRhom1qgHd533+3qGpoSnhXHPnS9gti6RbMLCCjHTA5NyZkcPnyMOnYMNQ7Bf/Ddd78SNCGIt7cXq9Tz5/9l1MnLTkZGBh0/HsZa3JEjJ+ieeyYo83aVoWUuXrxSkd2rSovrzW8ztA1/4osvvqu+OOl8M+fMmaEIbgafh+PQCLt370w//riA63z22bfKRN9mfNEaNapLHTq0JlwbBJrf3Xc/yjcXb14QIK4ZPsUGDeryW3jixCkUHX2O6586Fam00Yd4X/4TBIDAjh17WAObNu0x9X0KIGhk338/jw4cyJtPNzCwEruP3nrrBSY4aIV4uc6a9RP9+uufDPaOHXvZqlm0aAVbKqVK+fPLGFu83C9dSuJ6sFq+/von3tf/bdiwhceJzwcOHKaPPnqVypUra2h34eGRytx+V1d329YtZJeWlq6cmpnKfHM+s4i3SGqq/VvEfAP37j3AbzSggrcabrZZoI4fOxZmLrpiHzc0OTmFy2Eyr1mzUWluWdoi3jD16tUyzjtxIpxJDAVnztjMhiNHssj61KnTTOA4npqapnwoW1izgxkLExlbmBdajh8/afcWwzVBw4RAw4PGp4kOZdA4RQQBMwIgN0y2/fXXcuWe+U65U8rRO+9Mo8WLVyjL4ktz1Wz38SzgBTt58uNKCTjFddev36ImCp/htqHB/fTTb6ztzZ79Ae3ff1hZKXXZ347vNY5DMFNbp04t1uC44H//ob4W7EPTxGyuNmUPHbIpALqOu7ZuITu8BU6dimDAYMM7Csw9M7nhuPatYR9ESWSbwADhaF8DjkGctWk7YvsfZFqzZjWe1cRECKbZg4IC+Y2l64GMjh41Eyb6zJ3gTTl9+v/xLNXOnfsoPPy0+oLYv23N14NWcU16UgZvPfj5zJKSksLtmctkXxAAAnhpY4Lio4++YnMTPm/MlMJKyI3g5X306HGD6HAOJhjwnMLdBK0OL/Ann3yFmjZtyP7r2bN/Ij8/P+VTbs0aJc4JDW3Jmhu+q2Zx/K6DpM3i+Nl8rCD33UJ2uABMjcO5D78DTEMt8KNBFcYbKzeCG52TFufYzu23j1P+v1jlGD3DNwrmMWax/v57jVEVTlXHm2QczGGnZ8+uys92iR544Bnji4AZqtzK0aMnaMSIAeyjg8kNwfIcLDUQEQQ0ApMm3cxLOrZs2amLlBJxmjUmmLW5FUwMhoQE88tWEw985WXKlKGzZ2O4mQEDerIlsmrVBsOdhOUsmzdvN7rBc+vorzMOFsEdtz1Nn3/+HfvBXnjhEZ6qrlgxgEASr776jFqbs4hnaq3AB/4+LC7GkhasMn/00clqwWMvtQjzFYOI4KODExWzUhCsIfr22w+VM7X9VQ0BM1HoD345aGvwxXXo0CbXbYF8QXIYExZkYg0UHMea+HLdkFT0aASwzOPBB+9Ubo8G7MeGGYvJBWhcWOmQnWD9aZ8+3bkKyBKTf5gExHcWKyCefvpB5V8+asy+wvTEswNCgyWEZ3Xo0L7qWV1sdINj27bZT04YB4vgjts0O6jIIBy8HUB4IDtoaPA/YDLCKmnTpoVaJvIFa2mYjACRPPLIVOU/Cze6QJ/4omAxJJytIEisM9q48er8ZFgyc911Hennn79QJkayMg9O5snnBmyee+51hc9D7EcEyb3//kzlV6ljjFl2BIGZM+fwOri3336Bv7edO7dj0/Whh54z/NGuUOrRowtrbcuXr+WX/tSp05U76D71rMxkSwtk+dRTrxqnox58ynhWMVmH5S5Tp75N0AohVauGELRB+JqLi7j1FxQaFGg/AFBPl+vywthCi8K6tqs1Yc1jxs2HfxBrkq5WQMLw3+W0fOBq25fzij8CeH4+//wteumldw3yyemq4BKByarNVl0fzyH8c1iL6kzQF3zKWHJllqFD+7HWN23adHNxkd4vFLIr0ojI4ASBYoAAtH5oWSCqwhCs9wMB6oXzhTGGvPYpZJdXxKS+ICAIFEsE3DZBUSzRkUELAoKAxyAgZOcxt1IuRBAQBLJDQMguO3TkmCAgCHgMAkJ2HnMr5UIEAUEgOwSE7LJDR44JAoKAxyAgZOcxt1IuRBAQBLJDQMguO3TkmCAgCHgMAkJ2HnMr5UIEAUEgOwTcTnb4UbGO45bdwAr6GMKgI8YWflxtleCnYggAgLYhCA7gKoafY5/4qVlOkSvwu1+EhIfgB9w4B4LABy1bNuV9+U8QEAScI+B2skPoo/Hjb3A+GjeWIj4/IkjUqFHVsl4Rarpv3+4chQIBEh966C7+wXZuOkAquo4dXUdKwW94R44cyMSG9m65ZZQR9h3RLNq3b210g0CL9evXMT7LjiAgCBBZp9bkEk3kqURUkMIUaFzQwPDbPmhIVsm2bbto5849KqhABgdBRIAB5NjIjSBKcnYx/RBcACnz0DaIFFEnNI4IpZ2ZaYuDh75Aflu27MpNt1JHELhmEHA72eGh/uuvFS4BVkEWVHTURqyZIPQSwjrrMM9t27ZSIW681YNsC14Iourb93pCgEEE54Q41nHsCO0jUObKletVHol6hlmIev3792ACMefKQKRWEAuSBSHAJxKTILIyTFZETTaHhUImMoRbR4x9R1JHKCloX9AkEaUYxKj7gdaGwInIAQoND6YqAnqiji1Ks+26EPpp5869VKdODY7UAiJF/DtEoFi4cCmHycb5iMqM6BbIFPXPPzs4eRCSDpmJ9/rrO6v+4lSe0AOOEMlnQcAjEXCrGYuHunz5coZG4ogoiGzMmGEqNlwnDi6IOG833TTCMMlCQgKNgJs4FxpMaCj8WOW4KYSrQVJfRG91JaGhrZggkOMhIeGCnWaHIJ6IEKwFRILoyoguAeKdOHGsIqw26nM4h4IHsSCMNQTaFmLaIWIxBKR+7JhNg4VPDSYzfJUHDhzl+GETJow2TFwQY1JSkiLhgRx1NirqrEq52NMuCVGXLm2NMO5mIsV48YccHiBDna1p06btjHNaWioTNLDXgmQ/iFqhx6fLZSsIeDICbtXs8JAiqYyrOHZIoYjsRQhSqKP0gsAQERWaDh5kkA4E0VVr167BOSt9fGxliKaKVIQgC2cC7WrgwJ6c+QgaE8gOBKcFZidid2kBmSLAKDQ49AstDtmStLaFCQM9AWGvbdmbmQjDA1MTwUQhaA8kDU0RYiO7ZM4VC4KH4BgmHUBajmYriHTDBlugUTPxQTOEab5z5792kWvN14VMbUjriOThjnkvuGP5TxDwUATcqtmZH0xHPKFFwY+GNIOa6FAHxKgDCyKxh57p7NfvemVarldJQpK5DJGPQSDmvBKOfcDMRFISHbUYxKBnN1EXZKHJDjO1bdo0V+NZzs0gQxo0PJjYMAHHjBnKkwUIbw0xX5uZ+HAM1xAfn6jaa8EpE2+8cShnaNKkD/JavXqTkZ4O5yCDk9bSzO05Iz6zhoZxmD+jrfh4kLht5rZDh1DWLLdutbkCcFxEELgWEHCrZpedvw5+MPiUHKP8gnSgCUGSk1N5qQj8XjDLEEoaPjRoQdDCoAWBwJwJtEBoiBAk6oVozYo/qP9ACno5x7Bh/Yzkw/DzDR3aX6VxbMwZyOAfhL8LyX+QxBviStuCNjl+/EhlvpelEydOsW8RJAfChV9Nm/bQGM2CdIw6haMrIsULwjxRAT9itWpVlPm60NwU+/dA4hhL797d6KuvflR921WRD4KAxyPgNrLTD7WeQXREFuag1mT0Mfi44GxHImoINDuYk5hIgBkGsoDWB1MUSy3effdzfardFr5A+MMwkbF5c5ZGg/OwhAPaEnJHaHOvdevmTIS6brNmjXky4513PjPGOHhwH4OwHEkH5KQnLrDEBebiV1/9wOPFwO6/f6IyNffxGFEX12FORwcfZLNmjVQeii+4jisiha/OPONbq1YN1ddFQgIgs4DEgT/8gEiTp0nUXEf2BQFPR8BtZiweav0QQnvSf1hkDIGmBGLDA4y493Ci33rrGE7Io0kQhABNDxqZzpEJ8xJZwVauXMdOemc3DFnGkL4RM5J48PWfTh6il5+AOMqUsfn1fvttiUFOyAurE32jb2iInTqFGuaimXRAfNDKNKnjXIwb2mHJkraZU5v2FcZDBS7INwHzEteN65s48SaVcvIfNc5Enkwxa2+or81U7Ot+0FhgYEUO0w2/J/rTgutCXWRTW758nS6WrSBwTSHgNs0ODxtmJZGyzSzffz+fM4BhIgCJf2+77UZ+6M+di6U//vjbLnk2SA+/eEBCYC0gMZi+W7c6X1eGXyVg4sKsWelzoQUh0Q7IDpMaCQkJ3DeWY0RGntHVeNIDhPrcc1OY9GAuY5Li+PEswtKkYyY+NIDkw/DvYbIB48fSEfjjzObvb78tZq0Ls7sY09q1m43sZOb2HIkUmGoNEn0huzs0yeeee5jX5OmXBNwDMGO//fYXw/+J+iKCwLWEQJHLQQHNCdoeCKEoCbQuEGdiYiJrYnkZGwgaEwTQsGCyOhO0b0XGNbRjcwlkZWmHPxPLdvBiEREErlUEihzZXas3oiCuW68b7NatA3388WyXS34Kom9pUxAoagi4zYwtahfu6eOBdoxJmdjY82rd4vdCdJ5+w+X6ckRANLscIZIKgoAg4AkIuG021hPAkmsQBASB4ouAkF3xvXcyckFAEMgDAkJ2eQBLqgoCgkDxRUDIrvjeOxm5ICAI5AEBIbs8gCVVBQFBoPgiIGRXfO+djFwQEATygICQXR7AkqqCgCBQfBEQsiu+905GLggIAnlAQMguD2BJVUFAECi+CAjZFd97JyMXBASBPCAgZJcHsKSqICAIFF8EhOyK772TkQsCgkAeEBCyywNYUlUQEASKLwJCdsX33snIBQFBIA8ICNnlASypKggIAsUXASG74nvvZOSCgCCQBwSE7PIAllQVBASB4ouAkF3xvXcyckFAEMgDAkJ2eQBLqgoCgkDxRUDIrvjeOxm5ICAI5AEBIbs8gCVVBQFBoPgiUGCpFKsHtKXgck0owL86+fuWJy8v4dXi+zWRkQsCBYNAZmYGJacnUHxyBJ1NPEAR8dsLpiPVquWpFEFyDYP6UCm/CgU2aGlYEBAEPBOBpLQ4Ohy9vEBIz1Kya1ZlKNWu2MUz74JclSAgCLgNgbDzG2lf1B+W9meZbSlEZ+l9kcYEgWsaAShNzUKGWoqBJWQH01U0OkvvizQmCFzzCNSu1IXALVaJJWQHH52IICAICAJWI2Alt+Sb7MC8Mhlh9S2W9gQBQQAIgFus0u7yTXZYXiIiCAgCgkBBIRBcrrElTeeb7LCOTkQQEAQEgYJCIMC/hiVN55vssGBYRBAQBASBgkLAKo7J9y8o5JcRBXWLi0a7A29sSC07hPBgMjMz6ezpi7R78xnatSmS1EdLZMrLXWiPanPFwmOWtCeNeBYCXl5ellxQvjU7S0YhjRRZBEJqlKXGrQPJ28eLfP18qGu/WvTA1E7039e6kX+pfL8r+bpBptXriIVQZL8EHjIwa76tHgKGXMaVCOiX6nvPbqCU5MtUoqQPXTegNt18fyua9GR7+mjqJj4J9eo0qkgNmldW2t8F+nfrWUpPz1C/iSZq1DKQzsckUUj1slSlZjk6vDeGwg7HudQMS5Xxo2Ztg6hs+RK0+58zfK5fCW+q16QSxcUm05nwC8ZA6zWpyPvHDpw3ymRHEHCGgJCdM1SkzCUCqSmX2dxsorS9ttdVo/IVS9KlxDR64u3uVK+pjXhw8omDcfTWE2spLTWDHn/rOrqo6pQp52e0e2BXDE1/cr0iPHtbuE2XqjT52Q5Ki8wyOv784RAtmL2P7nm6PSXEpdC0/6zkdtD30+/1oI3LTpKQnQGt7LhAIOsb5aKCFAsCzhCA3w5SRZm5MG1BdN99sIseGv0nffLyZqrdqAKNuqO5cSpM3rcVuT046k+a9/U+YrLsVtU4jh1ogbc80JrS0zKUxvgPTZ28gsKPJ9DgcY0oqGoZ2rj8FNWsF0DVattM3k69avI565eetGtHPggCzhAQsnOGipTliEDSpTSuU1KRGDS85EvprOX1vaE+VVdkdCEhlZq3Czba+XfbWTqwM5qSLqbR0l+OsEncqlMV4zh2YAZXDPSnuTP30s6NkUx0n7+6hRAdDBqfJrVOvWxLETr1rEExZy7Rwd0xdu3IB0HAGQJixjpDRcpyRCCoShmukxCbQhUq+5N/aV8afqv9AvOMy1km6rmzl4w24ctLUL63ikGljDLsVPrf52g146vlbKRtH8ciTybSySPxSpOsSeuWhFGdxhWUebtfV5WtIJAtAkJ22cIjB50hAHOzo9KqLl1Io1PH4in+XDKV9PelpycuNaqDnGKjk8jbW1VWAjNUi5+a5ChfyZ8O7T2ni3h7XrUDCWQijeZ9fV7c/47BPzducks2d+HuA+mJCAK5QUDM2NygdA3X0fMHDVsEUov2ITTi1qb04ud9qFaDAPa9ZWRk0ubV4YrMSlMfZcJiiUqztsFc556nOxjIwaTtP7oBVQ4uTXc+1laRow/t2BBpHMfOiYPnKTryEg2b0IRadgxh39xtU9qwD2/butNcd9OKcEKfLdoH0/4d0aRJ0K4h+SAIOEFANDsnoEjRlQg8/EpWUNY0NSP761f7aNUfx7kifGkgwvH/acl/KExJSqdFPx0yGko4n0Jj72nBfyjcsOwU++WMCmoHJPapmtx48p3uNOWlrP5mv7tDkaDNnE2MT+FlLSDDdUtFqzPjJ/vZIyBklz0+1/zRJXMP0yY1C2qWk0fiKElNSJjlMzWRsOjnQ2qWNYiL/1kZTvHKL6fN2B0bTtO+7dGERcqYUDh+INY4HbO0cWodHiRMtf3YzYupudIOy5QrQXu3naGYqCx/H+pg+QlM6B3rbdoeykQEgZwQELLLCaFr/HiUWsCLv9wIJg/w50q2uyAnzNKaBUS2ZU2EuYj3m4YGUWjXqtSlb036S629wxo+EUEgtwjk22eH7EAigoA7EMAvMLqrX2/AbF4454A7upQ+igACjgvPr3ZI+U6407PB4xK882rRl/MEAUEgRwSQcWzVkbdyrJdThXxrdsj3KCIICAKCQEEhEJ8cbknT+SY7JLYVEQQEAUGgoBA4m3jQkqbzTXbI4A01U0QQEAQEAasRALeAY6yQfJMdBoEM3iKCgCAgCFiNgJXcYgnZgXmRwVtEEBAEBAGrEAiL3WiZVocxWUJ2aGhf1B9CeABCRBAQBPKNAJSnfWf+yHc75gbyvfTE3Bj2keMRiW0ll6wjMvJZEBAEckIAPjqYrlb56cz9WU52unGQHnLKItUisgNJYh6NjGwFAUFAI4AfJSSnJxCWsGFlR0GQnO6rwMhOdyBbQUAQEASKAgKW+eyKwsXIGAQBQUAQcIWAkJ0rZKRcEBAEPAoBITuPup1yMYKAIOAKASE7V8hIuSAgCHgUAkJ2HnU75WIEAUHAFQJCdq6QkXJBQBDwKASE7DzqdsrFCAKCgCsEhOxcISPlgoAg4FEICNl51O2UixEEBAFXCAjZuUJGygUBQcCjEBCy86jbKRcjCAgCrhAQsnOFjJQLAoKAxyBQokRJ6+LZeQwqciGCgCDgcQhUqFCRvCtVCvS4C5MLEgQEAUFAIwCOS0tLJm8v7wwSwtOwyFYQEAQ8CQFwGzguIfE8ecfFnaPMzHQKDq5CsGtFBAFBQBAo7giAy8Bp4DZwXGpKEv0/u4gdwNvOTRMAAAAASUVORK5CYII="
    }
   },
   "cell_type": "markdown",
   "id": "555c5fac",
   "metadata": {},
   "source": [
    "1. Make an account (at runpod.io) and fund it\n",
    "2. Select an **A100** (it's what we used, use a lesser GPU at your own risk) from the **Community Cloud** (it doesn't really matter, but it's slightly cheaper)\n",
    "![instanceselect.png](attachment:instanceselect.png)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6d5e1585",
   "metadata": {},
   "source": [
    "3. For template, select **Runpod Pytorch 2.0.1**"
   ]
  },
  {
   "attachments": {
    "templateselect.png": {
     "image/png": "iVBORw0KGgoAAAANSUhEUgAABS8AAAFlCAYAAAAKxB1QAAABamlDQ1BrQ0dDb2xvclNwYWNlSVRVUl8yMDIwAAAokWNgYNJJLCjIYWFgYMjNKykKcndSiIiMUmB/wMDIwMogwGDJoJ6YXFzgGBDgA1TCAKNRwbdrQNVAcFkXZFY4t3TZsQUBq+bnBPfu07X7gKkeBXClpBYnA+k/QJyfXFBUwsDAmANkK5eXFIDYE4BskSKgo4DsJSB2OoS9A8ROgrDPgNWEBDkD2Q+AbIXkjMQUBgYmkPk6SUji6UhsqL0gIBqUmqyn4BkSqhuk4BSiZ2RgZKBrSMDdJIOS1IoSEO2cX1BZlJmeUaLgCAytVAXPvGQ9HQUjA0MzBgZQ2ENUfw4Ehymj2BmE2JpMBgb3/P///zcixLSAYmseMzCwr0WIKTAzMHCvZWA4/a8gsSgRLAQUYWCyeMLA8AIYH+ICDAwgtsie4jRjI7A8I48TAwPrvf//P6sBzZrMwPB3wv//vxf9//93MVDzHQaGA3kAWuRoHHs2YjEAAABWZVhJZk1NACoAAAAIAAGHaQAEAAAAAQAAABoAAAAAAAOShgAHAAAAEgAAAESgAgAEAAAAAQAABS+gAwAEAAAAAQAAAWUAAAAAQVNDSUkAAABTY3JlZW5zaG90n3xhBAAAAddpVFh0WE1MOmNvbS5hZG9iZS54bXAAAAAAADx4OnhtcG1ldGEgeG1sbnM6eD0iYWRvYmU6bnM6bWV0YS8iIHg6eG1wdGs9IlhNUCBDb3JlIDYuMC4wIj4KICAgPHJkZjpSREYgeG1sbnM6cmRmPSJodHRwOi8vd3d3LnczLm9yZy8xOTk5LzAyLzIyLXJkZi1zeW50YXgtbnMjIj4KICAgICAgPHJkZjpEZXNjcmlwdGlvbiByZGY6YWJvdXQ9IiIKICAgICAgICAgICAgeG1sbnM6ZXhpZj0iaHR0cDovL25zLmFkb2JlLmNvbS9leGlmLzEuMC8iPgogICAgICAgICA8ZXhpZjpQaXhlbFlEaW1lbnNpb24+MzU3PC9leGlmOlBpeGVsWURpbWVuc2lvbj4KICAgICAgICAgPGV4aWY6UGl4ZWxYRGltZW5zaW9uPjEzMjc8L2V4aWY6UGl4ZWxYRGltZW5zaW9uPgogICAgICAgICA8ZXhpZjpVc2VyQ29tbWVudD5TY3JlZW5zaG90PC9leGlmOlVzZXJDb21tZW50PgogICAgICA8L3JkZjpEZXNjcmlwdGlvbj4KICAgPC9yZGY6UkRGPgo8L3g6eG1wbWV0YT4KjfsS/AAAQABJREFUeAHsnQd8FFXXxk9IQkjovZfQe0d6711BQRRBRRFQQSwIiF2wYHktWFA/O1hQwQZKVxAsIE167xB6KGmE7z4nzjC72d1sQhKW8Fx/yczcuW3+M4ObZ08JCg+PvCAsJEACJEACJEACJEACJEACJEACJEACJEACJEACJBBgBLIF2Hq4HBIgARIgARIgARIgARIgARIgARIgARIgARIgARJQAhQv+SCQAAmQAAmQAAmQAAmQAAmQAAmQAAmQAAmQAAkEJAGKlwF5W7goEiABEiABEiABEiABEiABEiABEiABEiABEiABipd8BkiABEiABEiABEiABEiABEiABEiABEiABEiABAKSAMXLgLwtXBQJkAAJkAAJkAAJkAAJkAAJkAAJkAAJkAAJkADFSz4DJEACJEACJEACJEACJEACJEACJEACJEACJEACAUmA4mVA3hYuigRIgARIgARIgARIgARIgARIgARIgARIgARIgOIlnwESIAESIAESIAESIAESIAESIAESIAESIAESIIGAJEDxMiBvCxdFAiRAAiRAAiRAAiRAAiRAAiRAAiRAAiRAAiRA8ZLPAAmQAAmQAAmQAAmQAAmQAAmQAAmQAAmQAAmQQEASoHgZkLeFiyIBEiABEiABEiABEiABEiABEiABEiABEiABEqB4yWeABEiABEiABEiABEiABEiABEiABEiABEiABEggIAmEBOSquCgSIAESIAESIAESIAES8EKgcOGCkj9/PsmVK0KyZ88uQUFeGrKaBEiABEiABEiABEjgshC4cEEkLi5OTp8+K8ePn5CoqKNpXkdQeHikGY6FBEiABEiABEiABEiABAKbAETL0qVLyJkz5+To0WMSHX1GPxRfwKdjFhIgARIgARIgARIggYAhEGS+XcaXzLlz55SCBQtIzpzhsmfP/jSJmBQvA+a2ciEkQAIkQAIkQAIkQALeCERGljGWljll1669cupUtLdmrCcBEiABEiABEiABEghAAnny5JZy5Urpl887duxO1QpTLV4WK1ZYHnnkPqlevbL88MNceeWVqama0FvjatUqyxtvTJT4+AS5/vo7jFnpmWRN69atKePHjzSKbX555pn/ycKFS+02ERHhMnhwP2natIG6Ea1fv1mmT58pq1ats9tgx992Lp3MQVhYmAwfPkjq1aspRYoU1g/Nu3fvNWv+QD9EO9vXr19LBgy4VqpWraSmsUuX/iWffDJDzp4952wmJUoUldtuu1GqVKkghQoVNFYEZ40KvU8WLFgqP/44Tywrgh49Osro0UNd+h46FCXr1m3UdqtXr3c5xwMSIAESIAESIAESyEoEIFxmzx4qmzdvtz8fZaXr47WQAAmQAAmQAAmQwNVAANaYlStXUM+Z1AiYwaGh+Z/wF1Dbts3l+ecnSPHiRbXL5s07ZPnyFf5299lu4MA+KuIFB2eTgwcP64dTq0NwcLDcfvsAefDBYSo+ov633/4Q60KzZcsmzz33iLRr10Kwj2/jIQh27NhK/v13kxw4cFiH8redNa9zC/Gwe/cO5hv/XLJz524pUCC/lClTSpo3v0Z++mm+iq5oD+HSYrR//0FlVbt2NSNkVpT583+zP3CD4ZtvPqsCJ9TnvXv3GxPaCClbtpQKsOXKlZbFi5fpEnBjIco6CywPypcva66xjfzzz1o5fPiI8zT3SYAESIAESIAESCBLEChSpKB+7tqwYYv9OSpLXBgvggRIgARIgARIgASuQgLHjh3XMECJiReSGfl5w+F3tvFGjerKhAn3qUi3YsUab+OlqT4kJFjatGlu94Xo6CxDhtwkN910nWzbtlNg7ehe+vTpJnXq1BBYW/brd5cMHHiPTJz4qgqZY8fea6wms2sXf9u5jw9RsWvXdpKQcF6GDBktd901Rvr2HaLCaKFCBczam2kXWGeOGzdS533ssRdM2/vlxhvv0nYQNXv37mwPPWrUHQLRElaZN988Qtv26TNEXnzxLaNAx0urVk1UfLU7mJ1fflkk7dvfoD+wTn3//elmriAzR29nM+6TAAmQAAmQAAmQQJYhUKpUCfVysTxSssyF8UJIgARIgARIgARI4CokgM90O3fuVQHT38v3W7yEu/WSJX/KrbeOlN9//8tlfAh4P/zwibEs/EpFROvk3XffpnUTJ46zqjxumzZtqAE8//zzHzlx4qTUrFlVihUrYreFm9Bnn30jI0aM1Qu0T/y3A4tLFLhmx8bG6v6CBUsE39DDxRzCJoq/7bSx4xeuHeXAgUNGPN2n+xAYV65MEnERfBSlXr0axjIgn2zcuNUw+lvrTpw4pevCQfv2LbUOFpMQg2EhOmnSq8bSNErrcQNnz14gX3wxU487d26rW0+/jh8/aSwzf9dTTlae2rKOBEiABEiABEiABK5EArC6RFgdxri8Eu8e10wCJEACJEACJEACngngsx0SMBYuXNBzA7dav8XLv/5aJY8/PtmIi6fchhA5cuSYvPvup1o/cuQQgQ97mTIl5dpru+gHzsmT30zWx1nRsWNrPZw791f5+efFut+lSxu7yXvvfSb/93/T1fLRrvxvB3NVqhSpR5aYaLWByzgKzvvbzurr3B49elxdz0uVKi49enTQsSpUKCsdOiRZiP7zT1JczUqVymu3v/9e7eyulpeoqFw56TysMFEQszMmJkls1Yr/fiFW5/33P6HX7Kx33x84sK9Wbdmyw/0Uj0mABEiABEiABEjgiieQP38+k1X8+BV/HbwAEiABEiABEiABEiABVwJHjx7TnDWutZ6PQjxXJ691JpuBEIgCl2WrzJr1s3Fzbi3VqlWSTp1aS7du7dV9+vXX31drSqud+xaxGxHPEZaMS5b8YdyC9kj//r2Mm3Z7+fDDL7W5U+Cz5ra2sIpELMvo6DPJxM2oqKPaH+7Z/rZzXx+OExMT5emnXzaxLB81iXPu0h+r3YcffmHH58ydO5dWuwu8YIcfrAE/RYoU0nZIEW+V0NBQTYJkHWOL63KWhg3ryssvP6FVuCYEr0eszKlTP3E24z4JkAAJkAAJkAAJZAkCCN0DtyIWEiABEiABEiABEiCBrEUAOh4MH/0pfouX/gz27LOvy3vvvSRjxtytzWGtCWtKXwXxLSHS/fbbUhUwt23bpXGNkLimVq1qsnbtBl/dJTQ06RLOnYtJ1i42Nk7rQkJC/G6XbJD/KuDCjQ/Q7gWxKb///he1SMU8KDExydcCARbCJdaLBEQo8fHxusWvfPny2MKkVQnBs2fPQdahuqTDLd1ZSpYsLo0b15c5cxY6q7lPAiRAAiRAAiRAAlc8gezZs2s2yiv+QngBJEACJEACJEACJEACLgTi4uIEn/X8Ka6mff708NFm374DMmvWHG2B+I2wukypWHEgkY375Zef1B/LgrFDh6QYkb7GgI88Sp48SVaPzrYQBFEQK8nfds7+1n7t2tU12c6RI0flkUeeNa7jt5ikPQ+ZZDt/asbvm29Oct8+e/asdoE1qbPASjR//rxaderUadsStXDhJAtMnMBNW736X/2x3N0TEhKcw7gk7OnVa7Am98HYiC0aHp7DpS0PSIAESIAESIAESOBKJwBnHybqudLvItdPAiRAAiRAAiRAAskJ4DPef47dyU+61aSreAkBzYoDCVGtZ8+L2bXd5tVDmIfCzRwFFoR16lTXH8u6sG3b5kaFDdXz3n7BehExN3PkCJPixYu6NCtbtrQe79y5R60c/WnnMsB/B9dcU1f3Zs6cI8uXrxRYeW7dulPeeOMDrW/WrKFud+9OcgNHQh5nKV26hMbJhJs4bg4S+qC0b99CLGvNkyejNc4lYl0iiziKrxhPEGSR3AeCJyw6rXib2pG/SIAESIAESIAESIAESIAESIAESIAESIAESCALEEhX8XL48MG2hSHYXH99d4FFpbeC2Jgo33472wh5N7j8rFu3Ud20mzVr5K27XW9l9u7Vq5NdV7JkMWnVqrGcP39e/vlnrdb7284e5L+dxMQLuocxnQUJfFAQExNlxYrVOl/r1k1dOAwYcJ2etxL5rFy5Vg4dilKx9cEHh6uwqQ3ML7jQDx58gx4uW7bCqva4hUCM2Jco2GchARIgARIgARIgARIgARIgARIgARIgARIggaxEIN1iXsK1unv3DoJ057fccq8m4Rk79l4ZP36U3H77aJNMx9UFGmKblWV8/vzfkjFdsGCJ1KxZVZP/LFr0e7LzzoovvpgpXbq0NWJpD9uSEVabiC35ww9z5fjxk9rc33bOsbG/ePHvMmDAtXp9SLazfv1mzYiEOVHmzUtaPywlYQ3Zo0dHee65R2TZsr9NEp4q0qBBbY3nOWPG99oeguqLL76lCYAQ87N27Wpq0Xnq1CnDraFUrBgpsNKcPv1bbW/9Kl++nAwalCRswiX+mmvqqQB68uQp25rTasstCZAACZAACZAACZAACZAACZAACZAACZAACVzpBNIkXlqxhyyLxLCw7PLww0lJeqZM+VBOnz6jiXq6dm1n3MBrGEvCfsYVepoLq7p1a0ihQgVk//6DsmHDFpdzOIB4iViODRvW0WQ2VgZva25ri7YHD0YZsfB1s4Z7pE+fbqjSsmbNennzzQ//O/K/nd3hvx0kEZo48VXj1n2XNGpUV39wCmuAWPnJJzPsLpgPWcBr1KiiIiROIHHQ88+/oeu0GsL68oknJuuYRYsW1pia1jmIni+99LZmKLfqsK1YsZz+WHXI0I5xPv30azNHrFXNLQmQAAmQAAmQAAmQAAmQAAmQAAmQAAmQAAlkCQJB4eGRST7RqbgciI6IUXn06DHZu/eAui5DsDNynonBuN4eKV++vIKs4XCrds8aDsGuWLEixiryhOzevc/u49ypUqWixrLcsWO3WnTiXLlypSVv3jyakfzEiSSLSqtP3ry5jbBYT8VOCKJW4hvrvLX1t53V3tqGhoYaS8pKgnUhOdHatRvtdVltrC2sRhHP89ixE4Ks67BI9VRgHQqBt3z5MipWgh/GdpaCBfNLqVIlnFUSHX1aEMvTcll3OckDEiABEiABEiABEsgCBJo2bWA8WXyH0ckCl8lLIAESIAESIAESIIGrkoC/n/XSJF5elUR50SRAAiRAAiRAAiRAAplKwN8PtJm6KE5GAiRAAiRAAiRAAiSQLgT8/ayXrgl70mXlHIQESIAESIAESIAESIAESIAESIAESIAESIAESIAEDAGKl3wMSIAESIAESIAESIAESIAESIAESIAESIAESIAEApIAxcuAvC1cFAmQAAmQAAmQAAmQAAmQAAmQAAmQAAmQAAmQQJqyjRMbCZAACZAACZAACZAACQQqgRw5wkwCx7xel3f8+EmJjY31ep4n/CdQpEghQRLNuLh4vzshGWXjxvUlNDREvv9+7lWdgBLPaUJCgpw+fcZvfmxIAiRAAiRAAlcbAYqXV9sd5/WSAAmQAAmQAAmQQBYnULZsKenTp5teZbZs2SQiIlzOnDkrFy5c0LoZM36ULVu2ZzqFJk0ayMaNW1Xsy/TJM2jCO++8WaZN+1Z27Njt1wxBQUGCPvv2HZTNm7dfNuEyPDyHNGpUV377bbl5LvxaeoY06tatnRw8GCULFixJcfxAWXOKC2UDEiABEiABEkhnAhQv0xkohyMBEiABEiABEiABEri8BDZt2ibPPvu6LgKWbQ89NFymTPlATp6MvqwLa9u2mURFHclS4mVqgRYokE9y584lX331g8TExKS2e7q1z5Urp3Tu3EaWLPnDFrXTbfAMGuhKXHMGoeCwJEACJEACVxmB4NDQ/E9cZdfMyyUBEiABEiABEiABErgCCJQuXUL27j1wSSvNkSOHNG/eSJYu/cu4isfpWOXLlxVYQW7dusMeO3v2ULnuum5y4MAhI6rFSq9eneTo0ePSsGEdadaskUB0O3bshHGPThoDHeGe3qBBbWnZsomUL19Gz589e84e09oJDQ1VS1BYhFasGCl58+axLT+TxqgjLVpcI8WLF1ULUViJeisQsNq0aaZu1/nz55MkF/iLa8I64ZKNay5atLBeD9ySrVK5cgVp27a5rrtQofzGAvKQnD9/Xk83bdpQrVQrVSov7du3NNaUu5SZNWazZg2lSJHCxlLwsLo6o1OrVk3k3383Sc2aVc2c10iJEkVl//5D9nlrXmwjI8vItdd2kTx5chu3/jzKdPfufdoEbDB+vXo1levhw0dtUdHbupxjYx/X265dC3Nv60vp0iXlyJFjcu5ccoEUa+zbt5uuo2DBApIrV4RagmIM61o98atYsZxUr15ZQkJClE+5cqXl8OEjEh4ervekTp0a+nzgnqBY7RMTE6VDh1ZStWpFrce6rFKrVjXjMn7Wtlz1dg1pXbM1D7ckQAIkQAIkEIgE/P2sx4Q9gXj3uCYSIAESIAESIAESIIEMIwBREuIUYi9apWrVSkaALGusIk9pFUSlfv16SaFCBYzIuV0gVN12240SFham54ODg2XgwL5SrVol2bZthxEAE+Xuu28z4l4ha0h7m5h43oiVSUIptpaLdUhIsNxyy/VG+KuidXALhks1RExPBXOOGHGrip9r126Q/Pnzan/Uo8DKdOjQgeZ8buOevs2MU0SGDRuksSVx/ppr6kn//r2MBeopc36LEdMqGRGvO05piYwsLd27txcIiTgfF5egY9511y0q6mHtpUsXl+HDB6uAZ/Xr1Km1ctm5c4/yGDjweuuUyxbzOjnAdRwFoh5YQuSDWA0R03L7x3lP60K9s4BZ0rqCZe3ajca6M6fcccfN9rU720Icdq4DbtsoKfErXLigMmzatIHs2rVXSpUqLoMG3SDXX99DIFpHR5/WZ8R6rtAe4jcE20OHotTyF9fVoEEd53LsfV/XkNY124NzhwRIgARIgAR8EMCXfxMnjvX5gzaXq9Bt/HKR57wkQAIkQAIkQAIkQAKXhQBEtO3bd0nt2tVl4cKluobatavJqlXrbGs/VEZFHZVvvvlJz//99xoVBmHVt3jxMiNK1ZagoGzy8cdfGdEtKWgixLf27VvI9OkztY/1C8Lm6tX/CuIbYo5t23bpKQhbECynTPnQtn6ESNWzZyeZOvUTq7u9hTAKK8FZs+ZoghyIdLBmFEmav0uXNrJy5Vr55ZfF2mflyjUqXsI6dPnylSqQ/t//TbetDGHdeOut/SVbtiD7GmAV+MUXs+w5IbytX7/ZzPmzPSbWB2EOVqooGzZslUWLftf9vXv365y4LnerR1iurlr1r1oprl69zp4Tgh4S96xZs17HAKv77rtTLRe3bt2pde7r0krHr1OnouWzz762RUmIu+PHj1Qh2LLutJojfICndaTED/1hKfv557P0fmFtDz44TBBD9Z9/1urwJUsWkxo1qsivvy7X4wIF8strr72nVqCoAJ8BA64z7dfY168Nza+UriGta7bG55YESIAESIAEvBGwYi/jc4ynMn/+Er/iM3vqmx51FC/TgyLHIAESIAESIAESIAESuKII/PPPOnV5hngJoa1y5fLy88+LXK4ByXWsgmQ/OIY4hQK35Dx5cqn7udUG48Dt199SqlQJIwxusYVL9IPoBlfs4OBsas3pHAvu2nBJHjJkgKxYsdYkvNlmW3GiHSwp4cINy0WrxMcn2GuCGIt4k7VqVTVWpwXUwhLu8rAmtYTG/fuTrCGt/mXKlJCvv04ScFEHIXbmzDnWad3u2ZPk+o2DPXv26/XAKtQa06Wx2wHctOHav379JvsMrBghLoOPJV66r8tu/N8ORF9YQ1apUkGFVYiruDZcr78lJX4YB27ilps9sqyjWCIu9qOijqlFLPZRjh07bguXOIbFJ5IWFS5cSK0xUWeVtFyDP2u2xueWBEiABEiABHwR8CZgXm7hEmumeOnrzvEcCZAACZAACZAACZBAliTw77+bTVzLzlKsWBEjSBZXAQrinrPExsY6DzUWJuIbouTMGaHu2RABrQKh0Cl4WvXethjDKXyhHQS/0NAQI7xlTyb+QUB9/fX/U1dkuJp37dpOLTohJho9TPvBZRwxO60Cd+WDB5MsJCGKIt4lYn1ChINFJURbZ3Fm3saYERERKSbWsSxPneP4uw8G8fHxJkZmUtxNqx844JxVnOuy6pxbuPzffHMf5Ql3dFhbIr6ov8UffhgrpXW4z+f+DOEeIm5qRETSc+Rsn9pr8HfNzjm4TwIkQAIkQAK+CEDAxP9fLBfxQBAusV6Kl77uGs+RAAmQAAmQAAmQAAlkSQIQkGClCHdxiJcrV65Ldp0QNjdv3m7XI2mKJTYiKQ1iHH7//S/2+WzZshlX4ET72POO+YvgvwKRrUSJJEtOqw7HcK/2ZrUI8Wv58hX6gxiNo0bdoa7iEOt27doncG+Ha7pVrDXhD5HWrZuqS7glsOL6fBUIdbhOrMmKT4n26Idrh6XgpZYDBw4b0TVU448iFqlVMCeu09+CGKZ//71aZs9eoF1wvV26tPWjO+7HBRUlffHzYyCPTQoVKmisaINta02IyxAucd3uxf9ryNg1u6+LxyRAAiRAAlcXAQiWKPgcYFljXm4CTNhzue8A5ycBEiABEiABEiABErgsBOA6jniQZcqUtOMtOheCLONwE4ebLxLzIJv0unUbtQncu5F5GhaQEAjhej1o0PVqDekcw7mPmIZI/AN3ZhSMgXGRHAhjIDkQ4mKuWLHG2c3eRyKdMWNGGLfkfFoHl2sk/bEsLRHjEslzLFESiWMeeOAunQN/gJw+fUbP4XoQu7FDh5b22N52VqxYLR07trKTCME6cMSIwdrfW5/U1CMTOmJd9u7dRS1ZYXXasmVjzQRuJdXxZzyIqXAVR3+Iha1bN/No3WiNFR19RoVmZIkHQxRf/Kx+qd1ibCRBCgvLrpak117bVQXxmJiYZEOldA2ZteZkC2MFCZAACZDAVUcgUCwuLfC0vLRIcEsCJEACJEACJEACJHBVEUBcRcRwhHWlJ0tHJKGBKzJcpxGHEUlrrOQviD/55ZffGeu+diZjdw+NUQmxzUrU4gnkvHm/yQ039DRxMuubbJ6vasxDJH+B+zeyfsN9Ggl3rOQ37mMgpiPODx8+SAU6nP/ppwXqAo59WB7Cqm/w4BtUTEXczGXLVtiu7LAS7d+/t8bUhICJeSCe+ioQUhE3EmPClR3xHhED02kl6au/P+dmzpxtsot3N1akd6pQDPd9JEKC2OpvQRIlZDmfMOE+tRSBEIkkP94KxENcP7KFwxJ12rRvU+TnbSxf9bt3IwZooowbN1Ld8JBp/osvvvPYJaVryKw1e1wcK0mABEiABEjgMhIICg+PNN/DspAACZAACZAACZAACZBAYBFo2rSBim8ZtSpk2R49+i758cd5tsBnzfXII6NU0ILYBPEOVnHeCgRDxLuE+JhSgWgIKzzLWtJqj/iOEEjhFp5SMUPYa/LWHMmETp8+m8yNHfPny5dHr8c9zmRK86bEIaX+KZ2H2AqB1JOQnFJf6zyuOy4uPhlf67z7FpaRYIL75yze+DnbpLSP57d69Sry/vvT9Lowj3sMTE9jpHQNGblmT+thHQmQAAmQAAlkFAF/P+vR8jKj7gDHJQESIAESIAESIAESCFgCpUoVV3dtWBI641p6WrAv4RLtITr6WyBOuguX6Jua+JEQLE+d8i6mYjxv5zE/MpanpaTEIS1jOvvAQvFShEuM5e26nfM4970JuKkdxzmmp33EWPW3pDR3Zq3Z3/WyHQmQAAmQAAlkNAGKlxlNmOOTAAmQAAmQAAmQAAkEHIGuXdtr1me4C3tKsgOBzx9LyoC7MC4oYAhApM5owTdgLpYLIQESIAESIIEMJEC38QyEy6FJgARIgARIgARIgATSTsBfV6K0z8CeJEACJEACJEACJEACl4uAv5/1mG38ct0hzksCJEACJEACJEACJEACJEACJEACJEACJEACJOCTAMVLn3h4kgRIgARIgARIgARIgARIgARIgARIgARIgARI4HIRoHh5uchzXhIgARIgARIgARIgARIgARIgARIgARIgARIgAZ8EKF76xMOTJEACJEACJEACJEACJEACJEACJEACJEACJEACl4sAxcvLRZ7zkgAJkAAJkAAJkAAJkAAJkAAJkAAJkAAJkAAJ+CRA8dInHp4kARIgARIgARIgARIgARIgARIgARIgARIgARK4XAQoXl4u8pyXBEiABEiABEiABEiABEiABEiABEiABEiABEjAJ4Hg0ND8T/hs4eFkiRJFpUKFSDl9+ozExcV7aCFStGhhKV68qBw7dsLjeU+VhQsXlOrVK0toaKicOhWdrEn58mUlf/58+pM7dy4z/1m5cOFCsnZWRenSJSQoSCQ2Ns6qsrf58uWRQoUKmHlOS4EC+XSt1tgJCQlerwsD4Npy5oyQM2fO2uNZO1gjlhQTE2tVuWwLFswvxYoVlhMnTrnUp9dBrlw5pUaNKpKYmOhxfcHBwVK5cnlzzfnNGk765OdcU3rfG4ztvJ8RERH6PDnndO5nzx4qZcuW1nuOa3MWPI+5cuF5OOOstvcjIsKldOmScvZsjJw/f96uT6+dHDnCzHNbRcLDc8jJk8mf27TOg+cS7wOecfdrwzUXLlxI34W8efOYe31O77lzLrTB/XZ/FrNlyyaRkWUkOvpMsj7O/peyj3mrVauk7xjmcXIHp9Kli3t8B/BO4hm23i3ndaI+Ovq0z2XlzZvbvF9FzH1In/cryPwDEhlZWtfv/m9dvnx5pUiRQj7vOThUrBhpv/Puz67Pi+FJEiABEiAB8/+LErJ37wGSIAESIAESIAESIAESyIIE/P2sF5Kaa4d48OSTD6roFBV1VAWhd9/9VL755id7GAgTzz33iJQqVUIgAuLn8ccny9atO+027jsQxh55ZJTUqlVNMC7EmH37DsjkyW/Jpk1btTlEgHfffVEFl/PnE1TgDAkJkV9+WSSvvvqeRxHulluu176TJr3mPqWMGnWnCkIvvfS2XHttV+nXr5cRTJLELwijBw4ckuefnyLr12926Qsx47XXnpH4+Hi55ZZ7Xea11rhy5Rp56KGnXfpZB089NUb5de16k1WVbts+fbrJXXcN0rWXLFlcFi5cIi+88KbeA0wCsRD3BmJWcHA2gYg1ZszTsmfPfq9ryKh7Y7FKErYSJHv27EZoDpIFC5bI//73bjJRrVOnNjJ69FBzT94w93yxy3offvgeFeP69RuaTKhDw5tv7is33thb7rlnvGzYsMWl76UetGnTzNzrEco8T57cKpqNGzdJjhw5luahK1Uqb+7LCClXrrSOA8F7xYo1+jxCcEbBNVeoUM4I87F6H/HerVmzQZ5++mV9R6w2y5atkM8/n6l9rF958uTSd2nQoJH6nln16bHFPRwx4lbp3LmNCn5BQdnMO5Iob7/9sfz88yKdolq1yjJx4ljx9A7cdFMfFS8nTXpV2zqvE88IxL916zbJ++9Pk+3bdyVb8r33DhHck4ED75aDB6OSnU9NBUTQhx4aLnXr1pQXX3xLZs9eoN1xjb16dZI77xyorAcMGOZxWIiekyY9YtZ83vx7kSD4wuSxxyab+7TeY3tWkgAJkAAJJCeAL4Tx766vL6uT92INCZAACZAACZAACZBAoBNI+ozn3ypT5TYOgQ9Wkf373yVDhz4kI0dOkOHDbxVYR1ll2LBBKgpCSLrxxmEyb95vMnbsvQKh0VPBeK+88pScOxdjRKYR2ue6627TP/Cff/4RtWxy9hs3bqL07XuH9O59qzzwwBPSokVj6d69vbOJvQ+Rq1mzRhIWFmbXYQdWk02bNnARwVatWqfjYuyePQfLTz8tMGLRBG3r7Fy/fi21ZrMsHJ3nrP169WoJxEP3AlGqatWK7tXpclyvXk0ZMOA6ueOO++XWW0eZ/WGCdTjZPProaPn66x9l0KB7Deu79d5AePNWMvLeWHNa97Nnz0HmWXpYateubtZ3g3Xa3nbq1Fp2794rHTu2tuucO7in7dq1cFbpPq6hS5e2yerTowIWl3i2n3nmf4b7A/pe7Nt30IhaN6d5eFjlTpo0Tv74Y6U+jwMGDDf3crj+4fbMMw/r1hr800+/1jbXXXe7toF144QJo63Tl2V7xx036TsJUbxPnyGCdxlfLuDLgiZN6qdpTdZ19uo12IzzqLHA2a/vpvPfHQwMC1u873v27JMOHVqlaS6rE57Dd955QQXhAwcOW9W6xX3o3r2DzJr1s0u9+8F99w2VxYuX6buGd/Ljj7+S8eNHujfjMQmQAAmQgA8CcXFx+gWnjyY8RQIkQAIkQAIkQAIkcAUSgIESPuv5U1IlXsIS7K+/Vtku1Vu2bDfWTYfUog+TQTzo0aOjTJ36qVpd4VtyCA+wSGvUqK7H9fTu3dm0TTAWY6/YllKwDHzttfdl//5DauHkqSPGhhXd0qV/SpUqngVBWKudPXvOCBoNXYZo2bKxsWg7biy4NrrUWwewZvvii1lqsQi3XWeBiAbrQIgS2PdUYBHWs2enZKd69uwo27btTFafHhWwHsWaLStKWP5NmPCc7Ny5R4eH0LZ79z758cd59nTz5v2q7KB2eyoZeW88zYe1z5w5Wxo2rONyumTJYrpOiIR16lQ37tIFXc7jAFxhDedeWrVqbJ6Bsx4tMt3bpva4VKniKiZCaETBM4nnERaungos+CDuO4slgEK0RLnjjoGyevW/8t570+zQCUePHpennnpZrUthneypHD9+Uq0RMQdc7C9HKVOmpFx/fU8j0E2SjRu3Kg8wWbTod/nyy+/khht6XtKyMBae57fe+kj+/PMf8+XJHS7jtW7dVJ//jz+e4fXdtDqkdC/wSuDLkY8++lL/fbL6YYu5R4wYm+K7DNd8PA9W+e23P/TZhWU3CwmQAAmQgH8EEEYkd+6c/jVmKxIgARIgARIgARIggSuGAD7jIRykPyVV4iXEQFg2QaREgZBSsGABI7YkuUHCVx0xLnft2mvPDffqLVt2aMwiu9KxA+vAZctWqsjoqNZdCAQQcbwViAB16tQwruXbPDaB2AHLz/btXS3ycDxv3mKfLkgQ9GAtCndPq8A1F5ae8+cv0XHhnupJKJo5c46x9mvjcg7M2rdvacS5OdZw6bqtWbOqrFr1r1qEPfjgcGPx1lUOHYqy7w0EYbjvO2MnQpiFez44eSoZeW88zYc6WErimXGWjh1bGbfp1UYs2qWWcB06tHSe1n24JCNUgbtla69eneX77+farvPJOl5CxfbtuzUGY9u2zXSUkJBgadu2uRH4V3scFaz79u3hYpWL96lu3RrmXh3RPlg/xHH3AhG+e/eBPl2OYXmZLVuQupG798+MY6x9587dsmPH7mTTffjhF0YMfDJZfVorfvppvsbUdArvSV8s/Ca///63xnSFe7q3ktK9wL9p3kJdwOIyISHl2Kn499JpAYr1IQxGSnE7va2Z9SRAAiRwNRI4fvyE+ayZ/2q8dF4zCZAACZAACZAACWRpAtAT8VnPnxLiTyOrzaxZc4zwUky++updY7l41CTjKCgPPvik/cc4LNE8TXzo0GEjLCV3o8a4SNyxcOFSawq1ZENsOKtAJHD+sT98+GAV4EJDQ6RSpQrGdXOO/PDDXKt5su3cuYtMvL3J5lv7XDoOPgDD6gpxFZ0FcQDz58+rVbguWPHBld0ZI7FlyyZ63RAgIJrAGgDiEyzLnAUu6LCEa9OmucbkxDkIFxB1rRiezvaXug+RFYlKEGsQwh8sSiGs9u7dxcSJfEzX4j4HLP3g3vzSS++4n7KPM/reYCIwxw/MhStVitTYlJ999o29BnDu0KG1fPDBdK2bP/83E5+0p0yf7hrHEeIezkGshNUfCmIOVq5cQR599AXjuttH69LzF+Ivwo0ZMVBvu+1GfcZWrlyrFpCe5kE811Wr1hr39ubyyScztAnETljAQkDGPSxevIiK/Z76+6rDMzBgQB+N0eqeoMdXv/Q8BwtofFGRGQXWzQgVANdxuOrjeYaAj/i2sJyGxSPeuQ0bNntcTkr3wmOnVFY+99wbJm7voyZUw3vGWj1B382RIx9J5ShsTgIkQAJXN4HDh4/ql5Pw4jnlIZnj1U2HV08CJEACJEACJEACVyYBfLbLmTPcb40sVeIl3Ihh9QYrKlj1NW5c3yTweUjuvnucHD58RK0U4QbrqTgtGJ3nYSIaFpbdrkKG5S+/nGofI2HOvfde/IMf1kz79x9U67Jt23YbS7buKgp6slbDIDt27FFrMLiUQuSEqAdBEoKHs0B4mTHjPbsKwuXQoQ+qEGJVQgyB1SUKxCbMiRiM7uIlDBm/++4X4zre0RYv4Ub+1VffW0Ol6zY8PEzFVLiKW4lOIAC++OLjysfdehVC7rPPjjcJSBa6uLW6Lyqj7w3me/rph12mfffdz1wSQEGQgri5dOlf2u7XX5cbwfAO40ZeIZnF7Xff/Syvvz7JuBV/aITqM+q6j3uTUX/sIO4p4h/+++8mfRaQaAqJgW69tb9XAfPnnxfLTTddp+IlTKQRTgFJr1DwjiQmXlAhVytS+AXxGUIoEi/hywFkYx07dmIKvTLudGxsnAq4GTfDxZHxbwaEbUuoxXu4du0GTfiFVrC4RhKwKVM+8Gp16+teXJwp7XtI9gOLa7x/SFyGOJmPPfaAScb0lF+Wm2mfmT1JgARIIGsRQKzjsmVL6Zez3rxFstYV82pIgARIgARIgARIIOsSwN/y5cqVssMe+nOlfouXyA49dOgtms14+fKkGH8Qkh5//AG1anvllakqnhQpUkhFBeeHS5iCwqXZU4HLNyzurIL4fu3bJyVsGTZssLHudHUVQrw5pzUkLLCQJMSbeIlxkbgHruIQL7FFMh73Aos5JI9Bue22AdK8eSPblRd1iLOIpDj4QfZqFLgJo+TLl1esLNBaYX7NnbtY11WxYjkJDw9X93pYmHqzQLX6pWULoc6yPLT6wyoQMfZat25iVekWyYuQEAbC73vvfeZyzv0gM+4NLBc3btyiGeY/+ug1OXr0mMsyBg7sKxDEZ836yK4H9xtu6GWEw1fsOuzArXzr1h3G4q6NucfzVVgeM+YZlzbpeZBknXtO3fGt533NGsSrfFm+/fYnDaHgPt+SJX/IfffdqXExIcAiVidikaLgHiLZDN4HuDW7F4RlOHHilG2JvGTJn/pM414jFiTeHWeBGOr8YsA6BytXFAhq6VlgVdy0aX+PQ0LoxRcTuFZYByPbPe6ju/s11uvPuipUKKeZ2K1rhmCM8vPPn+sW4+MfRCTmwnvgqfi6F57ap6YOIRkQYuKGG+5QIR19Fy78XaZNe1Ngwe20Nk/NuGxLAiRAAlcjAVhf5syZ03hTlJfNm7d7DXdzNbLhNZMACZAACZAACZDAlUQAf6fDQxY6Fjwi/S1+x7yMiMihQgiS6DgLRJb8+fNpVZLgEqQx/Kw2iAmJDN27diUljrHqre26dRvMH/ON1e3TqsMWsQ8bNapjx2x0nnPuwwqzQIH8mgHcWe/ch7Uk4t9BeEQylcWLXd280RbxAiGk4Ofzz79VQdKZqRtxFzEXMqgPHHi3/mAfsN1jamI8xJaEQNGzZ2d1Qf/ll0XGijMOpzKkQBSCy7GzwJUWFrFWgQD9xBMP6Nqee+61FD/8Z8a9sbhDhEJSl9tvH2DHCsWzA8tLCOMWc2zffPNDk9Snti0eW9eHbZLFaydzT1pqwidvbsPOPmndh2i9f/9hF47ITA0xEUKdpwJLQYj+sJjED4R1Z1mzZr1ce20XjbfqrMeXAu+//7IR8wvY1RDxkUAL1siWiGefNDu4dk8JflCH9rCeTs8CK+nixYuqRbb7uEOHDtR7i3ok+oKFqXtMSvwjVrNmFeP6vsm9u8sxnotbbrle/vlnndZbSbXuvPNB+zlBhnZw9padHh1Tuhcuk6byANbCCHeBf5CtAlEWzL09G1Y7bkmABEiABJITQDzluLh4/VwANyMWEiABEiABEiABEiCBK4sAPsPVqlVVM4x7ypXh62r8Fi/xRzjcY++442a1QoQQVqdODZOYpp1JuPO3zgFx7pNPvjJu5LdJxYqRmh1y9Oi71FLSEhrcFwNhEQlOnn32ERUxEe+yRo0qJrvyQ9r011+XuXTBxeKPf4g4ED0hiqxcuUbFR5eGjgNYRULgGTdupFnrChdBwdHM3sW1Tp/+jQwa1M9YTebQeogg3347W8VKCJbWDxLwwJ3cU4EbM0RPrPP773/x1CTd6qZN+1bdla1s061aNdHM7wsWLNU5IAyNGTNC792rr76nGeDBET+WBan7YjLj3jjnhFs9hCkkG0JBPFEITLNnL7B5gzsESrhKX3NNPWd33UcWeMQvhUv199//nOx8elbAArlJk/rGwu4aFdvx7CImK9z3YQXqrUCwxHtTo0bVZBZ4U6d+qrEcH3vsfqldu7ruQ/x//vlHjBj3h8dkON7mwb2HdSfWVKxYEXVhbt78GpOle4hL1nlv/VNbj3vz8svvmCz396mLNKyV8TNkyE0qJluxTHFPYSk9fvwo/WID1sDIVD5u3L1qpYx31Fngeo3nFFbLsKqdNGm8PievvZYU5gHvH2Jcwgrbei+xhRiO0BaIJeqt+LoX3vp4q8fziGtFgaU51g3LYVidwnoYbuOwqoX1OAsJkAAJkEDqCeBD7sGDh81nzHLm/28V9bMg/h+CzzgsJEACJEACJEACJEACgUUAn9HwWQ36HT674TMcDL5SK1ziqvx2G0djJD556KER8uGHr2pcPiTn+fzzmSou4TwKEqlAsECiCiwSwuLkyW+6WKcltbz4+8UX39Q/+m+//UYjYpTSGIX44/+BB56QkyejLzY0e3B5RoHlFiw9kZwGlngpFbhxQ2hyt3Tz1g9C5XXXdZP+/XtpNnQIJ55cPSHCwG0dyWF2797vMhxcm3BTktyBXc+5NEyHA4g3sLR8+ukxKrjCwuuNN/5Pli9PEoIqVChrZz7+7LMpLjOOHDlBhWmXyv8OMuPeWPMizujHH88wFno3qms/RClYycI601ngdmxZ1SGztLPgHGJ5wqUbcQ8zsuDZhlUoxEGETzh/PlGtBu+77zGfz/vq1f/qNw34MuDUKdfnG88K4lbeeedAEx/xfo33iTokpvroo69SdTl49vAO3XffUBWE8YUDXO+++GKWcV/+NlVj+dsYFsZw2e7Wrb26x+MfKyRQGjPmabW4tMbBvwnDhg1SARNWirhvsCRFGAF3K1IkW8IPRE9Yd8Ii+Ouvf9L3ChbaSIz18stvW0PbW4x3+HCUZoDHlwyeiq974am9rzp8mdO5cxvz7+MXmsxr1KgJJqHZcPMlyA36hzViko4f/2yq4nr4mo/nSIAESOBqJGB9SYXPmvC8wZdfCIdC/fJqfBp4zSRAAiRAAiRAAoFMAPlg4uLijPfvWRNW77jfyXk8XVNQeHikGS51BVZvSDbiLiw6R0EbWNFBcEhNQTIZuFxbMQRT05dtRUUSWB76ujdp5cR7450c3odz52L9itfofRTXMxD+kATo5MlTl/w+4IuEXLkikgmDrjOm7xEsDyHoIvu3r4I/QI8fP5mu7HzNl5HnIBC7i+2I44l6iNAsJEACJEACJEACJEACJEACJEACJEACqSOQJvEydVOwNQmQAAmQAAmQAAmQAAmQAAmQAAmQAAmQAAmQAAmknoDfMS9TPzR7kAAJkAAJkAAJkAAJkAAJkAAJkAAJkAAJkAAJkEDaCVC8TDs79iQBEiABEiABEiABEiABEiABEiABEiABEiABEshAAhQvMxAuhyYBEiABEiABEiABEiABEiABEiABEiABEiABEkg7AYqXaWfHniRAAiRAAiRAAiRAAiRAAiRAAiRAAiRAAiRAAhlIgOJlBsLl0CRAAiRAAiRAAiRAAiRAAiRAAiRAAiRAAiRAAmknQPEy7ezYkwRIgARIgARIgARIgARIgARIgARIgARIgARIIAMJULzMQLgcmgRIgARIgARIgARIgARIgARIgARIgARIgARIIO0EKF6mnR17kgAJkAAJkAAJkAAJkAAJkAAJkAAJkAAJkAAJZCABipcZCJdDkwAJkAAJkAAJkAAJkAAJkAAJkAAJkAAJkAAJpJ0Axcu0s2NPEiABEiABEiABEiABEiABEiABEiABEiABEiCBDCRA8TID4XJoEiABEiABEiABEiABEiABEiABEiABEiABEiCBtBOgeJl2duxJAiRAAiRAAiRAAiRAAiRAAiRAAiRAAiRAAiSQgQQoXmYgXA5NAiRAAiRAAiRAAiRAAiRAAiRAAiRAAiRAAiSQdgIUL9POjj1JgARIgARIgARIgARIgARIgARIgARIgARIgAQykEBIBo7NoUmABEiABEiABEiABEggXQkEBQWJ/mTLlrSVoHQdn4ORAAmQAAmQAAmQAAmkD4ELYv67YH4SE5O2Zj8theJlWqixDwmQAAmQAAmQAAmQQKYSgGCZzQiWISGhEpo9TMLMD/azBQeriJmpi+FkJEACJEACJEACJEACPglAtEw8f14SEuIlNi5W4s0P9hP/EzJ9dnY7SfHSDQgPSYAESIAESIAESIAEAouAJVrmyBEhuXLl1g+9MTHn5PTpUxJnPgjjwzELCZAACZAACZAACZBA4BDAF8/ZzZfN4eE5JVfO3JItd17z2S1aYmLO2iKmv6sNCg+P5Kc9f2mxHQmQAAmQAAmQAAmQQKYSgHCJD765c+eT4OAQOXY8SuJiYzJ1DZyMBEiABEiABEiABEjg0giEheWQ/PkLy/nzCRIdfUK/gIYVpj+F4qU/lNiGBEiABEiABEiABEgg0wlYwmXevAX0G/qjRw9n+ho4IQmQAAmQAAmQAAmQQPoRKFioqISYL6RPnjzmt4DJbOPpx58jkQAJkAAJkAAJkAAJpBMBuBohpiUsLhEficJlOoHlMCRAAiRAAiRAAiRwGQkcPXJIEoz1JT7j4bMePvOlVChepkSI50mABEiABEiABEiABDKdAKwuEeMSruIULjMdPyckARIgARIgARIggQwjAAETn/HwWQ+f+VIqKbdIaQSeJwESIAESIAESIAESIIF0JGBZXSI5z7FjdBVPR7QcigRIgARIgARIgAQCgsBxE8ccn/X8sb6keBkQt4yLIAESIAESIAESIAESsAhAvAw1SXoQxB3ZxFlIgARIgARIgARIgASyFoFYk4ARn/XwmS8l13GKl1nr3vNqSIAESIAESIAESOCKJxD0X4bxczHnrvhr4QWQAAmQAAmQAAmQAAl4JhBjPutlh3iZgus4xUvP/FhLAiRAAiRAAiRAAiRwmQgEibG8NAHcz507c5lWwGlJgARIgARIgARIgAQymgA+6+EzHz77+SoUL33R4TkSIAESIAESIAESIIFMJwDXoWzBwRJPl/FMZ88JSYAESIAESIAESCCzCCA8ED7zpeQ2HnKpC8qTo7jkzF5YwkJMkM1sOS51OPYngYAkkJAYI7EJ0XImLkpOxRwIyDVyUSRAAiRAAiSQlQjgQ+yFCxey0iXxWkiABEiABEiABEiABBwE8FkvJeESzdMsXkK0LBBR3gg6p+V07CE5cmaLJJyPkQvmPxYSyEoEYL4cEpxDcoTklVxhRfW5P3Z2O0XMrHSTeS0kQAIkQAIkQAIkQAIkQAIkQAIkkMEEihQpJIcPH8ngWVyHDw/PIRER4XL06HGXE6jPmTNCjhw55lIfiAdpchsvnKuK5M1RWg5Fr5cDp1ZLdOxBiT9/jsJlIN5hrumSCUCQx/ON5xzPO557PP94D1hIgARIgARIgARIgARIgARIgARIgARIwB8CLVpc45eloT9j+dumRImi0rlzGwkJcbVfrFu3pjRsWMffYS5ru1SLlxBs4B6+98Rfci7eVbW9rFfCyUkgkwjgucfzj/eAAmYmQec0JEACJEACJEACJEACJEACJEACJJBFCYSFhUmlSuWlatWKJvt2qH2VEB7r1Kku5cqVtusqVCgrRYsWlooVI9WisnjxIlKmTEmpUaOKHtsNHTsJCQmCflbBHMWKFbEOdZsnT26pXr2yjmW5cpctW0ry588rtWtX1zbZTFZw1NWsWVXy5Mnl0j8jD1IlXsJVHK6zB0+toZVlRt4Vjh3wBGCNifcA7wPeCxYSIAESIAESIAESIAESIAESIAESIAESSC2B0NAQadu2mZw/f978JEr79i3VOrNkyeJSrVol49Z9XAXDSpUidehGjeoKRM0TJ05Krlw5pUmTBip4JiYm9fU0/7Ztu4x4Wc4+Vb58WdmxY5d9XKBAPmnevJGcPn1GChUqqGPiZOXKFVTQPH78hLZt2rSh5M2bR06dipZWrZrq/PYgGbiTKvESMS4R25JxLTPwjnDoK4YA3gO8D3gvWEiABEiABEiABEiABEiABEiABEiABEggtQSKFy8qJ09Gy7FjxyUq6ojExsYKxMR9+w7IwoW/y5kzZ+T48ZMqKmLs+PgE+eefdXasymPHTsjWrTtlw4Yt5uwFCQ29aLlprSUmJlaio09L4cIFtQpCJvpYBVacGHP37n2ycuUatba03MzXrt0ge/bsl7Cw7Lqu3bv3qniJtrDCzIzi6vDuY0ZYlyE5D13FfUDiqauOAN4HvBd4P5iF/Kq7/bxgEiABEiABEiABEiABEiABEiABErgkAkimA0tKuHajREefUStMCINwBY+KOqrH2bIF6fmEhPO6tX7FxcVbu4L94OBsRuC0q+ydjRu3SpUqFQSu37DajIuLs89hDWfPnrOPsY+EPijWfGgDF3K4t1vl4MHD1m6Gbv0WL3NmL6xZxTN0NRycBK5AAqdjD5ks5EXSVbyMjCwj7dq1kMjyZdJEZMf23bJgwRJjBr47Tf3ZiQRIgARIgARIIPMJ3Fw5VD7b7OGvjcxfCmckARIgARIgARLIJALIAg6B8q+/VumMiF95+vRZQUKdP/5YqVnCYRlpiYlpXRbmgMt57drV1MrSOQ4yjpcsWUwtKhEPM2/e3OpC7mwD61BYgMJCE0IrrDhh0ZkZxW/xMiwkt7rIZsaiOAcJXEkEYhJOSsGcFdJ1yZciXGIhED3bSQt5//1p6bouDkYCJEACJEACJJAxBCBcUrzMGLYclQRIgARIgAQCiUDXru3s5Vgu2RAOO3VqLRcuXDAWkDHqvr11647/4lCeFcSzjPdkTmmP5N/Otm07jeVkpLqcOxMDbdiwWVq2bGIEzOIaPxNCKtbiLFjDqlXrpGPHVipaZs+eXRYv/t3ZJMP2g8LDI11X42WqioXay7YjCzIl3mWTJvWlceP68uqr73lZDatJIHAIBEmQVCjUTrYemZ9ui3pm4lgda8Ijz6VpzEvtn6ZJ2YkESIAESIAE0olAaGh2KVykuOzfdzGQfDoNHZDDWMIlFtfth7MBuUYuigRIgARIgARIIGMJIMN3cHCw7T6O2eDiHRISrO7gGTt70uiIc2m5r/uaD8Kn013dV9uUzpUoWVaiDh8w4uxFN3b3Pn5bXqJjZiTqGTXqDsEPCgTMm24aofv8RQKBSiAz3otAvXauiwRIgARIgARI4NIIOIXLSxuJvUmABEiABEiABK5kArB0dBcOYe0YF5eYaZflPr+3idNLuPQ2vnt9qsRL987pfQyLS1hbQrTEPn6mTXsz1QIm4gPAjLVevZqaZn7lyrUya9YcE5D0VHovOV3H++STN0zMgALSvftAXXd6DN6ixTUyYsSt9lCHDx/RmAnffjs73WITPP74A5I7d0558MGn7Hm4c2USQAwNZDpzL/gHDJnEruRSunQJjd+BOB3uJV++PJIrV07Zu/eA+ymXY3y7VKpUCdm+/eqwBHK5eB6QAAmQAAmkOwEKl+mOlAOSAAmQAAmQAAlkQQLZAuWaYG0JoRJbWFsuX75Sl4bgpKkp9evXlh9//FQeeWSUYDvAeIMAAEAASURBVL948SLy2GOjZd68L6VcudKpGSpZW6SF3759uRnv/mTn0qMiT56ckjNnhAQFpd9tQQBViMDICIXSqlUTmWhckn/55XOpWLGc1l3qrxo1KkudOjUudZhM65/R9zHTLiQDJqpWrbK8++6LMnnyoy4/48aNzIDZ0nfIZs0aSq1a1bwOesst18vdd9/m8fyoUXdK//69PZ5zViL+B/iwkAAJkAAJkMClEqBweakE2Z8ESIAESIAESOBqIRAQlpdOV3HLZRwCJkQ3S8T054YgHfzLLz9uYgIEGZFivPz88yINatq6dVN5663nZNy4e+Wuu8aYVPOVTcyAEFm9er0OC2EP6eI3bdpuMiedUAusZs0aSdmyJWXNmvXy99+rzZjBcs019bQ9BFGs7e+/16hJb2hoqDRsWMeMUV7+/HOVbNiwRQObIl5B48b1ZP/+Q5InT26pWbOK/PTTArX+Qrr7Ro3qyKJFy2Tnzj067vjxz+q6YOVWu3Z1iYhISktvXXti4gUz/j96CAuwJk0aSOXK5U3A1H91jVY7T9sXXnhTFi5cqvESYI356qtPy6efTpH27W+QM2eSYivBYrVly8YmOOw5XRdYoICXJaiCwZYt2/W+eAsW640HLGFh8mxxx9jly5eVIkUKarYqnPeXFfp6Y+DkjjZNmzY04vVvcuDAIcHavN1HjMkiGreib9+k0A1XEo927VrKvn0HBAGPPZVfflksTzzxoISFhUls7MWMaPjCoGnTBvLQQ0976sY6EiABEiABEkh3AhQu0x0pByQBEiABEiABEsjCBC67eOkULi3OqIPFZWqES/RFyncIcM8/P0Vmz15gDWeyHy0z1lKfGevA6kaECxK4OefPn8+4lvfXNkg///77L6mwCZFr+vS3VLCzBkAa+OHDx8pHH72qVZ07txH8NGjQxYiSueTLL9/RFPFWe2RlsmJ1wpr02LETUqBAPj398MP3GKvHxXLDDT30eNy4BLnzzgfl11+Xm3U9qO06dOgnL7wwQYVJa0xsITLWqtVOBb/p0990mXPatG9lwoTnnc097kM8xFz/+9+7Zr77pU2bpsZSdb6MHXuPDB060O4D19p+/YYaoXKH8qpYMVLy5s2t/NDo4MEogcAFQdBZypYt5ZXHrbf2l27d2hvxsJuKxOj3zjsvCFyV27Tpo5a3/rKC6OmNAcRrcD906IgULVpIl4d7Pnr04+aZWuHxPlpCrfNauO+ZAKwUv/32J816VrlyBdm1a498/vkslzAENWtWlRYtGkuhQgWMmLhevvvuFztTGfp///0vRjhvYe5PEfPFwofmHRgoU6Z8INHRp+1Jb799gKxbt1EF++uu62aE7YP6rEBgx/O3cOES2bp1p7ZH27Ztm+l+0aKFzXP1sXnGTtpjYWfFijUqzMNCE0K+VTDekSPHdS7UQczs0KGledeqqxi6aNHvsmPHbqu5yxbrQiiGpUv/tOuHDRsk8+cvUZHfWjfef/z7BHfzzz+fKVWrVtJ/Q86cOWe+0Jjn4pKP+cEG/y7hPQTrK91l34bDHRIgARIgAc0oDvGShQRIgARIgARIgARIwD8C6eef7N98Lq08CZdogLiXqRUu0Q9CCsrSpX/p1vnrlVemyq233mcLKM5zEDRRsIVbNSwNJ09+y7idd5L/+7/PVSREPLzq1Vtru48//soIiE1UgBs9eqief+GFKcZq8VojRMxXkaJv3yRxEh3Onz8vPXoMkl69blWxs1evTrqWdu1ukNOnz8jNN/fRcZ2/xox5RgVQiKBz5izUU88++4ZuR44coiIOztWr11GmTv3UtL1OxRDnGL72LfGmQoVyKoZCuPzyy+/M2rtK7963qRD1+usT7SEQE3DSpNeMANxBBg8epdfsFDuthr54fPPNTyZzVjbDooM2B+cKFcrKV199b98Xf1n5wyA0NMQIsHdJ1643G/Fnrzz11ENy9Ohxj/fRWj+3KROASDhmzN3G8jW7rF+/2Qh9rYxV80W3clglIzRBXFycOb/JPE9d1OrZGhn9x48fqVnUICYn3ZNKavVrtYHoOWDAdSpYog5fPNx2W3/p2bOjsdJda57/MHnllaf0+cH5f//dhI0RNQ8LvjyIjU2epQzBj/HlBIRBZ8HxvHmL9RmEZS7Wjn8HVq/+V2O5Yh6I954K1oVn2Flg2QxrYhScv+OOm6RSpUgVRyHeP/PMWK3bunWHiuvPP/+o/aUALMKffvphgeU3xFZYYb/99gtyqSEvnOvjPgmQAAmQwOUjQIvLy8eeM5MACZAACZAACVy5BC6b5aUv4RLiZVpKjhxh2i0mJiYt3bUPrKhQmjdvpAIILMYmTnxVhQ1rfG1gfiFlffv2LY0V2VwjMHyi1aNHP2EExVqaMOjrr3/Qurlzf1WRBwewBoPQAutHFCQI8ZQgBe7qKBCCOnVqo9aR06Z9o+IfxKI9e/YZC8b6+qMNzS+IJnCV96dER5/RZhER4Wb8Vrp/7lysEVL76v6hQ1HGdb2abd0JkfXDD79UIfa33/5QC8Z27ZrLk0++ZE+XEo9hw8aoFWrPnp3kk09mGCGqk3KFJZpV/GEFAdQXg3nzftXhYP0Gl3+UH36YJ/fcc5u5ngJqzaaV/OWRQEhIsBEXR7mcQ7gCiytO/P7738a69Rttg2cR7tjol5Bw3rhfj1DLXksgR9vbb7/RxV0bzylEa6vAGrlduxYq/qOuTZtmsnHjFpcEOnhWR46coG7taIPndejQW+Thh59RS+2FC39XS8n583/DaY9l7txF5l2dbETJXGrlWbBgfrVwhCUyCsRFfFExbNjDdpY3JPqCWI6501JgJfraa+9rV4RkePDB4TJw4N1qPYovJmbMeM9YYlbUcBPdurXTdxwhJGAljQJBf/Dgfi7vmp7gLxIgARIggSuKgD/C5dhlKX+GLZ4zSEwkITl01vxiIQESIAESIAESIIGrgMBlES8zQrjEvbKER1gpWe6k1j2EdRSS1yAmpa+CeJUvv/yOcWO92Vg/NdSmy5atkEGD7k3WDW7UOXOGu7h0Ig4krL9KlLiYsRkWX86C2JXOAjHOU4H765Qpz6pQOWZMUjw+xM5ELEzE6MSPs+CcvwWZl1Gioo5q9mTsDx58AzYuBSIPyoEDh1VEsU4itiBiSTpLSjzOn080LrCzZciQASrY9u3b3YhgK9S922LgDyt/GcDyzyqwYEOBZRuLbwK4B5s3b3NpFBWVJOpblRAWrbJ27UaNJYpQDOCL++N0o8b7AKtdZ9m8ebvzUIXRQYNu0LAJCB0AYRzxYZ1l5cq1tnCJeswBsS81ZceOPSbG7G5BHNwffpirIine+X37Duow1apVUstt63lBJdzGYQUKcTYtBRaWVsGXAqdPn1XhEnWwEEVdsWJFVLxEwiQIqtde28XqomJqZGQZ+5g7JEACJEACVx4Bf4XLNUeTvrjydIUdSoVIp9IhUj5vkBw3oZtXHD4v8/cmyJaT3vt4God1JEACJEACJEACJHClEch0JSejhEuAX7LkT7Xku+66ruoeat2M4OBgFQFhOdmmTV8jwiVqXDvrPFxUUSDaIMELrP+mTPlQxcEHHxymFmGI3+ee+fzUqWh1r7aEQIwB60NkJLbcWFGXlgIRaOrUyUaoDJcbbxxmrMySvom35pw5c46xjnvWHhrWm+7xJ+2THnbguo4CZm3bNtf9pk17qJCIAwi9EJFg9YWCJEVgExcXr8elSpVQKzc9+O+XtTZfPCzxctKkcSrSzJhx0frOOZavfWsebwwsIdTXGDznnQDejxkzkqyGvbWynguctwRnhF1AeAEIcvHxSWKxt/6ml8spWCeuW7fJiIrNzHu2QiIjy6po6GyE+JDOghiwERERtsWn85yvfVh5wlUc4iW2TpEU63eKjRgHcTjx7CM2a1qKxcefvkgeVqhQQfOFQnG7OUR4WK+ykAAJkAAJXJkE0kO4LGGsLftXCpWSZosSYT7Bl4wMkQTzv9MtJy9+WXtlEuKqSYAESIAESIAESMA3Ac8mf777pPlsRgqXWNSRI8dMBu2vTYzDdvLYY/drBmG4f3/66Rua/Oajj77StcNqsFixwhonsnv39saNc5h9TUgcMnv2Zxq3cuPGrSqo4GRYWHYjyMSrUFOpUnk9D5EHLuNdu7Y1bqB91Xrq6afHaLw7JCS5lIL116pVVWbNmqMiH9zH8QMhFi7QvXt3NjE5a6u4eeut/Yy12Cy9Hm9zwkoTlpJIbPLZZ1PUjRcuq5s2bVNXc7j7Io4hxFK4sH7wwStG2JxlWyrClfb55ycYi9JixqV2kFqlQgRyFn94IEYirGJbt25ikqfE2PE8neOktI950sLAGtf9Plr13F46ASSkgdBXsmQxezAI+jVqVFHrTLvSw87cuXAdb65i+rJlf2s8WGcz99iSeA937tyjrupWOyt+rXXsaYtwArBwRHZ7JH5avPh3uxneh8qVy9vH2EEsXVgeW6EWnCcRTgHCvlUgcBYuXMg6TPV2y5btaoUKN3PrB1+kTJ36SarHYgcSIAESIIHLTyA9hEtcRaMiwbZw6byqOoWySa7QJEHTWc99EiABEiABEiABEshKBDLV8rJx4/rJ2CG+ZVpjXCYbzFQ8/fT/VCSBCypEPasgXuMHH3yuh0je07z5NSZxxsN6DAtDqyBpDYTIzz9/y6qSBQuWmOzBv6nFJmL1denSVoVRJLeZNOlVadiwtiaDQUIYFMTJhFXgpVgA9uvXU8fq16+XSTrTS/fxq1mzXjonEoHMmDHVrocgOGvWz/ax+86YMSPsKliCOa0WkUn58ccna6ISWK1aBUmDLPdZWH7CSg2iKQrcfq04gVZ7bH3xsNrBqg/ZzSHMekqsYrXztcU8qWVgjQfx0/0+ImkMSxIB5K9CdmxnQaiDEydcs3c7z1v7uJ8LFizVuI5PP/2KWu/eeOO1mqTp5pvvtpp53C5evMzEJb3dJLEpbEI3vJ2sTd26NTQpFZLuwDJxyJCbzHP8k90OX15Uq1ZRk+zANdubxSOuA8lwkGQIISGcoiRcxF977Rl1J0dsVwiTw4cPNiK7qwu7NSnc6/GFByyYt23bqbEpnVapVjt/t5j/5Zef1AzrSDyGL03wRQZE2rff/tjfYdiOBEiABEggAAikl3CJSwnN5lmgjAgJEs9nAgAAl0ACJEACJEACJEAC6UQgU8VL9zWnt3CJ8SG2wZ36nXc+0YzBiMO4ePFyWbt2gz094tt16NBPLRvhBg1hANZWmzZt1wziPXsOkv79exuX8BgjWi7R81bnUaMeMxaPNTU7MIQaWF516XKzWkXCEgyZjtet26jNIfggI/ihQxfjBQ4dOkbOnElKloNGjzzynJ1p+MknX7QtHQcNGmlc0JN/HD127Li6bnfvfoux/qyj1peImQcX2JgYEwDJrUCAwRqcZc+e/XaMP6t++vSZmkQIbrsFCuRVt3tYnlpl//5Dct99j6mAuXr1enVjhQUjCpL2WLEkT56M9srDGssSCiEUWyW1rJBExRsDWN+5c//66x9VEEKMTxT3+2itg1vzB5LJuP3VV++6oMCz1b37QJc6bwcvvfS2seIdYRI8vaphFBCL9rHHXvAqJlrjIJkNBLv69WuZ9ygp0ZJ1DtuZM3/WLw5GjbpTwxnAcviLLy4+Q999N8cI8ONMuw9NNu8HBKK8twIrT1gyu1sPo88zz/zPWBffopbIcNmeM2eRWnR7Guunn+Zrwp/HH3/AiLun5N13PzX/rlTz1NSvum3bdmmCsLvuQiKie/S9+uuvf8yXKTP96s9GJEACJEACgUEgPYVLXNH64wjjE5rs4racSJToeNdQLMkasYIESIAESIAE0kAAXmXO5MIIp2X9PZ2G4bx2gTecVaCx7N9/MMW/Ha32vraRkaXNWIc8Gky5XxtC08EYJhALjHvwdymSP6emIGwiPHBxz8DBn1KxYqSGTYPG5KtgbHjl7t69zzZ489U+Pc4FhYdH+vWJp2Kh9rL1yPxLmhNu45b1JeJHpqfF5SUtjJ19Evjyy3cEyVg6duzvs11KJxFPENZuSISEf5D697/orp9S30A/nx7vh/Man5k4Vg8nGHE7LeVS+6dlTvc+SHCDcAMQ9fwtEOwggL/99kcuXZDNHC7pH3/8lcbVxJcGCHXgqWBOnL/UgucVa/Fmwekc3xkP1ll/Kfv44iUmJs78zzb5lxKXMi77kgAJkMCVQCA0NLsULlJc9u/bdSUs12WN6S1cWoMPMDEvbzQ/of8FfVprkvt8sz1e/jjk+f+HVj9uSYAESIAESCAtBBCmbvLkR20PPOQbQN6Bjz/+Ur1N0zKmex+Exfvll8/VG+78+QQTgiy7/v0FI5MpUz7w628x9zGt49mzp6kB1qZNF42yrHPu15YzZ4T+7fd//zddvTStdr62FSuWk7JlS6uXrq92l3rO+bewv2O1b99SHnhgmOYpQY4XGMk8+ujzdi4V93FwLU8++ZAkJiaav7fz6t/eY8dO9Ni+bt2a6mkJL8WBA+9JVe4V93mt4xIly0rU4QMmVKP3ON6psryEY8oFt0Qb1mT+bClW+kMp67bBNyotWzbWbzSGDUty2c8KV5sRDls7tu+WyPJlxBIh08IJY1zOAnHRX+ESbuCIi4ks4EOG3O9z2SmNmR7CJRaQ0jzORVqJrJx1l7oP4ZSFBEiABEjgyiKQUcIlKEzfEi/LjVBZOW82iTl/Qf402cbPpZQf78rCx9WSAAmQAAkEGAH8ndO37x26KuQyaNCgtgk594Ds3XtAVq5cm26rHTduomzYsEXHi4w0fwebEH/wOp027dt0m8N9IOe1wfCmceMGxot3pHqpWt607n2cx8ij0KFDywwXL51z+rNfsGB+48V3twq3yDkCQ5tnnx2vOVLef396siHgRTt27L2azwUef7CqfOqpMZrrBCEXnQUhDfv372XyqHwjd999m/NUhu/7LV4mJMZISHAOiT9/LsMXxQkCi0C/fnely4Lwj1uzZkmxPNNlwAAZBO8F3o/0LIiz2k5aqICZlnEhXGKMK6X06NHJiJeV1WX6wIHkJu2ISws3BRYSIAESIAESCFQCGSlcWte841Si4CelUrtgNlljLDNZSIAESIAESCC9CMAqD2Hy8AMRE3/fX3ddN0GYsKVL/7SnQYJfhN9DMlK0gxHTv/9uMvkL2hohLUTmzv3N9F1jt3ffQSivb775SQ2fLPESlpEQCmvVqq7WhMgV4AwTBvGxY8fWOh+MUKZPT53oCcMbXANyIzRsWEetCZHbANafzr9DkQAZwmbu3LmMOHinLh3C3+zZC2T16n/1uESJopqEtkKFcgLxEPlTLJdviwdCF3br1l4TGMMrGdeHHCN16tQQhNlDeDK0cZY2bZoZPaWRWr8iQTQ8Ez2VIkUKqRCJuVEg0v7663LDs4mn5pqMOm/ePGKF9UPIuLfe+lDee+9lDceI8G5WOXr0mBobwWI2YMXL2IRoyRGSl+Kldde4JYH/COC9wPuRngX/EL///rT0HDKgx3J3E3df7GuvvedexWMSIAESIAESCBgCmSFc+nux1lo+2xwv+GEhARIgARIggfQkAEs+K/8FkuhCRHOKly1aXKM5RyBelilTUnr27KR5DZDkFPkBJk0aZ3JQTDA5R7Z5XRaSoEIgQ0FOhokmrBrqFi1aJohl+corTxnX5Sdl69Yd2gZ5EeDJh1wgefLkUhdoWIqmtmAOCKEQEKtXr6QCKoREFLhfDxhwndx22yiJiAhXEbNmzaoq5h45kpRbo1ixIiYB7ESBIAmBFtacXbu2k5EjJ6joCB5IkoyYkshPcvDgYc1zAEtTiMMLF/4ucOF+/fWJRhx9VN29MTeYQiOAQNqqVRN56aXHzTpG2y79aGMVWLBaVqxWXfXqlTU+pXXs3JYuXULHxfxW2bFjj+ZUwTnnfYIojQLrzswufltenomLklxhRSU69mBmr5HzkUBAE8B7cSbucECvkYsjARIgARIgARLIGAKwcoRg6KuMXRaTKZaQlnCJtWB/7dHzmTKvr2vnORIgARIggSuXAITK/Pnz6gUgtwBymNSrV8vEvZzh90Uhl8Cddz6giXN+/HGeCprXXFPPRRRzDoaYi716dRaInSiwUMTcCD2HBM0osK4cOXKIioJIMtOuXQu59dZRdkKhPXsOqAu4NvbzF9zAGzWqq1aL6IK4mxjXEi9h+bhx4xZ1mcf5efN+w8bFbRyJV5GE9pVX3tFzP/44X1588TG58cbexjgpyWUbbtkPP/yMWkSiUe/enfX6RowYZ4vC8EYsV660LV6ePHnKuH6/rmP+/PMiY5n6vtSsWcVmpCe8/GrSpIFabN5++2iPLRDCzbIMdTaAsIpzTvHSeT6z9/0WL0/FHJACEeUlPDS/nIs/ntnr5HwkEJAE8D6EheSSA6eSZ8cOyAVzUSRAAiRAAiRAAulK4LmmOXyOdzmES2tBWFu3H85ah9ySAAmQAAmQQKoJzJjh6gU3fvyzRtDa6vc4sBhEFnGrwO0aGbSd5Y03JhmhbqeEhWXXDOcQAJEYCKVatUoqCFrCJergNg4rSFhJIqP25s3bbOES55cvX4GNzwJhdurUydomT57cAtfpjz760nZpnzfvVxk06AYpUCCfib95wgiZzY2QucDnmFWrVjJJjqbYbZD8dfHiZS4u2zt37rWFSzSEaLps2QpbuETdF198h41d1q9PigeKCnAAf1h5plTA5pFHRml4Nm+Z4uPjE5S7p7FwLlCK3+IlFnzs7HYplLOS7D3x1yUl7gmUi+c6SOBSCCBRD94HvBcsJEACJEACJEACVx+BQLS4dL8LzzUNk7HLYt2reUwCJEACJEACKRJALMju3W/Wdg0a1FF37D179qXYz9kArtjOkph4QYKCgpxVKhrC1TkuLs7EytwhzjiLsNy03MOtTohDCfExPDyHxp9EFnRniY2NVTdsZ52n/Xff/UyrT506bcTTHUYYvLjWgwejjGv4JpNUtplxA19h3NXLqmjqaRzU4Zrgso6xnAXHuIaL5cLFXbOHc97iV1oNPTG0znnbwmpy0qTxJnblZz7FXCRfatKkfrJh4BqOc4FSUhUEANaXMQknpVie2ka2cX3YAuWCuA4SyAwCeP7xHuB9wHvBQgIkQAIkQAIkcHURcLpoe7pyxJsMhKQ5tQsGC1zbWUiABEiABEggtQQQBxGCHn4Qx3H9+k0mYctN9jCnT58xlpIXLQAhJhYuXMg+7++OlQho9er1LsIl+sNtuXLl8i5DVa5cwSTVOWyS6ZxRi00kx3HGuITLtfPYpfN/B0hkY80LS0ancGm1nzsXruPNNQHPsmV/C67XW4GVJURWT2vdvNm7wRNig1aqFOkybLFihaVkyWIudak5gPD4wguPyrff/iSzZv3ss+uuXXukatWKKgJbDRE/NEeOHJocyaq73NtUf5KJOr1JMyuXytdIXcgv9wVwfhLIbAJwFcfzjwzjeB9YSIAESIAESIAErj4CvqwuMztZTkrz3WTiX7KQAAmQAAmQwKUSmDr1U2OJ2FTFLowFd+0WLRpL8+aNjBtzYY1D6W4leKlzwkW8adOGgpiTSOIDi8LhwwebTN1JLtzIZA6RFaIq4klCuBs2bLBflpcprQ0u38iW3rt3FxMDc5FL8yNHjpm1lDBibUF1X8fJOXMWmkQ6N5qkO5FqiYns4r16dTIZ1n916es8WLBgqbF8bKDZxnF9cKl/8cXHpXbt6s5mfu/nzp1Tnn9+gqxZs0HjdcLtHT9W7FIM1K9fLzNfSx0TWeMhECN7OpISwR0d+5988pW6+8M1H8fuoqzfC0qnhqlyG7fmhGCTJ0dxKZq7usmyfFpOxx5SC7SE8zF0J7cgcZtlCMDKMiQ4hyCrOJLzIMYlXMVpcZllbjEvhARIgARIgARSRSCQhEtr4RAwaxkLS1hauhfL+jIQLEHd18ZjEiABEiCBK4fAxo1bNUs2EtOMHv24imN169aUxx9/QJPovPvup5pRPD2vCDEzn3nmf0aQvEXGjLlbXcvnzFkkn376tU4DF/MJE56TsWNHSp8+3VS0fPXV9zRr96WuA2Mj/mb9+rWMlaZrnosVK1ZrAp/p09+SKVM+MFaOs2X27AUqEj777HjJmTNCY2X+73/vyqpV67wuBdc3ceKrctddA+X++4dpu++/nytIzJOW0qlTG4mMLKM/HTu2soeAZWnnzjfqMQTow4eP2MmGJk9+02SAv8PEGX3dCJaxGqfz889nadtcuXKp0Llz5x4jVnu3ILUnyqCdoPDwSFeH+1ROBBEzZ/bCRtDJLSHZfAcsT+XQbE4CAUMAVpaxCdEmq3gURcuAuStcCAmQAAmQQFYlEBqaXQoXKS779+0KyEv8qUeE13VdzgQ5cA/3lkBojck8ztiXXm8bT5AACZAACVwCAcSfhBt2RhfEhzx5Mlrgou2p5M2bWxD/0pMLuKf2/tQ9/PA9Oufbb3/ksXlERLhaKLpbnCJr+okTJz328VaJPojn6T6Wt/ZprYdLPRi6c4TlKu4jLFmdBRahGbmmEiXLStThAyZp0cXETs75sZ8my0vnILA+owWakwj3SYAESIAESIAESIAEsioBX/EjYf14OQssKyFS0vryct4Fzk0CJEACVx+BzBAuQfXEiVM+4ULYTK8C93TEfoSV4pAh93sd1plcyNkotcIl+qalj3NOf/fdxUmrX0yM5wR/GSlcWnOntE11zMuUBuR5EiABEiABEiABEiABEsiqBGp5cMu2rvVyi5dYxzQfAipjX1p3ilsSIAESIAES8E2gR49O0qNHR3XpPnDgkO/GPJvhBC7Z8jLDV8gJSIAESIAESIAESIAESCBACCCupKcSCMIl1uXL+tLTullHAiRAAiRAAiSQnIA3N/HkLVmTGQQ8f/rKjJk5BwmQAAmQAAmQAAmQAAlcYQQ8uWTjEtYad+1AKd6sL63EPYGyTq6DBEiABEiABEiABPwhQPHSH0psQwIkQAIkQAIkQAIkcNUT8BXvkpm8r/rHgwBIgARIgARIgAQyiADFywwCy2FJgARIgARIgARIgASyFgFv8S4DxWXcom25jlvHzq23a3C24T4JkAAJkAAJkAAJBBIBipeBdDe4FhIgARIgARIgARIgARIgARIgARIgARIgARIgAZsAxUsbBXdIgARIgARIgARIgARIIGsT8JZwKGtfNa+OBEiABEiABK4MAtmzhwp+WFwJULx05cEjEiABEiABEiABEiABEvBIwJvwF0jJeqyFrz2aaO1ySwIkQAIkQAIkEIAEqlevLCVLFnNZWf36taR48aIudak5CAoKkvDwHKnp4lfb0NBQadOmmV9tM6IRxcuMoMoxSYAESIAESIAESIAESIAESIAESIAESIAESMALgXz58kpERITL2WLFisjBg4dd6lJzkCtXhHTt2i41Xfxqmy1bNilWrLBfbTOiUUhGDMoxSYAESIAESIAESIAESOBqIYAkOMw2frXcbV4nCZAACZDAlUYgJCREYOW4b98BqVgxUtat2ygFC+aXmJhYOXz4iF5O0aKFJTQ0RPbuPSBlypSUc+diBOJi/vx5Zf/+g1qPhs5z+fLlUaFxz579NpKwsOxStmwpyZMntxw6FCXOc8HB2aRcuTI6N8Z0L7ly5ZT4+ASJjY3TeTE/1li6dAk5fvyE7NixRxISEqRy5fJy5MgxOXbshA6RI0eYVKlSUTZs2CINGtQRjIPtli3b5NSp0+JtTZ64nDlzVooWLWQsQkvoXFu37pCzZ8/ZS8V1VahQThITE834213O2Y0yYIeWlxkAlUOSAAmQAAmQAAmQAAlkPQJXkiu2dxd3upNnvSeTV0QCJEACJOCLQHBwsNSsWUVq1apmRL+jKsrBNbtAgXx2t0KFChjLwiJ6jHNNmzYQiHvR0aelWbNGLueaN7/GiJO55OjR4zpmzZpVtR/ad+zYSvLmzaPCYvXqVaRu3Rr2HBgHQmRU1FEpX76scQ9Pms9qgHkPHDikhxAJ69ataUTJCtoea2vVqonALRwu3FWrVrS6SWRkGcmdO6ecP3/eFmMPH44yImi8XoO3NXniUq5caWnSpIERPU/pXN27dzAxOLPrXGhfu3Y1OXHihF5/hw6t7DVk9A4tLzOaMMcnARIgARIgARIgARLI0gS8CYVZ+qJ5cSRAAiRAAiRwBRGA8LZkyZ8qXPqz7P37DxlLxs3aNCIiXEqUKGq7c8MKcsWKNXoOYmPv3l3UmhMWl9HRZ+Svv1bpub1790ufPt3MuU0ah7Jw4YIyc+YctVrcuXOP6dfZZSmYY9OmbXYdLDV//XW53b5Hj44quG7fvkt69eos2bKt1HMQHFetWqfi5Z49+1TYtCw+YSXpbU2YyJ1Lgwa1Zd68X+XkyWhdx7FjxzWBECxC0fbPP/+RuLh42b17n/Tr11uvC1aqGV1oeZnRhDk+CZAACZAACZAACZBAliDgLTFPbeM2HmglENcUaIy4HhIgARIggauHQHx8vN/CJaicPn3GhgPry7CwMPsYLttWgXAXGxtrLB9zqTs4XMWtApHvxIlT6gIOK09YasLdGuXChQt6bLWFRSXEzaioJDd21B8/ftKlPebFOJgTYyHZD1zEIa4ePHhxXmtMbOEe721NOO/kAvdyWI9awiXOQwS1WMTFxalwifrz5xONy/hZdUnHcUYXipcZTZjjkwAJkAAJkAAJkAAJZAkCvuJa1i4YOB+rb64c6pX3Z5vjvZ7jCRIgARIgARK4WghARPx/9s4DwKribP/DsgsssMDSe+9IlSaCDWNBRewx0ZhEk2gS9Yv5jNEkxhRLevubmKjfl5h8mmgsiUasiIIgvUvvvbOwwC71P8+sczh799y7d/vd5TfKnjZnZs5vzr1n7nPe9x0JdT7VqRP/2enz+KViTIaThE3FpszLyysi5vljEhxjzwtvS5SU0ClR0CeJieGk/N7KcfXqdS5+pqwu167d4MTQcF6/nqhNPo9fSmyVtaf+pVpKvRalGiHaAwEIQAACEIAABCAAgU8ILNx9PJKFJu0hQQACEIAABCBQPQjk5Ox3E+tIIJTloly+k02KMektMXv06GrjQx5wFoxyNZebtj+m+JayqJTlpibXUSxMWVcqSazUJEE+heNd+n2ymmzZsrnb1MRBLVs2cxP1aIfcwzX7d/func2qVetcHv2Re7euSS7eSona5DKE/sgaVBacffr0dHtVxiWXnO+sN0PZqmT1lMxcJdVTKQQgAAEIQAACEIAABKoPgWet5eKAs4oKlakU9zKe5SVWl9XnPqOlEIAABCBQsQTWrNngJrq59trLnbAo92gJjckk5R037gKX9dix4+a99z5063LrXrRombn88gudBeXx48fMpElTP8l3zMWv1KQ7mlhHVo7r128KqlO8Sx8r0+9UPM3Bg89wcSVlJTpt2mxn4anjqlczlksQlXjqkyxAFY/yuuuuMO+/P81OALQjbpv8OeHlhx/ONOeeO8r06NHFCaCKrykXdS/IhvNW5nqtzMwuJyuzQuqCAAQgAAEIQAACEIBAIgIZGXVMi5ZtzJbN6xNlq5Jjcg9/7Kx6kXVLHKxqgVDCZTzx8tvT80wi1/fIi2InBCAAAQhAoAYTSEtLC+JKJnOZI0YMsTEh95tly1Y5C8f8/CORp8kNXQJlVJJlZPg8WThOmHCJefHF/wTZZQkqC8/Jk6fFrWfUqKFOWAxP8uMLSE+v7QRUWVP6lKhNPo9fZmSku/N9jE6/vyKWbdt1Mjt3bLVWo9EsVSdu4xVBnjIhAAEIQAACEIAABGokAYl/8VzH44mGlQkiURsQLiuzJ6gLAhCAAASqA4GyiHNhATL2WuMJl8oXe54EVM0qHi/F5tckPcOGDbJu4y3N6tXRL3plmRkWLlV2ojbF1i3387KwiS2vrNuIl2UlyPkQgAAEIAABCEAAAqcVAbmOx0uJxMN455TX/kR1V7VFaHldI+VAAAIQgAAEqpLAhg2bzI4dp2YEL4+2aMbvnTt3FypKM41rUp6olJZWy1l/Tpw4qUQzqEeVVV32EfOyuvQU7YQABCAAAQhAAAIQSAkC3vpyQMQkPV5ArGyxUPX6uqMgVXZ7otrAPghAAAIQgEB1J6AYkpWRFMcyHMsyXOf+/bn2WG54V41fx/KyxncxFwgBCEAAAhCAAAQgUN4EirO+VGzMykoIl5VFmnogAAEIQAACEKgKAmW2vNQ079nZTUzDhvVNnTp17OxMVXEZ1AkBCEAAAhCAAAQgEI+AYrUfOXLE5OYeMnv37ivimhTvPPbHJ5DI+lJnaVKfypggpzjhUvE5sbqM348cgQAEIAABCEAg9QmUWryUaNmhQ1tz8OBhO7vRHjfFuwbFsQFBUx8BLYQABCAAAQhAAAI1m0At+3ZZL5mzshqYZs2aujHcxo1bEDHL2O3fnp5vRcq6Jsp9XEVLwJRwWFHiYXHCpdqQyEJUx0kQgAAEIAABCEAg1QmUSrzUdO0a/K5atS6uD36qXzjtgwAEIAABCEAAAqcLAb1czs/Pd/927dpjGjXKMp07t7eeMw3M2rUbThcMFXKdEgcHnFU7btk+DmV5CphySf+MjXEZTzT1jVGdzDDuabCEAAQgAAEIQKC6EqidkZH9UEkaL+FSb+6XLl3pBsAlOZe8EIAABCAAAQhAAAJVTyA//4idKXO3admyhRUyG5p9+3KqvlGhFtSuXds0aJBlDhxIrXaFmhisbj9sffJtSiQk6pgXMRftPhGcW5oVlXPPoLqmVf3EMTUr0uKzNO3mHAhAAAIQgAAEIBBFIKtRE3PoYK45ceJ41GG3r0TipVzFmzXLdsIl7uFxmXIAAhCAAAQgAAEIVAsCe/bsdS7kJ06cNIcOHU6ZNlcn8VLQvCCZSMBUvrCIqTDxXvjUsURJlpbfGFTHiZbF1aFyEC4T0eQYBCAAAQhAAAKpRCAZ8bJEbuOKcSlXcYTLVOpm2gIBCEAAAhCAAARKR0BjunXrNpnu3TsT/7J0CIOzvFu4t7AMDkSsFOTJcEc0oY4XP7Vjkd3uby01+38yW3kyYmW4CoTLMA3WIQABCNQsAm3btja9e3c3GRkZdt6RjWbZslWRF1i3bl3Tt29P06pVc7N9+y7z8ccrIj1n27Vrbfr06WFmzVpgcnL2R5ZVlp2Kud2/f2+jdh85ctSsWLHabNq0tUiRdepkmMGD+5ujR4+auXMXBcfbt29jevTo6rZXrlwTeW6QuRxXxE48Nm/elrDUeO0ujmuvXt1s+J4O1tLwhFm5cq0di22MrCc7u7Ht7x42bGNDs3HjZmdIGJkxzs569eqaYcMGmylTPoqTo2y7zzprqFm0aKmdEPJg2QpK4uykxUtZXWpynv37DyRRLFkgAAEIQAACEIAABKoDAY3tNMbTWG/nzt3Vockp20YJhxIfNVFPskniZGGBskDUTPb8cL7KmN08XB/rEIAABCBQsQQk/nnjMXnBjhw5xEydOtMKkUfMiBGD3bHly1cXacSoUUOd+DZ9+mzTq1d3o+333vuwUL6zzx7mYmAr/rVEuIpIw4YNMhLQpk2bbTIz65kxY0aYd9+dYvbuPRUWpmXL5kZt2b8/1+h6fdL1SnibOnWGSU9PN2eddaY5fDjPThi912epsGWTJo2tkHosYfnx2l0c1+7du5hu3To7JrVrp1kmI62we8Rs2bK9UH0SqM87b5SZP3+JWb58lRk6dKA7rhCOySZ5skjArqjUsmUzK6QnLSuWqRmJg+WEis7ObuJmFQ/tYhUCEIAABCAAAQhAoAYQ2L17j9FYj1R2ApogZ9xrhypshvGoFsp6E+Eyigz7IAABCFRvAmPHjjZDhvR3FyHru4ULl9qY1bucMClrSln4xSaJfhLWZMGYk3PAWlXON40bN3IhAH1eiYT79u03b7zxnhPO/P54y379erlJ/gYM6OvaozqUJHzqWDh17drJvRDVvj179pkZM+Y5I7jt23dagW5bcMyfI3Hz3XenWi/ftX6XWw4a1M9aGq5wL1a3bt3uRDztKy5JrBsyZIBRW+vXz3TZJX7GnitLT4mHSlpKUBw+fLCRtWc4paWlWevPLlY4PtMMHNjPibE6HtXuZLiqfbK2VLxxCbGaOLFVqxbhKt16z55dHT8dF0cJwLHXUOQku0PWmoMHn+Hy1q1bp1AWta9Tp/buOmVxK3FTSXU1bXpqHKhr07X6pDafeeYAV6Zn6o9V1jJp8bJhw/o2aHrFm4JW1oVTDwQgAAEIQAACEIBAAQGN8TTWI5Ufgcpy35Zo+e3p+cwqXn5dR0kQgAAEUobAokXLnLilBsm1OCzwyWNC4mRsklCpl5I+yXJz167dTsAM71uyZHlg1en3x1sqvMy5557l3LrlrSGLQAmkipd9xhm97SR7BWMIiWOyEMzLy3dFqb35+QXrjRtnOZFOImY4bdiwOdLDV9exa9ep65Boq32JklyxJTLu37/fWXFedtmFbsJpiXRytw8niXcSJpVGjRrmYoDLA0Xia5s2LYOsslqV4Ldt2w4nWF5wwWh3LKrdYl0cVwm4nTu3d+2SENihQ5siVpeqoKAfT1mZyjVbrveeddDA0IoEyLFjxzj+hw/nOxahw0601CTcYlmQt+BaZOUZ5qM8WVkN3KliKlFXYvfx4/YF7bixRmEJKjslbd+pGcZlykqCAAQgAAEIQAACEKhZBDTG01iPVL4EvICpGJfJxMIsSe2+7JKcQ14IQAACEKheBGKFPt96WcJJZHvttbf9rmAp0Snslq0DEhm9GBVkjFiRxZ2sKX2SYHZKiFzn3Jd1LC8vz4mWkyZNdbE3JXBJtJPop3A0Bw7k+iLc8jOfucq5fn/00ZxIwbVQZrshEVQiXfg6Dh8+7PbpmCwKo9op68B33vkgqEMTE8olPpELuCxaJQS/8sobLgalROIrr7w4aNLs2QucaKj4lBs3bjE33jjBuUonKjM4OWJFZajvbrhhvLvO1avXOTExNquuT3FNw0n9qP3iH/aYURgBMZcV7IIFS5xlp87T+E5MlMRT/fOvf73pRGtd5xVXXOREzDVr1pvx4y+2Yu5cx0D9KXd1JVn+ykLXT+wo9/9u3Tq5OKouQyX9SVq8tPdH0qp8JbWdaiAAAQhAAAIQgAAEyoGALAU01iNVDAEvNHoB0y9LWpvcw5+1cTXlmk6CAAQgAIHTk0CTJo1cnETFsJRoFZskNmZnF7hL+2MS6GQ5WFxq1qypneinR5BNEwJJbFMKW3PKQlHxLJVWr15v428OceKlLPa0HZueffZlF19TVownTpwsZEEam1fbGpfo2iQ8Hj583GVp2LCh26djUe2UNaHcw8PWqL7tiSwFZYEo922Jk77ucFxNiXVyw27aNDuI76h6SiteapKb3NxD5u9/f8VZfmpbLuxz5y509fs/EihjY5GqH7Vf16M2+aRJmRYu/NgJmppAx6dwn8nVv1GjLCNrVJ/kdaN9ckuXlavCEEgwltCp+0UvtiWWemtTnSeBO2rSJV9mRS2TFi8rqgGUCwEIQAACEIAABCAAgdOBgERMJb8Mi5h+dvEwBz8LuSYBQrAMk2EdAhCAwOlBQLEWJXQp5qOSRCUJSdOmzXKCWxQFiXcSEcNJwlXUxD7hPFrfvHmr+xe7X9vh+IkSsCSiKUnIlJinWIsSvxRjU0kWkhJavfWkLDIVp1Ozh4fd313miD/Kr/M1SY+SrkH7lKLaqfoUu1L/5N4cTidPnnBCofJI/JQbuY93qfJ1PeHkt+VWfvHF55uZM+e6GKISVD/96SvDWUu83qFDW/PyyxMD8VNioyYyihUvNYGRJg4ypsD6Un2vPlCoHwmtb7/9QZG61Se+7ToYFm0lakugfOedKYXOO3asYGKiNWvWmc6dO1p39X1GlpjipNnfjx8/7qxZJTr7dOJEgaDstytjmXTMy8poDHVAAAIQgAAEIAABCEDgdCHgLTK1VNzK2H/+OMLl6XJHcJ0QgAAEChOQtZ9EQSUJURdeOMbMmbOwSIxEiYdesJSYKAGrY8d27rw2bVo5sUv7leQSXJpQMb1793CCpATAvn17WkH1lCWnxC7Nsr1t207nYq16JH5JaFV9SjqvY8f2gZgpN3a1LV6SwKlJYzSbtURExd1MJHqqPtXfp09PV6QEyksuOd+JnooVKZHOM1J8R7VHSVaHii8p13ElWWL6CXQkcKal1bJi6TZn9dm2batSsVN5ivmpJDHX94224zHRRD0Sr717vCbV0T5vIapzY5P6RNfvr0195pNc6GVF2qhRQ+dOLl6ymPWzhW/YsMW0bt3CcZYru1IB0x2uD+WCLjGzf//ezsLTZajEP1heViJsqoIABCAAAQhAAAIQgAAEIAABCEAAAskQmDFjbpBt4MC+TmST67X+KcmaTlZ8Ejjlfrxly3YrsuWbKVNm2O0z3UQrsizUtizoJMZpv7ZL6vorF+RrrrnM1Su36unTZ7t1/ZHYpZmw581bHOzTyuTJ05yoqbiLqnvLlh3WvbkglmKnTh2sKNbexu0ssCotdKLdWLVqnRPJrrzyEieiaYIc7UuUPvxwpp1YaJQT/SReSlT1LuCzZi2wE9gMcTEgFy9eZsXMAutBiZoffPCROeeckY6RhM716ze5auQavmTJCjNhwqXWAvSwm7QmylU/UZt0TIKfJrxR/MypU2eY0aNH2H19nMgoMVP7lMJMcnL2O2vPiy461+XT+VOmfOTyxfuj6xo9eri59trLncgpa1vFR1XS9Rb0x3BbnnEirI7769H9oXtC91LY9X7qVDE9y4qiBRMciefevfviNaHC9tfKzOxyyvYzQTW6wadPn5MgB4cgAAEIQAACEIAABKorgVQa62Vk1DEtWtrZNzcXjZtVXfnSbghAAAIQgEBFEpC1nSzlwkkWmd4t2O+PyuePxVtqAhsJfBK1JELGxnuUS/Oll15gXnzxP0XaoDJl3SfxLLZ9ybTFzwieyOIwtt2qT67jsed4i8TYdvjz5ZbtxTy/T8v0dLmZpzthOLw/2fWo61SZakesi3u8vF5sTaZO9btc5WPL9ufqOiXSxuPg84WXKlMp9n4K5yntett2nczOHVvtfVU0hqsvE8tLT4IlBCAAAQhAAAIQgAAEIAABCEAAAhCohgSihKgooSkqX7KXKzEwVhDs3r2L6dWrm5HVX7yyY8VOX1+8/P64lrH1hY/FWy9tfVHCpeqQcFgS8TC2XVHXGa+8kuSNrcdvR/W7P6ZlvOsM54ldL67M2PzlvY14Wd5EKQ8CEIAABCAAAQhAAAIQgAAEIAABCNQQAkuWLDeHDhVMnBN7SRIX589f7OJCxh5jGwLlRQDxsrxIUg4EIAABCEAAAhCAAAQgAAEIQAACEKhhBBLFmlRcSRIEKpoAs41XNGHKhwAEIAABCECgzAQUKPzqq8eVuRwKgAAEIAABCEAAAhCAAASqFwHEy+rVX7QWAhCAAAQgcFoSOO+8UcEMl6kIoFmzbHPLLdeb888/OxWbR5sgAAEIQAACEIAABCBQbQngNl5tu46GQwACEIAABGo+gQkTLjHnnDPSdOrUwdSvn2l++cuHzPbtO80f/vCM2b//QACgXr265mtf+4IZN26s2bhxi/n85+8OjsWuKLD8HXd8znTs2N6VqfImT55mnnnmhUJZR44801xxxUWmf//eLrD5ggUfm9///n/Nnj37CuXThmbq/tznrjN//esL5q67bjOalfOnP33cvPnm5CJ52QEBCEAAAhCAAAQgAAEIJE8A8TJ5VimXU1YegwadYd59d0rKtY0GQQACEIAABMpK4JprLjNf/ernCxUzcGA/t92hQztz553fcbNaSox86KFvmjZtWrljtWoVOqXQhkTOhx/+tmnevKk5cCDXCpF7rTDa3llNHj6cZ1544VWX/4ILRpsHHrjLrW/YsNlkZtazVpWjzODB/cztt99ndu7cXajcM88c6LZnz15odC4JAhCAAAQgAAEIQAACECgfApXuNt60aRMzYEBfU6dORvlcwSelqNyuXTu5f126dCj38su1sUkUpmuoW7dOwpyf/ezVQR79qOrSpWNk/qyshqZduzaRx8p7Z9u2rUyTJo3Ku9iUKm/YsEGmZ8+urk3irh+9iVKrVi1MdnbjRFmCYyr7sssuDLZZgQAEIHA6E/jMZ65yl//tbz9sPvjgI2dxKeFw9+69pk+fHkbfmQ0bNrDWkI8afde+9960YnFJEJVwOWnSVOuGfpu5+eY7rZj5G3eeLCdr165tMjIynPVkLauCfuMb3zdf/OI3zI033uGEzSZNGpsvf/mmQvUo35Ah/U1eXr5ZunRloWPa0DNdz4tEqXXrFqZ7986JsnAMAhCAAAQgAAEIQAACpyWBSrO87NGjq/nWt75qOnfuYHJyDpgGDeqbjz6aYx577HfOFaus9CdMuNRcf/14c/DgQZOWluZ+JCxfvtq89NLr5v33p5e1+Eo///e//4n5r/960Cxfviqybom15547yvzxj391x/v06Wl+9rPvmV/96o/mtdfeKXSOJjm44opPma985VuF9lfExn33fd1Mnz7H/P3vryRVvH6oyRUwVa1HJfxKTPzHP/7lrHt0UTfddI2RFc4vfvGE/fHc09x//53muuu+FPd677jjFqMZ2GLdEaNO0I/eb3zjy8598eDBQ1FZ2AcBCEDgtCBQt25d+zKssbWOPGhmzZrvvoslEq5cucY+735v+vbt6dzG09NrOzfxRx/9nTl+/JizjkwESG7gR44cNb/97dM2/3GXVUJmo0YNTePGjawYWt+98MvKauCewYsWLQ2Ke/rp58z48RdbN/azrEv4783Ro0fdsZ49uzkR9cMPZ35S5km3X/Evv/zlm91LvWPHjptp02a5th86dNjWlWXHKP9j9u7NMdu27XBi7K5de8wNN3wlqC9qRZanAwf2jToU7JN7+4IFS4JtViAAAQhAAAIQgAAEIFCdCVSKeClrgkceud/GfXrPfPObP3A/Njp2bGckdD3wwN3W1evngTBUFpjz5y82ss5QklXFiBFDrGD6NXPs2DHz4YezylJ0yp0rQW3hwo+LCL933PF5M2/eYrN587aUa3NUgyT+XXjhmJQVL2Ux+aUvfdY8//y/g3v0nnseMidOnIi6nDLvy809aGbPXmBjpV1inn32pTKXRwEQgAAEqiuB/Px8Z2kpi8qvf/2LhS5DYqb+KaWnp9uXc/faZ/1xZ+FYKGPMRuvWLU2LFs3MqlXrzFVXXWpfAp5lWrZsbp+nS80rr0y0/95wZ8glXen48cLf9RIrNaaQZ0T79m3M2rUbXL6hQwe45Zw5C93S/5FlqFzRN2zY5OJrKnbn3r37nHDq8+g5o38qS+JlccmLkvEETITL4ghyHAIQgAAEIAABCECguhGoFPHytttuchYATz31bMBHlmvf+c5j5qmnfmFGjRpmxcWZ7geFXLZkoSZLyjZtWpq5cxcWsSQMCkmwoh8A//nPO84qQgLp+PG3BBYWcm2WNYQsJdat22j++c9XnWWHijvzzAHO9XzlyrXmkkvOd0Kr4l/JmlOTBsgq4513PrDtWhTULjdptbdbt87W8vOQmTjxXXu9H7vjvrwlS5abiy8+37qzp5u3357irssXIKuRT33qXFf3vn37zXPPvewPxV3KRXzZssKuabIkefvt962Ae6e5++7vJRTYzjijtxk9eoT9wdTIXcu77051P8hU4VVXjTNbtmxz1rGjRg11ViH6Qbd589a47Yk6oB+Iifpz7Ngx1rq0wGJRbZ44cVJgKeLb16pVcyvGLnL3gBcM7777S+bVV98yY8eOtm6CLc2Pf/yroM26ZyRa//znf3BWvr16dbMi4Cmel156gbO4kaWn7xv9aB0//iLHSyK3LIJOmtr0AABAAElEQVSVZC0sK0glWVfK+ubf/37LWd3s2LHL3bPuoP0jgf6GGyZYZplOfHz77Q8CsdPn8UtZHavtilcqK+SXX37dWXL64xIv5ZKIeOmJsIQABE5XAj/60a/td/x9TmgUAz3nbr/9c/a5/Vog9ElMPJUKgl3WqpV2aldoTc8lJVn9659PI0cOMcOHDzbf+95P3DNg/fpNdvxwzOgZonwSO5UuvPAc92zUur7LfToV73KB3+WWH300172gleipyX8Ua1PPXll9hpNid3788YrwroTr8QTMmiZcnjx50sjaVksSBCAAAQhAAAIQgEDNI5DsWC96dF/OPHr37u5iS8UWu29fjhPO+vTp7g4pbpVEvHvu+Yr9UbLbbN26w1pTfM4JU7HnJrv9zjtTnPCoeFNKsuD49a9/5IRSiVTdunUyjz/+aBA7UhahErKuvvpSs3jxUvfDRVaj3/72152QqcD+P/nJ96xLWWtXnmY+/e1vHzb9+vWy7tKzXRyuRx/9jhO+lEHlyUXt5puvNStWrLaD8DRnhaofRD5JjLvuuivcD5fc3Fzzgx/c69rsj0ctFcNSs6nGpieeeMa5ot1444TYQ8H2WWcNdT8G5V63bNlqe62XmXvvvSM4LmuOL37xRjN06EAzf/4Sey1drIv0940mOShJKq4/N27cbBkvc0XKgkZ9riTx8Yc/vNfGDjts61/sxELx90kTJmgSBcUl80Kj2vzVr95izjijj+uH/Pwjjr2uIZzUT569+kYWrLrfZPGiyRfuvferRgKnku5PCYlKs2YtsK7fG9y66tJ945O4PPjgN92kDxK9b7rp2iLx0HxeWQj96Ef3OcFeFjr60f3EEz91QqvPI6FboqxirpEgAAEInM4Eli5dYa3fv+le5uzYsdvFs9bz8rnn/uCeUyVlE463Le8FhVO55ppbjdzB09JqWe+Q292zRRP56MWlnjN/+MNP7MvWu92LQb3I8sm7nOsZcMYZvYzaF+v1oBez3rVczyu5i2uyPZXrBVY9e0oiXPr6JWD6F6XaVxOFyxPWrT+jTsnGHp4PSwhAAAIQgAAEIACB1CdQx471NOYr7mV1hVteKqaTrOEk6kQlxa6StYNPGRnp5m9/e9FIwFHSj4mzzx7mrNN8npIsZa23du16Z2UpywlZtL311mRr8VlgBfrGG+9Z8fIRa2V5gfnXvwrcxWRN8cMf/soJS5o19P/+73ErWD7uzlPdstgcM2aki+so8UlxJ2fMmGPzF8TO6tmzi51EYKCLy6X8sszUjy8JarIGlWima1ZMTs2QqllJP//5u4OZSzdu3BrMcKrzo5LE0/37c4sc0mQBivv1i1885FzqVqxYUyTPf//3Hc7qQzG+lCZNmmL+539+7SwRvctbTs5+G8/rcXf8/fenWSuXp50gG4795Q4W8ydRf6ptEpeVwjEvFRtV1+BdAhWz9LnnnrAC38vOUlb533xzcjAjrLaVNm3a6qwwC7aS+ysr3FtvvScQgpctW2W+//1vuvL1g1JWtpoYSUtv+RlbsgRsWQt7IVXLJ5/8hbPS3Lp1e6Hs48ZdYH+0ptn+fTQoTz+Ab7nleita/8Ll3bNnn12edJ8bWSiTIAABCJzOBPSdKHGxffu2pn79etYS/233UkvfzR9/vNx+985NGo+sKZX0ff7ggz8NvC5k6X7RReeaDh3a2md0Wztu2Gifi8+5F5tXXz0umD1czyU9f9u2be28ElSWYlBKjJw5c542P0kFFqBe4PR7/bY8LnwqbqDm80UtvQWmjoXXo/JWt30n7XPw6LGjNoZ5A3MkP6+6NZ/2QgACEIAABCAAAQgkQUBjPY35NPZLlCpcvNQPhRMnTlpriTqR7VDcqCNHjhQ6JgHJJ1nmjRkzwm3KlfgHP/hvf8gt7733R24ylEI7Yzbq1avnZgDVbrmof/DBdGttOC7IJVGxa9dTM3WvW7cpcKHevn2ny7dq1SnxVRaPEmSV5MKmAPyy6JPbuMQwLdevPyU6yapPdfika5IFqJKsAGWRKaHMJy+C+e2opSw5JAxGJU36ohiNshDRrKzh1KZNK6NJCKZMmRHsluuyLBw1c6sXL8OzpWqyBLnXyzVa4uUNN1xpJ0e6Ijhf8bxuuumUZWRw4JOVeP0Zm0/bikemCRo0g7d+RPokKxi5yqsdSlGibNQ+f368paxkwhas+mEqs2XV72OZxTvX75doPXt2Qew17VN5im+mGcljxUvF+JTVjUIQ+CQL1fBM8fohu3//AccC8dJTYgkBCJxuBPRdKcFSVvBy49Z3sywXn3zy/6yl+z5rbf95G36jf4nEy927C2JK6nw928JJ+/Tc0ZhBL6UUOuStt953YUqaNs121vWyrnz++T+6czXJjpK38J8zp7DLeLjsilyvaaKlZ3XSCsxHjuSbhg2yTI7fyRICEIAABCAAAQhAoEYRqFcv0+QePGA09kuUotWvRGeU8Jh+aMg9WK7HUTET9eMgLG6peG+ZoHUJn8YUWDBIQJRYGU5RZYaPSxzVjxHN2q0fPnIZa9GiuYtN6fNJpFq9er3ftMvEim8oo5sYSC7VsuKYP/9ja/23xSxdekp8Vd7w9Whb16S2KGk269hZpTVJQTwrP3eS/SPRTT+m4qW//vWfzrrztts+6370+XyyApWQ6t3Y/H6Jg4rn6VNsm0+ePBG0WZar3ipS+XUsUQqXFe7PqHM0aYGSROBwmjp1hvuxempf0T4qjfWK7s9wUhkSYzXrbLJJ4ru3uvXnqNxGjbL8ZrDU9TVv3sxN9OB36vxp02b7TbeU2K8+IUEAAhA4XQno5Z7CbKxeLa+Je50rif+e10tCJYVaKS7JKlLPvt2797pnp2Jia1K/rl07BS8/mzZt4l4m6tkooVQvB3/5y4fc8/naa28LXnLJ2lMvnLzngupW/GS1KxwLu7g2cbx4AmJ61IqXaVmN7ditrhMyiz+LHBCAAAQgAAEIQAAC1YVA3br1XMhEjfn8OD9e2ytcvFTFiislS7OpU2cGFo3arziUGvS/9NLr2iw2ySV6zZqwyFjsKdYd9wYX1H/Llu0OhtzRNaGOJlTxST9swgKb35/M8rzzRtkfT4fsTKgPBLAHDOibzKkuj36UXXnlxa7DvGDZuXMHt52oEE2oEyvwhfPreh577HcuVpeP26jjqk+CrtzewrG5ZCXoZ1kNlxO1vndvTuAuF3W8LPvUv2r7c8+9UsgatTR9pJm7ZeEqodh/EGRRGbYqlQu/4lD6CR80mYNCHRQWs618/onYHHVtstCRxai3wlF56kM/wUP4HIVJ0A/n8GQNsdcmgV1lhvsnXAbrEIAABE4HAnp2yXpdQuVjj33HzQqusC7f+943jJ69emYqrEiiJPdsxRWWBecDDzziBMZ//ONf5mtf+4Kd2O1B88Ybk90z51OfOsfFdf7rX19w1p0q8733PnST+ykMi6wqFTJGMZn1Eu4f//i3q9Z7CihmpZ45pPIjoOf2MetClJt7wL6s1TO2wPOi/GqgJAhAAAIQgAAEIACBqiSQnd3CjfU05vOaTbz2pMU7UJ77//Snv7lZOR988B4jYU/WZ/rh8cgjD1jBbGK5xWmSpYSsJ1q2bG7jZA53gfcvvfR8G2T/4QCEYlzeeutnnMWFrlEi3l//+jvrTj60VJcsNzPVJ6sOCVyKZTls2KCky5KYqh9gapOC/stN7vbbbynW8lLWnUOG9E9Yj1yOn3rq/+zMpsODfHJznzTpQzuL9lfcpEUSMuUGLmvAsDVlcEIFr0jI049KiYb6kSmr0MmTpxnFvfRWmJoF9qWXnnZsStIcxRSVqKjrU/k33DDeWuG2K1SE7hn9iJVYqB+hmrxHccv8j1D1r/pn0KB+cSfQkdXlnXfe6u4BlaPy9u7dZ8XLU6EGfKW6NoVBOOeckS5GmvJr9lnNnutTt25dnEuiXMdJEIAABE5XAnpe3X//Iy5ciJ6revkkq0eNH+Ty/fDDvwksIk8xKrDK9x4Bepmkl1j6rtfzQOnllyc68VHW8XoufOYzV7lnhMKp/OUvLwRF/epXf3Lxt/v27Wkn3bvOCZd6Rv3sZ78Pvt99zG4fciU4OakV39aingRJnX4aZNLzNy/vkBWYj9kxQEH/nQaXzSVCAAIQgAAEIACBGk+gWfNWboynsZ7GfMWlSrG8lAutBMQvfekmGyD/HidKycLu9dffdZPzFNfIZI8PGnSGncTlSWdFIas3CYP33PNQIbdp1SnLukcffcAJVvpBox8y06fPSbaaQvk0k6jEweeff9IOsPNcnMRkYlb6QsTmu999zPK5y8XhVKf95jdP2Yl8Ovsskcs33pjkfnDJ7TyRe7GsKTW7uK7Zp1/84gk3q/b//M+vnIWnRE5ZpMiisrKTrFmWLVvpZo59/PH/dX3xs5/9wc1+/swzv3PNkRufLBXl8leS5MXbm266xk26M3HiJGv9eyrWp8qStYysLv/5z6csi1pudvUf//jXQTUSMTWBlIR2zSb/0EM/D475lZ07d1nhd56bVEgGmrq3v/Odx4q45iu/LDr1g/srX7nZ3Hff153Vp879+99f8cVZgb+PmTdvUbDNCgQgAIHTlYBiCGtSNcUF1vemXjg+9tj/c3GPY8OfiNHWrTvcc1/hV5TksXHHHfdZL4+B1utiktunt7p/+tNfnWWlXo6lpaW58zQxW/iNr0K63HXXd93zuH//vs4KVPElFVrEp6FDB7jVsIeDdrz44n+cVajC5oTTd7/7E/fiSsKsvAw0RtE6KZqAt748cGCfHcc0dQLm7t0FsUajz2AvBCAAAQhAAAIQgECqE5BwmV473eTk7HGeNuExeLy218rM7JLUK/+zzjqz1AJfuHJZJ8rty1u2hY9V9rqsHPft2+9+QJS1blmDyIqxpAJbuF4JjPqxFBs/MZwnvH7zzde6GJ6ahbU0SZaOmZmZCcXP0pRbmnNkgSiLlrD7vn5QyvqyLEzVFrllS1SM5XrVVZc6K0j9eMzMrOcsZ2NjYPprycjIcOKm2hgvKY/KSdZiUv2dl3fEXnfBj2xf7hNP/MTOFv/HYLZ6v58lBCAAgdOZwC23XO9iWIdfMFU1jxdffMpN8DN+/C2Fnl9V3a7S1l9eY73S1h/vPI0HFPcyK6uJfaanmz17dzIDeTxY7IcABCAAAQhAAAIpSkAxLuUqLq8avZzW5IzJWF3qcirF8jLMTYpqKgiXalNZRbHwdemaynpdmvW7JOmf/3wtcH8vyXk+r8S8RFabPl9lLKNEQ93E5dFHYUE03rWELWmi8kRZ+MTmU55k8vnzovpbrpGygFVsTBIEIAABCJwi8Je/PH9qI0XWrrnmthRpSc1uhsYDGtzq7Xy9evVNc+tCrn15eYetJexBdyyZN/Y1mxJXBwEIQAACEIAABFKLgIwX9QI6M7OBHcNlOo8nxTOXq7jiXCYrXOqqKl28TC2U1bs1EtzkGk8qHQFZue7aVTJX9NLVlPxZmtxq/vwlyZ9ATghAAAIQgMBpQECD26NH5aFxzHosHDYZdiBc1/5r0qSZSXMeFtbFggQBCEAAAhCAAAQgkDIE9HL5hA2V5CZhPHjAaFZxL1qW9MUz4mXKdCsNqWwCb731vtG/VEqJ3NJTqZ20BQIQgAAEIFDZBDTIlTeFFzJldak3+vqPBAEIQAACEIAABCCQegROGvufHcOdtC+i3dKulyYhXpaGGudAAAIQgAAEIAABCFQJgWDgawfBJAhAAAIQgAAEIACBmk8greZfIlcIAQhAAAIQgAAEIAABCEAAAhCAAAQgAAEIVEcCiJfVsddoMwQgAAEIQAACEIAABCAAAQhAAAIQgAAETgMCiJenQSdziRCAAAQgAAEIQAACEIAABCAAAQhAAAIQqI4EEC+rY6/RZghAAAIQgAAEIAABCEAAAhCAAAQgAAEInAYEEC9Pg07mEiEAAQhAAAIQgAAEIAABCEAAAhCAAAQgUB0JIF5Wx16jzRCAAAQgAAEIQAACEIAABCAAAQhAAAIQOA0IJC1enjxpTK1atU4DJFwiBCAAAQhAAAIQOL0IaIynsR4JAhCAAAQgAAEIQAACqUYgafHyyJEjpk6dOqnWftoDAQhAAAIQgAAEIFBGAhrjaaxHggAEIAABCEAAAhCAQKoRSFq8zM09ZLKyGqRa+2kPBCAAAQhAAAIQgEAZCWiMp7EeCQIQgAAEIAABCEAAAqlGIGnxcu/efaZZs6ap1n7aAwEIQAACEIAABCBQRgIa42msR4IABCAAAQhAAAIQgECqEUhavNy5c7dp0CDTNGqUlWrXQHsgAAEIQAACEIAABEpJQGM7jfE01iNBAAIQgAAEIAABCEAg1QgkLV6q4Rs3bjGdO7dn4p5U60XaAwEIQAACEIAABEpBQBP1aGynMR4JAhCAAAQgAAEIQAACqUigROKl3sgfOHDQ9OzZDQEzFXuTNkEAAhCAAAQgAIEkCUi41JhOYzusLpOERjYIQAACEIAABCAAgUonUCLxUq1bu3aDm42yf//euJBXendRIQQgAAEIQAACECg7AbmKayynGcY1tiNBAAIQgAAEIAABCEAgVQnUyszscrI0jWvRopnp0KGtOXjwsNm9e497a68B8MmTpSquNE3gHAhAAAIQgAAEIACBJAjIyrJOnTpGs4prch7FuJSrOBaXScAjCwQgAAEIQAACEIBAlRIotXjpWy0RMzu7iWnYsL4bFNuxMQkCEIAABCAAAQhAIIUI6N2yXjLn5h5ys4ojWqZQ59AUCEAAAhCAAAQgAIGEBMosXiYsnYMQgAAEIAABCEAAAhCAAAQgAAEIQAACEIAABEpJoMQxL0tZD6dBAAIQgAAEIAABCEAAAhCAAAQgAAEIQAACECgRAcTLEuEiMwQgAAEIQAACEIAABCAAAQhAAAIQgAAEIFBZBBAvK4s09UAAAhCAAAQgAAEIQAACEIAABCAAAQhAAAIlIoB4WSJcZIYABCAAAQhAAAIQgAAEIAABCEAAAhCAAAQqiwDiZWWRph4IQAACEIAABCAAAQhAAAIQgAAEIAABCECgRAQQL0uEi8wQgAAEIAABCEAAAhCAAAQgAAEIQAACEIBAZRFAvKws0tQDAQhAAAIQgAAEIAABCEAAAhCAAAQgAAEIlIhAeolyF5M5K6uhSU+v7XKdPHnSHDiQa44fP1HMWRxOBQKNGzcyjRtnmQ0bNsdtTpMmjUxu7iFz7NixuHmq0wHdr0eOHDH5+UdSptkNGzZwfPPy8lOmTZXdkJp2n8Xya9u2tb3n8s3u3XtjD7ENAQikIIHmDbqbbs3PC1qWdzTH7D60xmzbv9gcO1G67+oh7W8yB/K3mZU73wnK9SsD211v6qU38pvBcsXOt83eQ+vddt9Wl5useq3N7I3PmOMnCj/D6mc0Na2y+pgmmR3N8ZNHTc7hTWbbgSUm/9iBoCxWIAABCEAAAhCAAAQgUJ0IlKvl5ejRw82ll15gPvWpc83FF59vbrzxKrt+jqlTp051YnJatrVr147mzDMHBNfeoUNb06pV82BbK5dccr5p1iy70L6q3IhqY0nao/u1U6f2JTmlwvOOHDnEdOvWucLrKa8KytoHUe1Itfssqo3J7qtbt44544zehbIPGNDH9OzZrdA+NiAAgdQlUCe9oWlav4vJqJ3pGtk4s53p3+ZqM7Lzl01ardK9A25av7PJqtsq8qKb1OtgGtdrF3lMO+ukNzAdm45wbWqd1bdQvmwrWI7ueqfp3WqcaVK/o2nZsLfp2/oKt69BncLP9EInsgEBCEAAAhCAAAQgAIEUJlC6UXeCC1q4cKlZvHiZy5GZWc+MGjXMnHvuSPP22x8kOItDVU1g3rzFZsGCj4NmdOnS0VnObt++K9iXaivVoY2pxqy820MfJCZar15dM2RIf7NkyXIja3SlN9+cHKwnPpujEIBAKhCo9Ukjlu94y+zMXe62OmWPdKJg60b9zJacBXZfLdOoXhuTXb+Tyc3f7iwkT5w8HjS/dlqGaVa/m6lfp6nZdXCV3V/wfRBkCK3UqlXL5B7ZaWasfyq099Rqu0aDbW0F757bNR5iNufMdwe1b0C76+x6LTNv83Nm+/4ldjXNtG9ypunXerw5o81VtswnTxXEGgQgAAEIQAACEIAABKoJgXIXL8PXffhwnpk3b5G57LILTe3ata0L+XGTlpZmevToYlq3bmk0QN+0aYtZtWqdO2348MFm7doNZufO3UExXbt2MhIAPv54RbAvvNKuXRvTpUsHZ925e/ceKxKsKOTWLMs6WYfJfV11bdy4JXx6sJ5M3S1bNjcdO7az7alntm3bbtas2WBOnChwi+/Tp4c5ePBQIbfroUMHuuuJ5x6and3Ytr2jkfuy2pWXl2eys5s4ocM3rHPnDqZdu9b2mo67PFu2bPOHEi7btm1lWrZsYebPX+zyiftZZw21bV5ntm7d4faF82i9WbOmZtGipWbw4DOM6lVq0KC+mTNnoVFfKqWl1TL9+vUyLVo0M/v3H3BtTeR2rWtT/zRtmm327cux/bjSuWq7wuyfREyVR5ae3bt3MbJgU7vXrFnv7qN4bZRgrr5o1CjL7Nq1x+U/dOiwr67IUvdgjx5dja5f16Pyc3JOudY1adLY9OrVzTRsWN+KuQfN0qUr3NIXFK99/rjvP9Wje3vz5sL9p77VPZWWVtusWLHanxZ3WadOhuXZyVnFKizDunUbzd69OUF+WW4uXbrS3vPtTPPmTR3zxYuXF/pMBJntSqLPo8/n+6h+/UwjMVvtlBAXrw98GzxH9d2wYYPM9Olziv0O8HXGWybzOUv0mS+uP3X/6DtH9/eePftcn8lycvr02UGTPA99NrZt22F5rIkUI3VvjBx5pjtv9OgRZseOnWb58tWmd+/uwXeF7jt95vUd0b17ZxfCQEKn7pfevXvY75o67p70n1nfiOLuK5+PJQQgUDEEduQuM33NFaZhHY1lapsRHW91wqWvTS7aH63/kzl0ZI+zsBze6VZTp3YDd/jEyWNWupR4GV/A9OVELds3GWqOHDto1Ib2TYY4F/O8Y/td/XIZX77jDefS7s61AurGvTNdGyS46viho3uiik3pfRozDhzY140dS9NQfVfrBa2WJAhAAAIQgAAEIACB6kegXN3Goy7/xImCwbl+jCvJVVcChARK/Rs8uL/7Ma9jEjj1wz6cBgzoa44ePRbeFaxLVDrnnJFOCJD4J5FS5fuk4xIlJWLt3LnLrcdzEy6ubpV9wQVnW7HypBUa9rhrkFWpT61atXAihN/WUqKUBJ+opLh+co+VS/2WLdutq3An57atAbpPcuOWQLRv334rbObbaxvmhDZ/PNFSYmP//r1NRkaBPi0xRnV069YlOE2ioI9RKrFPQorSjh0F4rHEWIml4RiX6i8JaNu373S8zz//7KC82BWJOwoj0LBhQ1dO8+bNrJA91vWz8hbHtGnTJo6Rrl2xOCX0ShBWimqjhDgJ5apXop7iR15wwWgn0LmTIv7oHlE71AfqK/WJzlNS/QXCe5otb5ORsKVwCLpX/HHlj2qfjsviTj+2JIJJyBszZkSh/pOrvkQt9dWePXvNiBFDrJCd5cqO+lO7dpq7Ht3D4q97R+1RO32SqHX22cNdG/X5khAX/kz4fH6Z6POoPO3btzFjx452oqPK6927W1BeVB/oHLVBAr9P6enprh1qv1JxdfrzopbFfc4SfeaL60+188ILzzFt2rR0InODBpnWavws2/aOQVP0skT3vD4T27btdMJ2PL7qV/+yQUsvMoevQZ87uZDrO3HHjl1OcB47doy7LxQXU2UoDIdeAvhU3H3l87GEAAQqjkC2dftWkhDY3lo/yuJSMTAnr/qZFQ/fMnXTs0z35he4PFqmWYHzo3V/Mu+s+JFZu3uqqV0rI650KWFTMS97tBgb/OvStOBZm2WtOxvWbWE25cwxW/bL4rKWadt4kKvHu6HvPrjGbYf/7PlkX8O6p8YY4eOpvl4W4VLX5sXPVL9O2gcBCEAAAhCAAAQgEE2gQi0vJSZJQJN46AUwWSnJws1bw0nQkHi0bNkqs3r1OisejHHCi6w0JTZIiFq/fmNk6yUevPXW5GDii5yc/S7GpoRSWYZpYgyVqbKVZDXmXTdjCyyu7lGjhpoZM+Y5SyydK6vLCRMudoKfhK+SpiFDBjiL01mzCty9VP/ll38qKEbipoTcf//7Lee+rQOyGLjoonNtG9Zbnqfc0YKTQisSSiQ+tmnTygl/svCTAOgFSjHS+nvvTQudVbC6efNWJ/7Jsk/XGU6yXlVoACWJWePGjXVWkVHWlxIaVedHH81x+WVhO2LEYDcxkAS94pi2atXSCX8LFljXN5skAEnsUYpqo+4XCYyy8NP9JgFTQlGiJI7vvfehuy9WrlzjhLVBg/qZqVNnuntUx7wApfJuuGG8uy917YnapwmQJKS98sobgdXq3r37rLA4zKgeJYmVH344yzHSttheeeUlWo1MshCVYPnaa28HFr8STlXOxImTgnMk5MuCVkmfifPOG+UEXG8lHGS0K4k+j8ongV79t3ZtwWdQZUtQl+gd1QfhsuOtF1dnvPOS2Z/oM6/vnET9qf4So3ffnfrJ/WCstfKZ7n71dav/pkyZEdwT+m665prLTZMmS91LBp9PS9UnS97+/fu4ZbzvHr0MeP/96a5ufZdcffU4d1/oO0FJFpwSrBWOI5n7yp3EHwhAoNwJ9Gp5kenabIyztMzO7GAn68kzOw4sNQPaXms0ic/8Lf+w3x0nzJrd79uYlTZutI1HKVfuFlm9nAv33sMFk+2s2f2B6dr8nITtC4ufyphvLSvX7vnQCqWD3Xmbc+Za9/Rdrl65jqvMNOuarnT85BG3DP859smkPnJfr44p/GK3tO0vjzJKWzfnQQACEIAABCAAAQiUjUC5i5eyCpKFnMSxRo0aWqEy17zzzgdBKyXASZiTcCZrIrm3etcpWR5JjJGYKaFI5WzcuNm6GR8Nzg+vSJiRWCVLL9Uld05ZxUkMkJi2fv0m67Y5xAot6VYY2mrdjrfHFS8T1Z2V1cCJRhs2bAqql1WUxFNZE5ZGvGzePLuQe7hEE4lkEieUVK5cSSUg+iRrO12XXLDV3uKSrlliToFo2drMmjXPCW8S+bz1YDLlhOuRaOeTRGkJMrJUjBIvZe0pcc4nXaOERaVkmErMk1Am6zf1pVyuVWe8JMFWrt0SeCX86PrFLFFSmWFRSaxUp5LuRfGR9aH6RcKp7iXdc0qJ2teiRVNr9Zrh7k2X2f7RuXXr1nXnyyLW35f+uNquezpekhu4Pg9hEVKfE7k160WB379r16k+Uvt1TG2WmB2bEn0e1Udy+d6w4VSohdzcg068iy2nJNuJ6ixJOVF5E33mi+tPiYS6Z8L3g7Zlvaqk+1whLPT91bhxgYiu/UeOHLFW142LiJc6lkxSOAXfd76PZInrk6x2vTVwcfeVLDVJEIBA+RKIde6WULnExpncbmfwPnL8kKmX0dgcPrrPfnecCCo+dHS3SU87w2TWsc9ba2Wp4z5phvJjx/OtsBmdatkjOXmbzbS1vy+UQUJo20YFVpZjuv5XoWONrFjqZxOvbyfmyc3faRpntnd17z20zjSoWzBZj9zLT6ek501ZRcusrIb2eV3gcSF2elb7F/LlwVLjNI0p44UYSlSHxiXeqyE2n4wANAav6KQxt8bhK1euLfT8LG29MlrQM7G451nBeKpuqZ+9pW0f50EAAhCAAAQgUDUEyl28lPgjyyoJkhKTwoMPCZrnnz/qE8Fvm3OllXjUrl2Bu7IQrF693g2CJMpoMCSrr3hJos0A61Yu0U+ijxfpfH5ZPakNsmCUBZUGQxLT4gl28erWAEkWeoqbGU4S7CRmlCZJuNLAMpzC2ypXokhs0r5k65TQJcs55Zf4ITdf8dZAWQPxTZsKC3exdUVt+zAA/lhY6PH7/FLCV9Q16HgyTCXcvvzy686l9owzerk+VPzNgvvL13Jqqba8+upbLqaqLNVk+SnL0XC8wlO5C9aOHi0sjEsoV9uU9INHLsISkvSjQsKtt1zV8UTt827T+mERTop5qM+BLCjV31608nmOHIkOkaDj6kdZrIaT+OqHi+4nzzrcR4n6p7jPoziojeH7Mlx3adaLq7M0ZYbPSfSZL64/JSjH/iANb/vPnX7IhpO+d8Lfc+Fj5b1e3H1V3vVRHgQgIMfsghSesCfMJe/ofhdTUqJjQSxLYzJtbEmJlHlHcqwl5FErYjYNTtGs5em169q8JUvNG3Z3M41vP/CxOZBfED9ZwmgXaw3azlpkyuJTY692VuCURahmFx/Y9jqzetdk06JhT9eenMOnXsKWrPbql1teG/PnLzHyphg4sF+pL0ChQfSCSuPAgud3hhtHTp483YmOpS74kxMHDOjjxsOJxirx6ijwDih46a0Xpkp+XKMXY2+9pXuiYpNerA4bNti9yNcLzrImxcjW2N173cQrT548+p3w0kuvm/KoN1497IcABCAAAQhAIDUIlLt4Kcs472Ybe4my+JNw9sILrwaWeor1Fk4SH8aPv9gKmm2cxVgiq0a5Y37wwXRnLaUyNLgMJw0yNXjzA0LFv1TcpHgzn8erW4MoiX0SLcKWkLLUWrZspatSope3jtIOiUl6exwvSQyTS3PYklCWil4gVVxNubHqGrwApbfbakOyb+flJq9z+vTpGVidytW3b99eLhamnxU+XhvLul9CmxiF26s+ktCTDFPVL4F47txF7p8ESbmaS2SOJ6iJlcIE6J/4T5hwiXXPX+uEx6jr0T0ZTmqvFwj79u3pXLxnz9ZMsgVJcUjDKV77JHhqMD1jxtxw9qA/C0THwveU+lpWfbLojEriqPaFk6xwVY8XLsPHilsv7vOoPpKFrgRYhXpQUhv1Q0XX5+/V2Hr0WZDVprd69dbEyldcnbFlxW4X9zlL9Jkvrj937Sr4TGrCHJ9kze2TeOj+0ufGW0jqWPgz6vNW1LK4+6qi6qVcCEAgPoGt+xeaFtZ1vK+d0Xv17smmVcM+prV1Gd+cM8+cMMedy3jrRv3N3uwRdrbyFXbm7yuszFkQAzh+qUWPyD1c4uTiba+4CXt8juYNe9i4lwPNsu0Tzfq9M4wm5ulz7HIrYH5s9h7eYLo1P89l1UQ+4RnQ/fk1cemFy/K6NoXL8WMmjcMkGmpMOWXKR2Wu4s03JwfjvJIWpnN9UlgTPaOmTZvtd1XKUuP05557ucjL2IqsXEJthw5tnegr74iFCz+uyOooGwIQgAAEIACBFCBQ8tFzGRot0UourJrxV0mu3rKKDCeJg/qBroGhrOa8cBfO49dVnsQQiQcS6by7rz+uyTe0T3XquMQsiR/xUry6JZRppmi1SZO6SNCR1afWvbgqsbFTp3ZuMCURc8SIIQkHcpo9XUKqrl8ipgbBYZFn5849TqBTbExZhMnqS7OFSxDywon2KQ6irj8qyapPlpb9+vV0S+WRC5Vc1tUH8URm5VO8Prmuy0KwtEkuROLvBUJZvl122YXOnToZprJGUB9qkKo+lIgmwcxbK8a2UULTtddeHojI+oGh87wVQtR1aFIlDYDFUKKcBHHFFFU6fPiwcw9Wf6sctUfWpD4lap+sXNVOhVGQkK3yxUIxQpVkwaH4kbqn5NKtPLIU1TJekjWy2ipLA5UnHsOGDXSxU+Odk2h/cZ9H30cSjP3EU7rvNUmWt+6M7QPVJ5FVLyXEU30vTj4VV6fPF29Z3Ocs0We+uP5cs2ada7OsPvSZ1DV069Y5aIp4qA9kzexDByikwKc/fWWwHWT+ZEXXq+8wTQIUz7Uv9pxE28XdVzq3X79ehSYZSlQexyAAgbITkEi5df8i0zF7uDm/+7esiHmFnchnr1m2Y6IrfOmO113Myn5W3Dyv+3+bRvXaOSvIeDV7683w8fS0ujaGZh+z6+DqQsKl8my2LuyayVwi5tLtr5uN+2aZTk1HGs1wnp3Z0RWj2ck37D0VxiVcdk1bDwuXZbW6jGKjsaJiV2ss5ZPGCHrhrDA3ijPdvXtnf8gtdVzfzTqucV34RaTGgRqLhJOe8xIj9bzRy/yyJI2F9BxWuzQOCY/r5E2idsmwQG1TXXrea7ypl5Aam2oSQL1c11hIYySNASQY+mea8muc4JPCNSm0isYLqlN9EDu2UT2yaNWkihoLxx73ZcVbdu7c3r1UXbhwiZuMMiqfxoQaV6m9vXp1LzJWTsRYvxl8f/rrCdehMYDGd7o+/3vAH0/U1z4PSwhAAAIQgAAESk6gdkZG9kMlPy36jB49urjBxI4d0fEYNWu4xCcNWCSoKaalXC41sJI7rU8SZjTD70cfzXFxB/3+2KWswc48c6AblMmaUJOKSCSQ5ZSEBomg2j906AA34JKVnMpMJGbFq1sWi5r8Rm4qmoQoM7Outfr8KLDElFVWdnYT55qk65Jrs8QbCYTeai3cfu1TLDtZE3bp0sFZiEqYkMAqoVTCl+Lt9e3bww2+ZD0poUh1eos3tWf48EHWGuCUpVi4Dq1rEKUBmiwAxV+ik0RECaASF31SW2X15ycJUUwnDdxkaag+UrxADdAk3njxVOfKbV+WjRJpYpMsGDW4lTuRBq8dO7a37ZgTWOQVx1SWcBrQFzDv48TdqVNnBO5BsW2UaCZX55Ejz3TtEldZbYpjVNL9qn7q3buHEwGVf+nSVfZfgTWtZnkXAw1+de2ycNSgX7FT1XeJ2uf7TwKYxDAN+OXyKytgz0rt0vXpx4F+PKg+CWzq53BsUd929cGePTluwKx+0Y8jzYI+b95in6VIH4m/6tY1xd73yXwe1UetW/v7vq/7MfP++9ODa4jtA7VRbdfnV/dGy5YtnOuXPhOyWtHx4r4Dou4zf4HFfc4SfeaL60+92NDnVdere0PfBfou0efHT4Dk79nhw4e475327duamTPnRfaX2qxrVTkSqfXCwIfDEAe9iIj93CmvfsitWLE6+O7Tjzzt1/2SzH2l+61+/fquLs+NJQQgUHoCR47nmh25y83+vC3WcjH6Beg2G/9y58HlLtbkuj1T7Yzjb5rjJwryarlx32yzL2+TOXRkl7WQfN0o/66DK2zMzKJutnLt3nFgWRDDUi3XRDsSLrdYoVITBYVTbv4Oe2ylLXu3OWpjcO7IXWZnIl/g6tpmrS9lGSrLTM00LpG1OiaNIXzSS9iwp4vfr2VxwqWOlzTFjm31faznqp7lirOsNGbMCPc810tJPcP1/NP3tfeukYAmAVAePhqXSRDTuXru6Jmn4ATeW0GimARN/52vbeXT8y1R8gKo2uCTxpSXXnqBa6vK10SDGsP6+JR6vsgrQeNXPf9Uh4REvYCXgYGe5ypj8OD+tv1N3TNN49dBg/paz5iCNqkvNO5asKDA+lFjfL0AzM095MZN4qdr1/NPSeMq5ddYSnw0XtC41B/XM9c/I/11xC41rhQ/8dS9oRj04u6TXixKWFT71V7VqTZoPKuUiLH6Z+zYMXa8k+X6QC+tJeYqFJPapeNXXnmx6xOVJw8YjeE0KaVeVibqa98+lhCAAAQgAAEIlJxArczMLiUNu1TyWmLOkKCSmZnpRLAoy0qJMhKUNKtycUmDSA2sJPp4US/2HL1BlWgXjl8Xm8dvF1e3Bi2yhJQQGpX0ZlrCaTJJHMJt1lt2pfAkN9rOyFD7rfNZTLkacLZu3cJMmvShslVIErtE1qrJVqq31F60iz2neKZinhGIObHnR7VRlgDhgWzsObHbap8GpVH3o8rSvROPg/oxUfv0Q0B54t0zOm5vYycux7Yr3rasblVeVHvjnRNvf3GfR52nPhJnMYpKUX2Q6LOQTJ1R9fh9icpWHrUn3mc+UX+qXeHPZDdreakfRi+++B9ftVvqe0f3TLL3mMrVj9PYz3ChQku4Ee++UtvK474oYXPIDgEIpDCBrHptTEZaPSNhVDE4q1u65ZbrXZO9OKnv5YExcSz9MWWMOq79f/nL81qUKEn8k4Cll3VKetGrl2FyGffjAolvEsn8M0GWjhLwfJiiG2+cYPTiz3vryPJS5Snki0Q2vZhT+1W2PFT+/e83g/okxEl4C7+ojLqAKLdxWVPKUlQvcpX0fJAHiF5UK7yORFJx/Oc/XwueTwrvc8UVF7kxuA+jc801lznPHT8+lXjbqlVzOyHnFBeySfmfeeYFV4e8EZYsWRG89PNC4rPPFriWy8tIoZh8SCFZSH7qU+cYHdezK8zDFRjzR2LpVVdd6p7L4j169Ag7RjtqjRNOhem5/vrxbpJKGTUo6RxZneoFstYTMZbQqReuYXd+vcTWbw31oTxKdP5zz70S/K5QX0so1Vg9UV/HXAqbEIAABCAAAQiUgEB8H9USFFLSrBIHooJry2pOb3Y1kCpukObr1EAnqix/XEs/uAzvi11Ptm4NTOKJUCozWXFCFnkaHMkiUm+nZTHauXNHI8vC2BRrMeePa4AZz6rQ5ynrMhl2ydQRT7jUucUzlXAbLZrp/Kg2+h8QOp5MStS+4srS/ZyofRI+7f9xUzKieuzJ8UTE2HzJbMf7PIbPVR8lqjOqDxJ9FpKpM1x/7HqispU3qj2+jHj9qR+TmqBJ1tmygtYPFE0UFfUZ0/dOvHJ8PeGlrre8U7z7CuGyvElTHgSqP4EDedEeCNX1yjQRj5IXMJMRLstyrRIXvUWjxC9ZE8oaUWM4JVmDSniUG7a8iTp0aGf3nrINkJeErPfkbq5wPl64cyeH/ihcjwRDL5TqkJ5BUc+h0GlxV+UdIAtFjTl9UtgaH75J+/bts5NKxbwc134vXGpdHjfhbQm13bp10qHItGvX7mC/vLH0AlQv/FSOJtjUuiwsZd0p8VIvJPXSMdH42hcol3WV6Z/BCvUjy9eZM+e78aTc3SU2b9hwygJVvxOmTCkYXxfHWEKxXnJK2PVJXjc+rr7uBfXPRRed60Rg9c12a9XqU7J97fOzhAAEIAABCEAgOQJVIl7Ga5oEPL3NlTuLXEEqM1V23bpGDYAllsiKTm/GZ82aF7ggJXPtcpOJN7lLMueTBwIQOEVAPyblHi4LC/341A8juYTNnbvwVCbWIAABCECgygh4sVLipRcw1Ri/Hs/isqwN1vMhPC7V+tVXj3OhRSTInX/+KBsGpJlzvVZYGQmU7dq1CqqVxZ9iVsuaX+GMJATKcjP2xanGg7LGLK8kq/8GDTKtZ0dWUKTEyrAQGRwoxxUfF1tFxr5Qk5u83Orlpi4hU89ZxdxMNqmPlW6++Vq3lDWpkiw8VZaMESTGRgmyylccYx2XeKm44j6pLC9e63peffUtJ2Ar9JPGDIrR7ycHTbavfdksIQABCEAAAhBIjkBKiZeKd+Nj3iTX/PLLVdl1y2pq1qz57l9pr8LHpyzt+ZwHAQgUJqDYu+H4u4WPsgUBCEAAAlVNIJ6AWVHCZdT1SnTUOE7CoKz8JL698MKrgeVg2NJR58vyUOKX/sll+qKLzrNxq7sErtW+DsWclMVfOPyHrP5kTRjPWtOfG7WUK3OsQUC47KhzKnqfjBQ++GB6YE3qLRqTqVdWkUoSD8MeFhIQJQxLvJRlpCw5JT7KQlRJ16wY0+JbHOPdu/fZl5d5gVWtPz8swmpdbvf6J3fyCRMucfHfxTvZvnYN4w8EIAABCEAAAkkTUCA2EgQgAAEIQAACEIAABKoFAQmY3gJPDa5o4VKxzuXqrH+Kb6kJZxQORBP3SciUYOVdseUKXdjlOMNce+3lzvJSbZVloMqJCgmkkCUSRSXGKa6xhM7Ro4cVcvtWGckmTSKjyWm8QCivAlmMygq0qpJ4KSyLBEW5iisWZbJJAqUEYAmUsnj1/zS5niwvvdWlJr7UDOiyoFSStacm0pFFaHGMZczQqVM7N6Gm2piRkeEm8FGfKMnNXf0p7yklMVX/qz91Pcn2tTuZPxCAAAQgAAEIJE0gpSwvk241GSEAAQhAAAIQgAAETlsC3gJTAMLrFQFE1oL6p6SQInK7njhxknNN1rYmxFEMRAmPR48ec15Ein+pJAvBmTPnWUFykJtcRq7ccjOWRWRs0vnvvjvFnH32cHPDDePtYc1CvqPUXjqKsSnLUM2eLWFNItvSpasCq8fY+itje+bMuTZG5chglnWFa1FszuKSuClOpnfPDufXrOWKH9+lSwdnDak8o0YNdxaRYnjo0CHz3nsfOhf24hjv3bvPWoZ+5ARk9YPqVRiAxYuXuSoVb1OisCbtEU/FNp09e4GbxEkZku1rVxh/IAABCEAAAhBImkCVzDaedOvICAEIQAACEIAABCBwWhPws42XFUJpZhtPtk6JXJmZmc4aMOxiHD5fFpealEaT4BWXZPGnciS2lUeSFaKsHuO1rTzqSLYMWTTK3frw4cPOgjXZ80qaT+KiRNt4Ew4Wx7jAklOTLh6PrFpM/cRBsRlK0tex57INAQhAAAIQgEBRAlheFmXCHghAAAIQgAAEIAABCCRNQG7kmtU6UZJ4mGyKcitP9tyofPFEtqi8Fb1PAmpxrMqjDRKJ4wmXKr84xvn5+QmbkYhpSfo6YSUchAAEIAABCEDAESDmJTcCBCAAAQhAAAIQgEDKEti2bUeZ21YeZZS5ERQAAQhAAAIQgAAEIFAqAoiXpcLGSRCAAAQgAAEIQAAClUFgwYKPTVnER52rMkgQgAAEIAABCEAAAtWTADEvq2e/0WoIQAACEIAABCAAAQhAAAIQgAAEIAABCNR4Alhe1vgu5gIhAAEIQAACEIAABCAAAQhAAAIQgAAEIFA9CSBeVs9+o9UQgAAEIAABCEAAAhCAAAQgAAEIQAACEKjxBBAva3wXc4EQgAAEIAABCEAAAhCAAAQgAAEIQAACEKieBBAvq2e/0WoIQAACEIAABCAAAQhAAAIQgAAEIAABCNR4AoiXNb6LuUAIQAACEIAABCAAAQhAAAIQgAAEIAABCFRPAoiX1bPfaDUEIAABCEAAAhCAAAQgAAEIQAACEIAABGo8AcTLGt/FXCAEIAABCEAAAhCAAAQgAAEIQAACEIAABKonAcTL6tlvtBoCEIAABCAAAQhAAAIQgAAEIAABCEAAAjWeAOJlje9iLhACEIAABCAAAQhAAAIQgAAEIAABCEAAAtWTAOJl9ew3Wg0BCEAAAhCAAAQgAAEIQAACEIAABCAAgRpPAPGyxncxFwgBCEAAAhCAAAQgAAEIQAACEIAABCAAgepJAPGyevYbrYYABCAAAQhAAAIQgAAEIAABCEAAAhCAQI0ngHhZ47uYC4QABCAAAQhAAAIQgAAEIAABCEAAAhCAQPUkgHhZPfuNVkMAAhCAAAQgAAEIQAACEIAABCAAAQhAoMYTQLys8V3MBUIAAhCAAAQgAAEIQAACEIAABCAAAQhAoHoSQLysnv1GqyEAAQhAAAIQgAAEIAABCEAAAhCAAAQgUOMJIF7W+C7mAiEAAQhAAAIQgAAEIAABCEAAAhCAAAQgUD0JIF5Wz36j1RCAAAQgAAEIQAACEIAABCAAAQhAAAIQqPEEEC9rfBdzgRCAAAQgAAEIQAACEIAABCAAAQhAAAIQqJ4EEC+rZ7/RaghAAAIQgAAEIAABCEAAAhCAAAQgAAEI1HgCiJc1vou5QAhAAAIQgAAEIAABCEAAAhCAAAQgAAEIVE8CiJfVs99oNQQgAAEIQAACEIAABCAAAQhAAAIQgAAEajwBxMsa38VcIAQgAAEIQAACEIAABCAAAQhAAAIQgAAEqieB9OrZbFoNAQhAAAIQgAAEIHA6EGhav4vp0WKs0bI0ac+htWblzneNliQIQAACEIAABCAAgepHAMvL6tdntBgCEIAABCAAAQicNgTKIlwKkhc/TxtgXCgEIAABCEAAAhCoYQQQL2tYh3I5EIAABCAAAQhAoCYRKK3FZZhBeZQRLo91CEAAAhCAAAQgAIHKI4B4WXmsqQkCEIAABCAAAQhAoJIIlKebePPmTc2ZZw4wrVu3rKTWJ1dNenpt07VrJ6NldU+6jnr16ha5jIyMDHeNtWrVKnKsPHeoftVVkalVqxYmO7txRVZB2RCAAAQgAIEaSQDxskZ2KxcFAQhAAAIQgAAETl8Cq3ZNMjPWP2W0LEvKympoHnvsO+bPf/6N+epXP2+eeea35vHHHzUtWjQLiu3evbMZO3ZMsF2WlQkTLjESuJJNTZtmmyef/LkVxJoke0qZ85W0jclUWLt2bXcd3/zm7UWyt27dwh3LzKxX5Fi8HSVto69fdVVkuuOOW8wVV1xUkVVQNgQgAAEIQKBGEkC8rJHdykVBAAIQgAAEIACB05OABEtN0FMe6ctfvsmcOHHCXHPNbebWW+8xEyZ8wRw6dNjcffdtQfF9+vQ048eXjyB1003XmnbtWgdlp+JKRbbxggtGm/PPP7vMl12RbSxz4ygAAhCAAAQgAIESE6idkZH9UInP4gQIQAACEIAABCAAAQhUAgFN2JNsCguXOq978wuCU0tjhfmlL33WvPTS62b16nWunKNHj5n58xeb3NyDZt26jc7i8t57v2patmxu2rZtbQ4cOGi2b99p3bjTzeWXX2g+/ekJRoKc3KFXrSooQwXdffeXzM6du821115uxo270MyYMc+onH79ejn3dJU3a9Z8V2eiPw0a1Hdl/POfrzlR9aqrxplmzbLNxo2bg9Nuv/1zJifngNmzZ6/R8caNs0yPHl3MzTdfa3r16m72799vj+1z+e+881Zz8OAh1zZfwIUXnmOGDOlvGayP20a149JLzzc33HClGTToDLN163ZXp8qQu/3o0cON8ohnXl6+2bx5qy/eLdPS0lx7dB2f/ew15p13PnDXo4Nq74QJl5rnnnvZiL9S27atrGB8sb2eS50VrOpTuXXr1i1VG339Eye+58q98spLXLkbN24xR44ccXXqj65h3LgLzHXXjTfdunW217jf7NuXExyXa/t5542yx68wQ4cOsML3SbNly7bguI7t3ZtjFiz42O3r3bu7+fznbzBLl6507Q8ysgIBCEAAAhCAQCECWF4WwsEGBCAAAQhAAAIQgECqEkgUxzKRcFna65k6daa5+upxTpz0ZUh0nDx5mtuUSLh48TK3LrFx167dbv3b377TnneZWbJkufv3xS/eaAW4S3wR1rpwlHnggbuM3JU/+miOOX78WCBWzp69wIpbS4K8JVkZOLCvFdU6FTpFwmHLlgVu7jr+hS/cYF2XP2XmzVtkMjPrml/96ofBORkZ6ebKKy8udP7NN19jDh/Oi9tGCbU/+tF9ZtSoYWbOnIXm2LFj5oknfmo6d+7gyunYsZ0TGT/zmavMokVLzbZtOwqVH94Q15kz55lvfetrJl6MS8Ud/e1vH3Zi8dy5C83Agf3Mz3/+fScsxuNYXBt9G771ra+ahg0buOsYNWqo+fGP7wviYCoe5sMPf9ucc85I1z9ZWQ0cu+7du/jTzW23fdZ88YufNuvXb7Ki5n53HePGRYvvPXt2NY88cr8Vruc6QTMohBUIQAACEIAABIoQSC+yhx0QgAAEIAABCEAAAhBIMQJenIy1qFQz/TGtRx3X/tKk55//txX46pn//d9fmzVr1jth7a233nfWlSpvxYo11kpwiiv63XcLltp47bW3zaZNW6yYuccdkyh41llDzSuvvOG29efNNyebF154NdjW+YqJKKvDuXMXBfvLe6V+/Uxz113ftRaFR13Rhw/nmy9/+WZz330/Nro2xfisW7eOyc8/YiTMtWzZwom1x44dN1FtlCVi7dppVox91LnYq9Djx4+bW2653vzgB79wdcjyVOX7Ot3OOH9++9unzFNP/cKJvS+/PLFIrq985Wbz4YezrHD4R3fsP/9514qXD1or1yvN008/V+o2qjBxf+KJZ1y5b7012fzpTz83lkqAVQAAIR5JREFUF144xkycOMlaXI51wubtt9/nBFplkkB51123Op4Sa2UJettt91hry+2uDInQP/3pg7ZNUy3PfLdPfxQn9dFHv2N+85unzJQpM4L9rEAAAhCAAAQgEE0Ay8toLuyFAAQgAAEIQAACEEhBAopnGXYBryjhUpcu9/DHH/9f87nP3emEvQED+pi//OU3xU7QI9dyWfBdcsn51hLvRhfHUe7c4SThsyqSBLqwiPjhhzONrACVZEUqN2gJrUpyeddxuZLHS4r5qWuTZamsVPVP196lS8fglHXrNhWqMzgQsSLmP/3p486KUVabsal37x7m/fcLLF917OTJk3Z7utH+eCmZNurc6dNnB0WIkaxpe/QoYNOnTw8nmsqy1CdZiuq4ZnvX8ZUr1wTCpfLIPTw3N9e56ftz5Kr/xz/+zPzrX2+4dvv9LCEAAQhAAAIQiE8A8TI+G45AAAIQgAAEIAABCKQIAcWv9PEvvYBZkcJl+LJ3795rXn31LRtP8UfOWu6//utLRnESo5L2y41arsyK/6hYjNHxK09GnV7h+w4ePFyoDgmT9evXdwKcDrz99gdObJXbttzbZY2ZKGVnNzbNmzcz7du3Cf4pTuS0aaeEQCsxJiqiyDEJrK+//q65//67XPxQn0FtatSooY3Tmet3uaW2mzRpVGhfeCO5Nhor0hZl06hRlitK5R84ULhebdepk+Gscxs3blSkXToxtm0jRw4xy5evcoxlBUuCAAQgAAEIQKB4AriNF8+IHBCAAAQgAAEIQAACKUDAT8Aj8TI8o3h5uor7y9TkLF/72hespeXzgZu4jmlyFblB659mHo9Nmshl2LBB5vrrv2yFqwPusCbKST7Vipu1f/8+zjVbsTSVJKzJRduLebJabNOmZXC+XN5btGgebGslNiamLAc1+ZDcwpXefvt96379KzN8+GAXk1MxOIumU22UtaHc43/726eDbIrlqXaVJT355N+sheJPrdXrdUExsrJctWqtsxTV0qeePbs5F36/XbAseRvlzh0uV2wWL17qilu+fHVgoerrUb1bt+5wEzWtWLHaxQuVeK0Z6pV0D2n2+LCV7bPPvmTvqRes2/sPrCv9181DD/3cWY/6MllCAAIQgAAEIFCUQPQr46L52AMBCEAAAhCAAAQgAIEqJxC2wFRjKkK4VLmySJTr81133eZmnta+du3amDvv/KKdcXxJIFxKuGvfvq3LI/dhzeqtZZcuBRPW6BzFQkwmacKfAQN6O2E0Kr/Etfvvv9MJYhJPVe7KlWuDeIoS0EaPHmHOPnuYad26hYvHGCsiDhrUz1x88XlOmOzUqb259dbPmEmTTsXrVLxGlXPPPV9x8SO9EOfbE9tGuU6PGTPCTWQj0VLWhJrYRrOclyXJbfuRR37nJgIKl/PGG+/ZSYc+7eJxyhJTs5mPH3+Rsxj1+UrbRk0qJFd1XYcYiZXiayrpOuVOrxnDdVyWpopR+sYbk9zxjz9e4axsNaO6GMhS85vfvN25ju/Yscvl0R/NmC7Xc4mWffv2NDfeeFVwjBUIQAACEIAABKIJYHkZzYW9EIAABCAAAQhAAAIpSsBbYKp54fXybu4DDzzirOP+9rfHnXWchEC5gP/sZ78PqpozZ4FZtmylee65P7j4mJpk5umnn3UTteTl5bmZuidPnm6GDh0QnBNv5c9//odzlb7yykusMPnFItlee+0dJ9o9/fQvnYAmi8kHH/xZkE+u1nJV//73v+kmk5H1oqw1w+mVV950sTjvvvtLzjpSYuA//vHvcBbnKi7xUpPWxKbYNq5evd6Klb8xmkhHloSa2XvWrHnm739/JfbUEm/LqvOZZ553cUP9yZo8R27gjz76gLNs3LNnn/n1r5+0gvJin8WUto1//vPzbgZxWaxKhH744V+bDRs2u3LXrt1gZx//tRVlb3aziMs1/o03Jpu//e1Fd1zhAe6//2HXfy+88KTbpzZJpIxKCkXwwx/+0t5L33OxMqNDC0SdyT4IQAACEIDA6UegVmZml5IFoTn9GHHFEIAABCAAAQhAAAJVRODSPg+XS80Tl36n1OVIkGvevKmRBV2sJaIvVNZ2mqHbWzpmZGTYiWyaWJfzXSVyC5Y1ocpKNEmOLDslsB04cNBXX2ipOIzhSXn8wYce+m83a/ozz7zgLAPlZu7dxX0eLTXRkCbg0czaUSleGxs3zrLWh0cCS9Coc8tzX5Mmjd0EQ1FllraNOk/iqETReElWlTk5B+L2q/pP94HuBxIEIAABCEAAAmUngOVl2RlSAgQgAAEIQAACEIBADSYgN99t23YkvMLY+JdHjx615+xMeE7UQcV1TCRc6hwJjvGESx2PEi61P5z27dsf3nTrEh81M/ctt1xvrUefK3Lc74jXRgl6lZk0M3q8VNo26rxEwqXqi2IXbkfsvRA+xjoEIAABCEAAAiUnQMzLkjPjDAhAAAIQgAAEIACBSiKw59CpiVlKW2V5lFHaulPpPIlysTNmh9s3eHB/66L9afOf/7zj4l2Gj7EOAQhAAAIQgAAEqooAbuNVRZ56IQABCEAAAhCAAASKJdC0fhc3KY+WpUkSLjUzOQJmaehxDgQgAAEIQAACEKh6AoiXVd8HtAACEIAABCAAAQhAAAIQgAAEIAABCEAAAhCIIIDbeAQUdkEAAhCAAAQgAAEIQAACEIAABCAAAQhAAAJVTwDxsur7gBZAAAIQgAAEIAABCEAAAhCAAAQgAAEIQAACEQQQLyOgsAsCEIAABCAAAQhAAAIQgAAEIAABCEAAAhCoegKIl1XfB7QAAhCAAAQgAAEIQAACEIAABCAAAQhAAAIQiCCAeBkBhV0QgAAEIAABCEAAAhCAAAQgAAEIQAACEIBA1RNAvKz6PqAFEIAABCAAAQhAAAIQgAAEIAABCEAAAhCAQAQBxMsIKOyCAAQgAAEIQAACEIAABCAAAQhAAAIQgAAEqp4A4mXV9wEtgAAEIAABCEAAAhCAAAQgAAEIQAACEIAABCIIIF5GQGEXBCAAAQhAAAIQgAAEIAABCEAAAhCAAAQgUPUEEC+rvg9oAQQgAAEIQAACEIAABCAAAQhAAAIQgAAEIBBBAPEyAgq7IAABCEAAAhCAAAQgAAEIQAACEIAABCAAgaongHhZ9X1ACyAAAQhAAAIQgAAEIAABCEAAAhCAAAQgAIEIAoiXEVDYBQEIQAACEIAABCAAAQhAAAIQgAAEIAABCFQ9AcTLqu8DWgABCEAAAhCAAAQgAAEIQAACEIAABCAAAQhEEEC8jIDCLghAAAIQgAAEIAABCEAAAhCAAAQgAAEIQKDqCSBeVn0f0AIIQAACEIAABCAAAQhAAAIQgAAEIAABCEAgggDiZQQUdkEAAhCAAAQgAAEIQAACEIAABCAAAQhAAAJVTwDxsur7gBZAAAIQgAAEIAABCEAAAhCAAAQgAAEIQAACEQQQLyOgsAsCEIAABCAAAQhAAAIQgAAEIAABCEAAAhCoegKIl1XfB7QAAhCAAAQgAAEIQAACEIAABCAAAQhAAAIQiCCAeBkBhV0QgAAEIAABCEAAAhCAAAQgAAEIQAACEIBA1RNAvKz6PqAFEIAABCAAAQhAAAIQgAAEIAABCEAAAhCAQAQBxMsIKOyCAAQgAAEIQAACEIAABCAAAQhAAAIQgAAEqp4A4mXV9wEtgAAEIAABCEAAAhCAAAQgAAEIQAACEIAABCIIIF5GQGEXBCAAAQhAAAIQgAAEIAABCEAAAhCAAAQgUPUEEC+rvg9oAQQgAAEIQAACEIAABCAAAQhAAAIQgAAEIBBBAPEyAgq7IAABCEAAAhCAAAQgAAEIQAACEIAABCAAgaongHhZ9X1ACyAAAQhAAAIQgAAEIAABCEAAAhCAAAQgAIEIAoiXEVDYBQEIQAACEIAABCAAAQhAAAIQgAAEIAABCFQ9AcTLqu8DWgABCEAAAhCAAAQgAAEIQAACEIAABCAAAQhEEEC8jIDCLghAAAIQgAAEIAABCEAAAhCAAAQgAAEIQKDqCSBeVn0f0AIIQAACEIAABCAAAQhAAAIQgAAEIAABCEAgggDiZQQUdkEAAhCAAAQgAAEIQAACEIAABCAAAQhAAAJVTwDxsur7gBZAAAIQgAAEIAABCEAAAhCAAAQgAAEIQAACEQQQLyOgsAsCEIAABCAAAQhAAAIQgAAEIAABCEAAAhCoegLpxTWhUb02pkGdFqZuepZJT6tXXHaOQwACEIAABCAAAQikIIFjJ/JM/rED5uCRnWZ/3tYUbGHyTWrXeIhpmdXbNK7XztRLb2Rq1eJ9fPL0yAkBCEAAAhCAAASqnsDJkydM3rH9Jidvs9lxYJnZnDM3bqNqZWZ2ORl1VKJl0/pd7SA31+Tmb7cF5phjx/PMSfsfCQIQgAAEIAABCECg+hCoZWqZ9Nr1rNDX2DSs28q+lG5o9hxaU+1EzPZWtOzeYqzJzGhSfeDTUghAAAIQgAAEIACBYgkcPrrPrNz5bqSIGSletmjYyw1udx1caQ4f3VtsBWSAAAQgAAEIQAACEKg+BDIzsk3zBj3cy+mducurRcP7tr7CdMoeWS3aSiMhAAEIQAACEIAABEpHYP3e6ebjba8VOrmIeCnhUu7h2/YvxMqyECo2IAABCEAAAhCAQM0hIGvM1o0GGLmTp7qAiXBZc+47rgQCEIAABCAAAQgURyBWwCwUIEiu4nInQrgsDiPHIQABCEAAAhCAQPUmoFBAGvNp7KcxYKqm9k2GYHGZqp1DuyAAAQhAAAIQgEAFEOiUfZZRjHOfComXinEpV3HiWno8LCEAAQhAAAIQgEDNJaAxn8Z+GgOmaurR/MJUbRrtggAEIAABCEAAAhCoIAI9bJxznwLxUm/cNTkPMS49GpYQgAAEIAABCECg5hPQ2E9jwFS0vpTVZb2MxjW/E7hCCEAAAhCAAAQgAIFCBDRBo7e+DMTLBnVauFnFC+VkAwIQgAAEIAABCECgxhPIzd9uNBZMtdSyYe9UaxLtgQAEIAABCEAAAhCoJAItswrGgoF4WTc9y804WUn1Uw0EIAABCEAAAhCAQIoQyDuWYzQWTLXUuF77VGsS7YEABCAAAQhAAAIQqCQCjeu1czUF4qVmGD92PK+SqqcaCEAAAhCAAAQgAIFUIaAxoMaCqZbqZjRKtSbRHghAAAIQgAAEIACBSiJQL71gLBiIl6qXiXoqiT7VQAACEIAABCAAgRQikJpjwDRTy/5HggAEIAABCEAAAv+/vTuB67LK9zj+BQFBAQFFVFRw33CprMlpWiZtc5nUpptZtlxb7ZbVbby3a814M5spG6tbOtpUpqk12eaSzkxq26SWS2654YaIouICqCAg3HMO8//HPzFsEnmIz/EF/2c5z3PO//306vW8fvzOOQjUTIGgoNKwZUDwsmZS8K0RQAABBBBAAAEEEEAAAQQQQAABBBBAwIsCBC+9+FToEwIIIIAAAggggAACCCCAAAIIIIAAAgiI4CX/ESCAAAIIIIAAAggggAACCCCAAAIIIICAJwVCPNkrOoUAAggggAACCCCAAAIIIICABwXqRoWpVYc4NUmK0sGsPG1Zd0AH9+dVak8bNY3UgNs7asXnu/XVJ7sqtS1ujgACCHhNgOCl154I/UEAgXIFYmNj1KPHeWratLEOH87R+vWbtW7dxnLrnu7BQYOuVcOG8f7qmZn79Omni7V3b5b/2I/ZSEhooBtu6K9ly1ZpyZLlP+ZWXIsAAggggAACCCDgAYH+t3ZQ38HtAnpSUiIt/minXh+3Unb7TJQBt3VU605xeu7RxSoqKlaLdrE67xdNVKtWMMHLMwHMPRBAoFoJELysVo+LziJQMwW6du2oyZOfV0xMdADA3LkL9PDDo8wLXVHA8dPdufHG/urcucNJ1SdOfEPPPDP+pOM/9IANjD744B2aMOF1gpc/FI/6CCCAAAIIIICAxwS6XNDIBS4P7D2mD6Zs0OovM9W2cwP16t9SF13ZXJvWZLkg5pnodmJylNp1aaDgWkGSedVdsjBdWZnHlJGWcyZuzz0QQACBaiVA8LJaPS46i0DNE4iLi9Frrz2nsLAwPf30eL3//nw1aZKgxx57UH379lJ+/nGNGDH6X4YpLi7RzTff566Pi4vVLbdcr3vuGaKFCz/XihVr/uX7ciECCCCAAAIIIIDAT0ug96C2OlFUorEj/uECifbbrVqyR1s3HNTYaVfpl/1a+IOXVwxsre6XNFHT5Ght33RIH8/d7oZ822uu+nVrXX9nij58c7Nad4xTUpsYZezI0eRxXyszPVdPvtpLdpi4LRNm99OKf+zW/L9s1n+Nu9hd8/7r6zV+Vj9lbM/RjtRDOveiJio+UeLa/mDqBnfdA0/0UJefJejuPrNNn4vdsWdnXK28o4V6/M6Fbj/BtGEzSdt3jVfh8RNau2yv/jJprQrMNgUBBBDwkgAL9njpadAXBBA4SeDmm69TbGw9/e53YzVp0hvaty9Lq1Z9o0GD7tXWrWm67rreatAgzl2XlNRUL744xgzTnm+Cj2/riSdGnJSteVID5sDSpSvdz7x5CzVlytuuStu2Ld2nDZ6OGHGfu9/WrUs0b940paS0D7jN7bcP0jvv/FnffPOJO3/VVZcFnPftREVFav786abfSzRgwDW+w3wigAACCCCAAAIIeFwgom6oG8a9eW2WP3Dp63Lu4eP6/UOfaeaf17lDNsh5w90pqt+wjr4081MmtYnVPSMv0Dk/b+zO+4aWX319G2WZLM6dW7PVygQxbxnezZ3fvvGQjuQUuO1UM5/m7h25ZttkYNrfpR+y49Nbdoh1gcevF+9RRGSo+t7UzmVruooV/IqMDtMjT1+k7hcnKtV8p20bD+qS3sm6/38vrOBKTiOAAAJnX4Dg5dk3p0UEEPgBApdf/gvl5eVr9uy/B1xlh4oPHfqwbrrpPhUWFroA54wZE9S79+Vavny1srIOmozKgW64eWhoaMC1p9qx82oOGfJrd3rbtjT3OWXK/7lMzCDzpvjBB39Vq1bJ5p7PqU6dCHd+2LDb9PjjDyo5uZnrY7NmTcww8d+re/euAc3YwOX06ePVrl0rjRz5B5dBGlCBHQQQQAABBBBAAAHPCkTH1nZ925txpNw+pqUe1ua1B2SDnNcO6eAW8Bk5dIGmPr9KT9z3sct+vOHuzu5aXwBy3lubNfmPK/XMI59r4+osMwS9vsIjQvTq2BXa8s0BV/e5/1msWW+UZlN+t+HjeUUaM/xTTX9ptV787RJ3uuuFjb5brdx9mxka2yBCk55apgmjv9LEMcv0t3dS1eGceBMULU0MKPdCDiKAAAJVIMCw8SpAp0kEEDh9AbtAz65de1yA0l5Vu3aYzjknJeAGwcHBuuOOwWrcuKFGj37eBBffcuefeupRk6F5rfr1u0LvvTcv4BrfTnBwkLZtW+rbdZ82UPrll18rPLy2yaRc6IaQv/DCK+YP3CXasSNdDz10l7p1S9GaNes1fPgdSkvLMBmgQ3Xw4GGXHbpo0UwTOL1Or776prtfnTp1XOAyJaWdRo36o956a1ZAe+wggAACCCCAAAIIeFzgNBfiad6qnmqFBGnJgnTZ4KIt+/ccdUOybeZlVL3SIKg9boeb+8qenbkmi7KBomJqK/+f1/nOneozIy3X38ae9NKgatn7n+o6e9xmetpih7r/sm8Lt23btsV+h21l+uYO8gsBBBCoQgGCl1WIT9MIIFCxQGFhkRke4xsfIyUkxMtmWJYtd975iAlodjYL95ww597zn5o27V0XvLQL/pwqeGkDkjZQaUu9elFq3761+vTppZkz5+iLL5bp5ZffMNmcvcxw9Cdl58S0mZO2xMXVU8eObRUaGmKyKOe5wKU9boObLVuWDrfxLQZ0223/Zk+5YObUqTPdNr8QQAABBBBAAAEEqo/Awaw8N99l4+ZR5Xb6MhMArGOGbtsMTFuyD+YH1Dt8oHS/dngt/3E797qv+F537R/WT6uYC0rKXO+7xi3w49v5ns+wsFpunkx92wXZ4e+bzM+xI4XfcyWnEEAAgbMvQPDy7JvTIgII/ACB9PQMderUTnXr1tHRo8e0d+9+DR48zN1h0KD++tWvrlR2dq7LyMzJMX99Pl46P5CtYOvaEhFROsTb7Xznl51zyHc/e8pmdb777itmSPpgrV27UW+/PUm++S/tXJs2u9LOwWmLzcy0xbZ7OqV//6s1ceJUHThw6HSqUwcBBBBAAAEEEEDAIwJ2QZuVi3fr/EsS1em8hvpmxT5/z+ww68HDumjd8n1aZeaftKV1p/paNHubv44dEm4X+8k+FBjU9Fc4wxtF/1ykJzQ02A1ZDw0LVl0TXPUFJg/sP6bkdjF6+Q/L/YHWJknRZih5uBvCfoa7w+0QQACBHyXAnJc/io+LEUCgsgXmzl1ggo/huv/+f3dN2eCkXWBn+fI1LkuyoKBQmzZt1Z49e002ZIzJekzyd8k37+Tu3Zn+YxVt2IDl0aN5bqEfG8i0gcspU2a6bMqBA4fKZnPaYrNB09N3u+0ePbq7T/srJCREDzxwh1sJ3XfwlVdmaMyYF1S/fqzJ5Bzr6vjO8YkAAggggAACCCBQPQTenrTOLaRjF7UZ8kA3s8p3Y9nFee777c9cgHCWWel7txn+vXNLts6/NFE9+7dyC+jYuS4Tzarjyz7dpcKC0pW/K/rGBcdL613ap4WatoiuqPpJ5+2q5bbc89gF6nx+gu43q4+H1q7lpkGyx+2wdpvlafve+YIE18/Bwzrroad+rsbNSlc6t/UoCCCAgBcEyLz0wlOgDwggcEqB6dPfc9mVd911s1ss58MPF6hRo3hdf30/tWjR3AQDpyk394hbAKd3754aN26Unn32TwoLCzML6TzkhpLPnfvRKe9vT1x44bnuvA0uDhzYx2R5Rrgh477MzebNE2UDlE2aJJjA5FBX1w433759p1k9fJGuueZy2YV7Vq5cY4KWV5hMzgF65pnxbi5MW7mgoMANGbfBVLsS+ahR/6nHHnva3YdfCCCAAAIIIIAAAtVD4JAZOm4X0PmPURfq0j7J7sf23A7/fu3ZlUrbUjpk/E9PfqX/HneJbry3dIEeW2fX9hzNmLDGbp5WWbowXRdcluhWLV+9NFNzpm9019lRQ6dTPp6zXRdfnaSU7g3djx22nnfs2+mY1nyZqQ/f3Kw+N7bV8NE9/Lf8ZO5211f/ATYQQAABDwgERUS0cP/7a92gp7ZkLfRAl+gCAgggEChgh2lPnvy8unTp4D9hMy5tFuTYsRP8Q8V/85thuvfeW/x1bIBxxIgnzTDwD/3Hym7MmjVZvnkpyx7/6KPPTDDyUZ04cUIvvTTGzHnZs+xptz18+OOaM+cjRUdHmQDmNLNYUIK/zurV63XrrcOVlNRUto0JE143AdWJLoN09uwpJgib5Pr1zjtz/dewgQACCFS1gPfeBYN1TYfRVc1C+wgggMBJAnZeyU7nNlQzs7CNzaRc8XmGW128bMXI6DBdfE2y7HDtozkF+uyvabJDz22Ji49QfOO6St+W7R/GnZAYqZj64dq26ZC/XnLbWNNOvDas2q/dZnGepDYxOrDvmLIyj6lNSn0dzy9yWZ72nrVCgtXaLMKTY+astIv/+I7ZrEs7F+eqJXtMRmWUeb81C1Bu/nYKIzsE3i3eY6ICm9dmMWTcyfELAQS8JDB/w0gRvPTSE6EvCCDwvQLdunXSeed10eHDOfrkk8Xlzh155ZWXqkOHNu4+ixZ9Yeat3HDKe6aktFdkZJ2A8zaLc/36VP+QGruS+cCBvZWY2Mitem4X8UlObqrU1O3+9m3Gps0ELS4u1mefLdXGjVvcPe08nZ07t1dGRqZ/iHnDhg3MEPTmJhuz0GRqrg1omx0EEECgKgUIXlalPm0jgAACCCCAAAIIlCdA8LI8FY4hgAACCCCAAAI1UIDgZQ186HxlBBBAAAEEEEDA4wI2eMmCPR5/SHQPAQQQQAABBBBAAAEEEEAAAQQQQACBmipA8LKmPnm+NwIIIIAAAggggAACCCCAAAIIIIAAAh4XIHjp8QdE9xBAAAEEEEAAgZopUKwS84+CAAIIIIAAAgggUDMFSkqK3RcneFkznz/fGgEEEEAAAQQQ8LzA8cIcz/eRDiKAAAIIIIAAAghUjkB+Uem7YEDwMkhBldMad0UAAQQQQAABBBDwrIBX3wGz83d51oyOIYAAAggggAACCFSuQHZ+hmvAH7wsKs5XSK3wym2VuyOAAAIIIIAAAgh4TsC+A9p3Qa+VfUc2eq1L9AcBBBBAAAEEEEDgLAnsyy19F/QHL48X5So8pN5Zap5mEEAAAQQQQAABBLwiYN8B7bug18quwyuVX5jttW7RHwQQQAABBBBAAIFKFsgrPKyM7JWuFX/w8mjBfkXWTqjkprk9AggggAACCCCAgNcE7DugfRf0YknNWuDFbtEnBBBAAAEEEEAAgUoUSN2/0H93f/AyJ3+PaodEKiI01n+SDQQQQAABBBBAAIGftoB997PvgPZd0IvFZl+mHVrqxa7RJwQQQAABBBBAAIFKEEg7tMSfdWlv7w9e2p2Dx7apQd02ZtkeFu6xHhQEEEAAAQQQQOCnLGDf+ey7n30H9HJZnzmHAKaXHxB9QwABBBBAAAEEzpCADVyuz5wbcLegiIgWJWWPxEe2U0hwuDJz1qjE/KMggAACCCCAAAII/PQEbOCyUXQXt1DP/iObqsUXbFrvXLWO72lGCsVUi/7SSQQQQAABBBBAAIHTE7BzXNqh4r55LstedVLw0p60AUw7cXvW0VTlFR4qW59tBBBAAAEEEEAAgWouYIeK24zL/KJsVZfAZVnyRBPEbBjVXvXCE807a7SCggIGE5WtyjYCCCCAAAIIIICABwVKSorNu2iOsvMzZFcVLy9o6et2ucFLezI6vLHi6rQ0K08e0ZHje93LbdGJfLIxfXJ8IoAAAggggAAC1UTAZlmG1Ap3f5y2i/PYOS7tUHGvznNZTVjpJgIIIIAAAggggMBZEDhl8NLXtg1i1g2LNy+5UW44ue84nwgggAACCCCAAALVR6CoON/8UTrXrSpO0LL6PDd6igACCCCAAAII1HSBCoOXNR2I748AAggggAACCCCAAAIIIIAAAggggAACVSPABEFV406rCCCAAAIIIIAAAggggAACCCCAAAIIIFCBAMHLCoA4jQACCCCAAAIIIIAAAggggAACCCCAAAJVI0DwsmrcaRUBBBBAAAEEEEAAAQQQQAABBBBAAAEEKhAgeFkBEKcRQAABBBBAAAEEEEAAAQQQQAABBBBAoGoE/h/5h+s6xzF93AAAAABJRU5ErkJggg=="
    }
   },
   "cell_type": "markdown",
   "id": "47d106fa",
   "metadata": {},
   "source": [
    "![templateselect.png](attachment:templateselect.png)\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "4. Wait a minute or so for it to load up\n",
    "5. Click **connect**\n",
    "6. Click on the button to connect to **Jupyter Lab [Port 888]**"
   ]
  },
  {
   "attachments": {
    "connectselect.png": {
     "image/png": "iVBORw0KGgoAAAANSUhEUgAAAT0AAAA8CAYAAAAUsG2MAAABamlDQ1BrQ0dDb2xvclNwYWNlSVRVUl8yMDIwAAAokWNgYNJJLCjIYWFgYMjNKykKcndSiIiMUmB/wMDIwMogwGDJoJ6YXFzgGBDgA1TCAKNRwbdrQNVAcFkXZFY4t3TZsQUBq+bnBPfu07X7gKkeBXClpBYnA+k/QJyfXFBUwsDAmANkK5eXFIDYE4BskSKgo4DsJSB2OoS9A8ROgrDPgNWEBDkD2Q+AbIXkjMQUBgYmkPk6SUji6UhsqL0gIBqUmqyn4BkSqhuk4BSiZ2RgZKBrSMDdJIOS1IoSEO2cX1BZlJmeUaLgCAytVAXPvGQ9HQUjA0MzBgZQ2ENUfw4Ehymj2BmE2JpMBgb3/P///zcixLSAYmseMzCwr0WIKTAzMHCvZWA4/a8gsSgRLAQUYWCyeMLA8AIYH+ICDAwgtsie4jRjI7A8I48TAwPrvf//P6sBzZrMwPB3wv//vxf9//93MVDzHQaGA3kAWuRoHHs2YjEAAABWZVhJZk1NACoAAAAIAAGHaQAEAAAAAQAAABoAAAAAAAOShgAHAAAAEgAAAESgAgAEAAAAAQAAAT2gAwAEAAAAAQAAADwAAAAAQVNDSUkAAABTY3JlZW5zaG90sswHoAAAAdVpVFh0WE1MOmNvbS5hZG9iZS54bXAAAAAAADx4OnhtcG1ldGEgeG1sbnM6eD0iYWRvYmU6bnM6bWV0YS8iIHg6eG1wdGs9IlhNUCBDb3JlIDYuMC4wIj4KICAgPHJkZjpSREYgeG1sbnM6cmRmPSJodHRwOi8vd3d3LnczLm9yZy8xOTk5LzAyLzIyLXJkZi1zeW50YXgtbnMjIj4KICAgICAgPHJkZjpEZXNjcmlwdGlvbiByZGY6YWJvdXQ9IiIKICAgICAgICAgICAgeG1sbnM6ZXhpZj0iaHR0cDovL25zLmFkb2JlLmNvbS9leGlmLzEuMC8iPgogICAgICAgICA8ZXhpZjpQaXhlbFlEaW1lbnNpb24+NjA8L2V4aWY6UGl4ZWxZRGltZW5zaW9uPgogICAgICAgICA8ZXhpZjpQaXhlbFhEaW1lbnNpb24+MzE3PC9leGlmOlBpeGVsWERpbWVuc2lvbj4KICAgICAgICAgPGV4aWY6VXNlckNvbW1lbnQ+U2NyZWVuc2hvdDwvZXhpZjpVc2VyQ29tbWVudD4KICAgICAgPC9yZGY6RGVzY3JpcHRpb24+CiAgIDwvcmRmOlJERj4KPC94OnhtcG1ldGE+Cutmaq8AABVlSURBVHgB7V0HfE3nG35JkEkRiVFErdYoglhVGlSNGh3/IkaIrShFa5Tqrg6lLYpS1IiiYo+QqopVxCwtShIj9l6Jnv/3vHFO703uvbk3De6N9/XLPed833u+8ZxznvOO717Z3Ny9NBIRBAQBQeARQSD7IzJPmaYgIAgIAoyAkJ7cCIKAIPBIISCk90hdbpmsICAICOnJPSAICAKPFAJCeo/U5ZbJCgKCgJCe3AOCgCDwSCHgnhmzzeNZjHxyBZCHe15yd/NQTWbLjGalDUFAEBAELCCgUfLdW3Qr+SJdu51Il2/GWdCxXpTtv6zTA9n5eT+pOr9MV2+doJtJF9RgbpKm/okIAoKAIHA/EMimjCp3N0/yzJGPfD2KKGMrD527ftBu8ssw6QX4VuROz1zbRzfunL8fc5M2BQFBQBBIFwGvnH7k71Oeja7Eq3vT1c8Q6YHw3N286OSlbWLVpQuxKAgCgsD9RgDWX+HHgpWneYPSIz6HExlwaWFWCuHd78so7QsCgoC9CCCkBk4CN4GjbInDpIcYHlxaidvZglXqBAFB4EEjAE46c20/5xls9e0Q6YFBkbSQGJ4tSKVOEBAEHhYCN+6cY46yZe05RHpYloIsrYggIAgIAs6KADgKXGVNHCI9rMPDshQRQUAQEAScFQFwFLjKmjhEelh4jHV4IoKAICAIOCsC4KiUL0lYHqFDpIdvWkgCwzKQUioICALOgUAKR1n/VpiDpOcck5JRCAKCgCCQUQSE9DKKnJwnCAgCLomAkJ5LXjYZtCAgCGQUgUz5lZWMdi7nCQKuhECBAgVowrfjjCGP/3oCbdz4m3H8KO988smHVPKJEgzB9u2/05jPvnRaOFyW9AIDi1PFCuWpyONF6OiRv2lTTAxdv37DaYF2loE99thj1LlzR/rjj4O0atUaZxlWhsYRGtqGQERfffW1w+e7u7tT37696fBfh2npshV2ne+RKxfVrFmDTp46RXHH4/icEiUCqVWrFsb5N2/epF27dqu/WLpz545Rfj93MJfn6tejkqWeoKtXr1Js7B7av/+AWZf58+ejOnVq0+PqeTl9OpG2bNlKJ0+eMtOpXq0qlStfjnLmyEEHD/2ZhtDt6adGjWC6cuWKWbvOduBypOfl5UXfTfqW6td/1gzLW7du0fsffEwzZ/5oVu7MBz/8MJVOnjhJw4aPtDpMe3SsnmyhIl++vDRwQH/6eXHkAyO9zJ6DPq3Qdm2pgnrxZYT0cubMyTiA+O0lPcqWkhFcErmMPvzoEx5GgwbPcTv6mPTtxYsXqVNYVyY/vczRLV7sn435mBZHLqHZs+dZPB3Pw5w5M6lqUBWjXtM0enf0+zRt2gwuK1WqJM2PmM0vCF0J5NipUzht/30HF40ePZK6dO6kV/N22fIV1Lt3P0J76fXz9tvD+ZyjRw5S9uzOHTVz7tGZXYKUgwFv9GPC2xSzmULbh1H5ClVoyFvD6Pz58zT63XcIN4qrSHD1avT00xVtDtceHZsNOEFlVphDejDOn7+A/vdaO+rQsTNNmPgd5c7tS19+8Wl6p9ms9/b2ZssysHigVb2XWrdkwsNLLKhqTXrp5dfo7NmzNGzoW+ThgR/0JcIzA4t49OgPqELFIBo4cDD5+PjQkCFvcv0Tyi0F4R08eIgaNmpCtWvXo61bt1HzZk1V/8GsY08/rOgCHy5Fet7eXuya/X3sGLVv34l+/XUjm9Jz50bQqHffJ5jf7dq1YdhxwUeOHE7rolbR3j07aMrkiVReme66wPo49vch6ta1C0WtXck6kyZ9w2806KRXDx3cLBMnjKfdsduVu7CRPv74A/L09EQVC96O7yoixhj+OLCbFi2MoDJlSvMNGB93hHx9falSpacJ+2FhHfXTeIub1JoOiB39xu7aRhs2RNFHH75HcFszIj16dOV+6tZ9xjj96/FjuczPLz/p9Zgb+tq3dyd9P3USVbiH5dy5s1i3cOFCxvlNmjTmsvfeG8VbS/METjo2aHP6tMlUokQgt4EXAea+YnkkbfotmvdhrTgqTz5ZlsaN+4JxQnvoI29ec5wCAvxp3rwf6eAfe7i/8PAwR7th/bj4eNq8eSv98suv6j4YQ1FR0VSqVCny9/fn+mefrUuzZk7n+2D5ssXUr18fwyLSMY5Q1tiunVt5vC1aNKdVK5fyuT17dmMMypYtk2ZshQoV5LIffpjFZId42tqo9ZRLueL+/gW4Djpwtad+P50uX75MPy1YRHiGihYryvWFC6VcO1iUh5RbG5+QQD/Onst1xYoVM9rAjq1+WNEFPlyK9BCTwMVcoC5acvJdM3ij1IXGm/bnnyOVF5KNH0wQWlLSHb4RGzUKoQh1cxd9/HE+Dya7m5s79enTk99q169fp2ZNm1Cvnt3tqsfDM2/uj9RMvQ23btvObkyoItxp6sGCYAxTJk+g8C5h7BVFR/9CVasF0cIF83gOiKn8889dFYe8zvGV06dP83n6R1JSEpen1kG/8yPmcL/b1A1+7uw56tAhVD1Q0yiHisXcL2kf2pbu3L7DL5qGjRqwSxUQEECLFy/hLk3jWq2V9QGJVHWW5glspk6ZyNjg+ixZsoyqK6sXLwW437pUrFiBLly4yG0gVuaIgGwiFy+gl1q34vjlhg2/UsOGDZT1NcasmSpVKlNl9eJZsnQZgRzeHfUOdeli7uaZnWDngZoiC+aKOODMGVOpRo3qtHz5SvWCykODBw2kESOGmrUGi/jw4SO0ffsO9lz27t3H9YghAscbN9JigPgdBJYaXvRPPfUkNWoYQqdOnaYTKnQCQXwR7nzX8M58DEItERhI27f9zscHDx2i27dvM1bAAPdYh/bt2K39/Z77a08/3JgrfODn4u39K1ekrd269rbpiF7HTl00FXzVWrV6xeY4Qho0Zr2Zs2Ybei+/0obLxo4dz2WoQ1t1nw3hYx/f/Jq64bQ1a6Lsqh867B0+v2XLf8cycuR7XFY9uI72XMi/Y8iR04fbbN+hM9djHpi3eqtqK1eu5n1rOKTWUfE/bqNHjz7GeZ9/MZbLXnst1Ciz1l7pMuVZ99sJk1i3d59+fFz/ueeNcydM+I7LAgoW1Xr1TqmfN2++ps+jTZv2XK9iqFruPH5aXFy8FrVuPZ+vH69evdZoL/UcdGw+/fRzQ6fR8025zX79B2pVgmrwfkTET0a9pfmgD/VgW9QpVryUhuvRuvWrRv38nxZqCQkJmodnHs03tx/3cfTo35py+VjHr0BhTVlK2p9//sU6qfsMLFGWz0G7et3zjZtx2cA3hxhlykXU4uPjNZXZ5TIVM+RxVq4SzMe5PHJr0dEb+LyixUpq+jV4441BRhto/+lK1dL0p/druv36mwmsh/sZf7getWrXM9rCva1I30xn585YDfPV22nbtoNZPdrp22+AUQ+99PqBzvHjcZqKq5udp/fxILe2uMqlLD1lnNklVasGsZ4eyMUBLJJr164R3uymsmPHLj6EJXHixAnKa2JpoMJavd5Hp07tOUiMQHFISH11BrHrFxSU0s+kSVOURfcPl8MNVze51aA0K6XzgYD13bvJNOvHOYbmzJmzeb9y5UpGWWbt6BYL+tPnAfcIiaNKyg1Fxnz1mrVU7qmnqHjxYtS0yQscZliwcJHVIejYwPIFbvjr+3pv1tfdZhzE7k6xYqw2ZKMCVs648d9QwYIFacaM72nBT3OplopPZc/uZhYKgCWDTDbk4sVLtFIlNhDvKlnyCRutp60a9OYAdkHhRs+cMU1lUa/R4MFD2YUNUvfcjh07SbfckpOTae68CG7E9H7MyHxDQuoTLGtYg9N/mMHJqezZs9GYMR8ZoZr+/V+n0qVL0759+2nKlO8Jcy5YMEDF+EbyGBCagNWJZwDPCVxbJGIQC0SIABISkn4/0HMFcansrb5MoLSKi6UWxIgQ+zii3AOY8pAzZ86YqZ1SN4aXV0rMDW4H5O7df91klLm7uXF5evUeHrmYfOAm64JkyubN5+mSipsULlKYizM7fZ9L9Xv5ylV2R/R+ExMTeRcxxPsl19RDbCp4gXh6pgTKFy1czKGBV15+Sb1UKrHbvmDBz6bqZvtwwyDJSclGeZJy5zdv3kLH4+I5NIAK/RoYSg7sIFY6U7n8+fLm5XjWzp272FVGrNRULl+5YnpoLLcwjc2aKVg5iFFjR/AfIYBdsbGKWHbyNcK9iHkkproXzySm3Jve6prp89S3VrqwWAxCu33rtnLdm/ByFSjBNf3oo/epefOmtEwtx+nRvSsTXeuX/me0gfh161Yt6bPPxlJoaBvl0uZV8fCOtPG3TazzvYr/Ra9fwzHdAQMGU3r9IJHjKuJSpLdZxTVAZIiTIYWvP+wAGxYXlmJ88OHHxvojrDtCYBbi5+fHb+9t9+IYXPgfPrDUBLGg1/sOMMgVSQpYFjExmym3SlJA6tSuZVgSgSoBgbfyhg0bCQ9hRkS5HRRUpQrP5ciRo9wE4mEQPYbDB/c+YIEEB1ejbt17cxA7d+7cXHPlcsrDDrKBYA2aLv4quA8xJfRgFY/6bVMMl2MewBOxIsh6Fa9EgDwsrANnLdetizbIgxVSfQA7yKRJk43lIiDsxo0bqQTAek4QpTrF4cOmTV9gwuveozepEAKfP1nFWJu80NisLWCJBBisL0hwcHXeAmdHBNd83Lhv0pyCBMKFCxfYwzDtp9q9a5agvIt8ag1dRsVfkfilS+p/I1RLUHQ5dm8NIeJz+fPl4/kdj4vTq3l79OjfvIUO2oCY6sTHJ/D1L3QvyZFeP9yAi3y4FOnBKhs8ZChNnz6FVq6IVMH0CNq7bx9Vq1qVunfvwhmpuXPncwZ1+LC36K23BtF5dcPB2npduU9wbRYuWpwpl2ahsm6aN2+m3IUJfLPDNejfvy+TnIrz8Dq4Pn16KhdnIJPi2XPnaOjbQ/jmB+lB4CLCJWzWrAntVq5cQsKJNGNLrQOrCksJxn31BX065gvKpSyJUaNGsNWJhEBqSUpOolq1atI7yn3Buqs+vXuyyl7l6kAOH04hzuEj3mbLDRZSbaUPAenphmzvXj04yH5RJRYGDR7A9YsjU7KLIAz0jYQKJDXGqeewavVaGj78bc7euqvkC5I4jVSSoVu3cBo5arR6IaSQKTdmx0etWjXMtEAC+gsR87l06RInE0CqqQUP/ScqM71w0c+cTHlGJctgsennp9bPyDESb92VtYU1d/N/WkBFixalMPWSVvFF1dd2q8uWgBvkmWdqU8MGIbRBrVbQX1L6OLaoscLCxjdFZs+Zx0Q/TN37kJhNmwmkin5atXyRjh07znMrV+4pCg8P43DPnj17CS+xV199mSZN/Ia+/HIc4Z7ppa43LE+0AUmvH1ZylQ9Hgou2goOOtPNfddu266ipN5VZ4HXf/gNGQBrtN27cnIPJCMjqf+PHf2sEWFWMissRVNbHo5bAaMp64eP06nGOnszQ28fWNDiPhAYC2nq9ugG1zz4fa/Q3alRK4gP1vXr1Ncr18WBrSWf4iFFGmzgXwXwEok3P0/cRxFYvBzP9deuiOQEBHSQnFGEZ9Rijsuj42D/gcSPIjsC3Pg9s9YSQ3k/NWs9yPZJBppii3tIckABAwN20zfXrf+EEgp7IAL56+5a2q1Uiw/R8fR9Jl3z5C2nKVbNYX7BQcSORoZaZmI0jNna3VjywtMV+7U1kpB6rl3deLTJyqdlYkNxBogK6eiLDNJmEcvcc3tqKFauM8ypXrp5mXIGBZTRF0oYOMMA1VFanodukaQtOmun4YKu8BE2tc2UdJHaQNDKtxz76RmIKY7GnH+i5QiLDof8CsmxASzqUGOkUfO6jgs3PqW9llFTrt04oC2nFylVpvoZWWq2TerFFMx6v/nUj7Z7pggAtUvNbtqj/xvJeGawCd3c3Tl6kV6+DUK9eXUJSA23AQoiJ2aJX8RZLCLBuDUtLVq+JMrMgsHIddVgrph54fhObnawOrOnAatHXHcIlxBvbliBgXk2NE8trLOmirkLFCrRuXTS7un4F8vPSifDwMBoxfKj6dkE4W6WI/cBiwNIgHTf0C9ce6xERBB86dASKDLE2ByQLWrRozhaFIkt2Q2HJ+Ph4U0U1loSEk6ReGkY7qXegA93UAksPyYk8efJQmzavEhb5buHQyFm1RtKP421IysDtP3f2PLuXderU4hjjHJVsOqesckuC5U4xMRuMqjcGDFLXLZqD/XAHLVnqujJc23ZtXyM/1T9imREqBqZbk0gkwOI/cED9h9UqTGAqWJv6srLksLQJy56whMeSwHqDxavIijYp6yx1LNlNxaqrqXBP5cpP0/59BwjLneB6m4qfX35lVdYhXx8f+kV5I5awt9bPyhVL+NsxaG/t2ijqEt7DtOkHvm+Lq1yW9B44io9ohz16dGXSaxfaKc13MQEJAvVYc1a7dk12015o8mKa731mFegQD+3atbMxHcQK9cyvUfiI7iAZgnWbELzALIVaHiQ0tkjPpWJ6DxI06cs+BILUEhpkzSFfjh2XZQkP84P1hJiXSFoEkFh0FRHSc5Ur9ZDGuXTpcnaH4XpZEriNNWvVJS9PL/rr8GFLKlImCDgVAkJ6TnU5nG8wiBHhz5ZYWipjS1/qBIGHiYBLfSPjYQIlfQsCgkDWQEBIL2tcR5mFICAI2ImAg6SnUTb1T0QQEAQEAWdFIIWj/v16aOpxOkR6yXdvqe+m/vt7cakbk2NBQBAQBB42AuAocJU1cYj0biVfJM8cGf+eoLVBSLkgIAgIApmFADgKXGVNHCK9a7cTydejiLW2pFwQEAQEgYeOADgKXGVNHCK9yzfjyMM9D3nlzG+tPSkXBAQBQeChIeCV0485ClxlTRwiPTRy7vpB8vepIAkNa4hKuSAgCDwUBJDA8PcpzxxlawAOkx4Y9GbSBSr8WLAQny1kpU4QEAQeGAIgPHASuMmWlYcBOfSDA6YzCPCtyEmNM9f20Y07502rZF8QEAQEgQeGAFxaWHggvMSrtn9tCIPKMOnh5DyexcjP+0mVKVG/3HrrBHeafPcmqZ+eRLWIICAICAKZjgCsOixLQZYWSQvkGRB2S8/C0wfyn0hPbwTk55MrQHWeVw0G//+BLGDWsZGtICAIZDYCGq/Dw7IUZGntJTt9FJlCenpjshUEBAFBwNkRcDiR4ewTkvEJAoKAIGALASE9W+hInSAgCGQ5BIT0stwllQkJAoKALQSE9GyhI3WCgCCQ5RAQ0styl1QmJAgIArYQENKzhY7UCQKCQJZDQEgvy11SmZAgIAjYQkBIzxY6UicICAJZDgEhvSx3SWVCgoAgYAuB/wOrg4Iq1yhCsAAAAABJRU5ErkJggg=="
    }
   },
   "cell_type": "markdown",
   "id": "359e5d1c",
   "metadata": {},
   "source": [
    "![connectselect.png](attachment:connectselect.png)\n",
    "\n",
    "\n",
    "7. Create a **new notebook**, and you should be ready to go!"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "99d49c86",
   "metadata": {},
   "source": [
    "# Preparing your data"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2bf81102",
   "metadata": {},
   "source": [
    "This is probably the most important and frusterating part! Of course you want to make sure your input/output data is quality, and that you have enough of it (~10K rows, for us). \n",
    "\n",
    "Once you've got that, you'll want to format it. You're gonna want to format it ***exactly like [this](https://huggingface.co/datasets/timdettmers/openassistant-guanaco/blob/main/openassistant_best_replies_train.jsonl)***. I would recommend downloading the jsonl file and taking a look.\n",
    "\n",
    "You need a `.jsonl` file structured like this:\n",
    "\n",
    "```\n",
    "{\"text\": \"### Human: YOURINSTRUCTIONHERE: YOURINPUT1HERE ### Assistant: YOUROUTPUT1HERE\"}\n",
    "{\"text\": \"### Human: YOURINSTRUCTIONHERE: YOURINPUT2HERE ### Assistant: YOUROUTPUT2HERE\"}\n",
    "{\"text\": \"### Human: YOURINSTRUCTIONHERE: YOURINPUT3HERE ### Assistant: YOUROUTPUT3HERE\"}\n",
    "```\n",
    "\n",
    "Here's an explanation of the above:\n",
    "1. **YOURINPUTXHERE:** Your inputs. If you were doing keyword extraction, this would be the text you're extracting from\n",
    "2. **YOUROUTPUTXHERE:** Your outputs, properly formated. If you wanted your keywords as a list, these would be the training outputs formatted like this: [\"yes\",\"no\",\"cool\"]\n",
    "3. **YOURINSTRUCTIONHERE:** This is the instruction. We think of it as a short (one or two sentence) reminder to the model of what to do. It'll really help training go faster. It should be the same for every one if you're training on a specific task. For keyword extraction, it'd be something like: ```\"Extract relevant keywords from the following text.\"```\n",
    "\n",
    "\n",
    "This isn't the only way to do it, but it's the way we finally got things to work. Make sure your file **doesn't have any extra lines or characters**, and that your data is **sanitized properly**. The formatting can be really annoying. We used a python script to generate the file. \n",
    "\n",
    "\n",
    "Once you have it, **name it `data.jsonl`** drag and drop it **into your jupyter lab directory**."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d3d66857",
   "metadata": {},
   "source": [
    "# Access to Llama2 (and the license)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ce2be4bc",
   "metadata": {},
   "source": [
    "The great thing about Llama 2 is that it has a commercial license. But you have to go to meta and accept that license. \n",
    "1. Make a huggingface account (if you don't have one already)\n",
    "2. ***Request access [here](https://ai.meta.com/resources/models-and-libraries/llama-downloads/)*** and make sure you use the same email as your hugging face account (**very important**)\n",
    "3. You should get an email from meta within 5 minutes. \n",
    "4. Now, you'll need to ***request access [here](https://huggingface.co/meta-llama/Llama-2-7b-hf)*** for the model on huggingface. \n",
    "5. Once they give it to you (it usually takes about an hour), go to your user settings, **Access Tokens**, and create (and copy) a new access token."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ffd23630",
   "metadata": {},
   "source": [
    "# Training\n",
    "\n",
    "You can copy and paste cell by cell into your new runpod notebook, or you can upload this notebook and use it directly."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "4cd63725-1dd9-46dd-840a-10a0a90b6c5d",
   "metadata": {},
   "outputs": [],
   "source": [
    "%%capture #Remove this if you're having problems, it just makes these not print output\n",
    "!pip uninstall transformers --yes #You'll need a special, different transformers\n",
    "!pip install -q -U datasets peft trl bitsandbytes transformers[torch] scipy tqdm "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "6aa3b836",
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "from dataclasses import dataclass, field\n",
    "from typing import Optional\n",
    "\n",
    "import torch\n",
    "from datasets import load_dataset\n",
    "from datasets import load_from_disk\n",
    "from peft import LoraConfig\n",
    "from transformers import (\n",
    "    AutoModelForCausalLM,\n",
    "    AutoTokenizer,\n",
    "    BitsAndBytesConfig,\n",
    "    HfArgumentParser,\n",
    "    AutoTokenizer,\n",
    "    TrainingArguments,\n",
    ")\n",
    "from tqdm.notebook import tqdm\n",
    "\n",
    "from trl import SFTTrainer\n",
    "from huggingface_hub import interpreter_login"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "da69f970",
   "metadata": {},
   "source": [
    "### Add your own token after running the line below"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "41cbc296",
   "metadata": {},
   "outputs": [],
   "source": [
    "interpreter_login()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "e12c183a",
   "metadata": {},
   "outputs": [],
   "source": [
    "compute_dtype = getattr(torch, \"float16\")\n",
    "bnb_config = BitsAndBytesConfig(\n",
    "        load_in_4bit=True,\n",
    "        bnb_4bit_quant_type='nf4',\n",
    "        bnb_4bit_compute_dtype='float16',\n",
    "        bnb_4bit_use_double_quant=False,\n",
    "    )\n",
    "device_map = {\"\": 0}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "9c71894d",
   "metadata": {},
   "outputs": [],
   "source": [
    "#Download model\n",
    "model = AutoModelForCausalLM.from_pretrained(\n",
    "        \"meta-llama/Llama-2-7b-hf\", \n",
    "        quantization_config=bnb_config, \n",
    "        device_map=device_map, \n",
    "        use_auth_token=True\n",
    "    )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "55e7a45f",
   "metadata": {},
   "outputs": [],
   "source": [
    "model.config.pretraining_tp = 1 \n",
    "peft_config = LoraConfig(\n",
    "    lora_alpha=16,\n",
    "    lora_dropout=.1,\n",
    "    r=32,\n",
    "    bias=\"none\",\n",
    "    task_type=\"CAUSAL_LM\", \n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "45344014",
   "metadata": {},
   "outputs": [],
   "source": [
    "tokenizer = AutoTokenizer.from_pretrained(\"meta-llama/Llama-2-7b-hf\", trust_remote_code=True)\n",
    "tokenizer.pad_token = tokenizer.eos_token"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "43ab95d0",
   "metadata": {},
   "outputs": [],
   "source": [
    "training_arguments = TrainingArguments(\n",
    "    output_dir=\"./results\",\n",
    "    per_device_train_batch_size=1,\n",
    "    gradient_accumulation_steps=4,\n",
    "    optim=\"paged_adamw_32bit\",\n",
    "    save_steps=500, #CHANGE THIS IF YOU WANT IT TO SAVE LESS OFTEN. I WOULDN'T SAVE MORE OFTEN BECAUSE OF SPACE\n",
    "    logging_steps=10,\n",
    "    learning_rate=2e-4,\n",
    "    fp16=False,\n",
    "    bf16=True,\n",
    "    max_grad_norm=.3,\n",
    "    max_steps=10000,\n",
    "    warmup_ratio=.03,\n",
    "    group_by_length=True,\n",
    "    lr_scheduler_type=\"constant\",\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "8363e637",
   "metadata": {},
   "outputs": [],
   "source": [
    "model.config.use_cache = False"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f60f7d97",
   "metadata": {},
   "source": [
    "### Load in your data here...."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "c27b3f49",
   "metadata": {},
   "outputs": [],
   "source": [
    "dataset = load_dataset(\"json\", data_files=\"data.jsonl\", split=\"train\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "e2093bc7",
   "metadata": {},
   "outputs": [],
   "source": [
    "trainer = SFTTrainer(\n",
    "    model=model,\n",
    "    train_dataset=dataset,\n",
    "    peft_config=peft_config,\n",
    "    dataset_text_field=\"text\",\n",
    "    max_seq_length=1024,\n",
    "    tokenizer=tokenizer,\n",
    "    args=training_arguments,\n",
    "    packing=False,\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0ade167c",
   "metadata": {},
   "source": [
    "### The main event... training\n",
    "\n",
    "This took between **2 and 8 hours** for us depending on amount of data and other factors. It is currently configured to go to **10K steps**, saving a **checkpoint after every 500 steps**. These checkpoints can be **very large**, on the order of **1-2GB**. It is noted above where you could change this as needed.\n",
    "\n",
    "Even if you stop early, you can **restart training** like this: ```trainer.train(resume_from_checkpoint=True)```\n",
    "\n",
    "Futher, you're saving a **checkpoint every 500 steps**, so you can run the model and test performance as you go along!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "2729b7c5",
   "metadata": {},
   "outputs": [],
   "source": [
    "trainer.train()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b2573884",
   "metadata": {},
   "source": [
    "# Saving your Model to HF\n",
    "\n",
    "At any point, you can save your model to HF using the below code. You can manage privacy and stuff on HF"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "d472ca8a",
   "metadata": {},
   "outputs": [],
   "source": [
    "model.push_to_hub(\"YOURUSERNAME/WHATEVER\", use_auth_token=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bbd2eb12",
   "metadata": {},
   "source": [
    "# Running Inference as you go\n",
    "\n",
    "You can stop anytime, and run the model from any checkpoint (multiple of 500) using the code here!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "ce66ae22",
   "metadata": {},
   "outputs": [],
   "source": [
    "from peft import PeftModel\n",
    "model = PeftModel.from_pretrained(model, \"results/checkpoint-2000\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "8ec6c19e",
   "metadata": {},
   "outputs": [],
   "source": [
    "def make_prompt(entry):\n",
    "    return f\"### Human: YOUR INSTRUCTION HERE: {entry} ### Assistant:\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "b95f68e1",
   "metadata": {},
   "outputs": [],
   "source": [
    "def run_model(entry):\n",
    "    model_input = tokenizer(\n",
    "        make_prompt(entry),\n",
    "        return_tensors=\"pt\").to(\"cuda\")\n",
    "\n",
    "    input_length = len(model_input['input_ids'][0])\n",
    "    model.eval()\n",
    "    with torch.no_grad():\n",
    "        full_tokens = model.generate(**model_input, max_new_tokens=100)[0]\n",
    "        decoded_tokens = tokenizer.decode(full_tokens[input_length:], skip_special_tokens=True)\n",
    "    return decoded_tokens"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "068ee94f",
   "metadata": {},
   "outputs": [],
   "source": [
    "run_model(\"Hello! Here's my example of an input you might test...\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "91ae0e9e",
   "metadata": {},
   "source": [
    "# Running Inference Later (maybe in Google Colab)\n",
    "\n",
    "We found that you have to adjust settings slightly to get things to run in google colab. Likewise, every environment will differ. What follows is the code we used, in google colab. If you're doing it in colab, **don't forget to change the instance type to T4**. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "9ed4aa13",
   "metadata": {},
   "outputs": [],
   "source": [
    "%%capture\n",
    "!pip install transformers[torch]\n",
    "!pip install peft\n",
    "!pip install datasets\n",
    "!pip install bitsandbytes\n",
    "!pip install tqdm\n",
    "import torch\n",
    "from transformers import AutoModelForCausalLM, BitsAndBytesConfig, AutoTokenizer\n",
    "from peft import PeftModel\n",
    "from tqdm.notebook import tqdm"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "3a5756fd",
   "metadata": {},
   "outputs": [],
   "source": [
    "from huggingface_hub import notebook_login\n",
    "notebook_login()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "bb4e2dad",
   "metadata": {},
   "outputs": [],
   "source": [
    "bnb_config = BitsAndBytesConfig(\n",
    "    load_in_8bit=True,\n",
    "    bnb_4bit_compute_dtype=torch.float16,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "f10820ce",
   "metadata": {},
   "outputs": [],
   "source": [
    "base_model_name=\"meta-llama/Llama-2-7b-hf\"\n",
    "\n",
    "base_model = AutoModelForCausalLM.from_pretrained(\n",
    "    base_model_name,\n",
    "    device_map=\"auto\",\n",
    "    quantization_config=bnb_config,\n",
    "    trust_remote_code=True,\n",
    "    use_auth_token=True\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "fb65f8af",
   "metadata": {},
   "outputs": [],
   "source": [
    "tokenizer = AutoTokenizer.from_pretrained(base_model_name, trust_remote_code=True)\n",
    "tokenizer.pad_token = tokenizer.eos_token"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "171159c2",
   "metadata": {},
   "source": [
    "### Input your own model here, as you saved it in the above step"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "c114ea53",
   "metadata": {},
   "outputs": [],
   "source": [
    "model = PeftModel.from_pretrained(base_model,\n",
    "                                  \"YOURUSERNAME/WHATEVER\",\n",
    "                                  use_auth_token=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "077e64b0",
   "metadata": {},
   "outputs": [],
   "source": [
    "def make_prompt(entry):\n",
    "    return f\"### Human: YOUR INSTRUCTION HERE: {entry} ### Assistant:\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "340790ef",
   "metadata": {},
   "outputs": [],
   "source": [
    "def run_model(entry):\n",
    "    model_input = tokenizer(\n",
    "        make_prompt(entry),\n",
    "        return_tensors=\"pt\").to(\"cuda\")\n",
    "\n",
    "    input_length = len(model_input['input_ids'][0])\n",
    "    model.eval()\n",
    "    with torch.no_grad():\n",
    "        full_tokens = model.generate(**model_input, max_new_tokens=100)[0]\n",
    "        decoded_tokens = tokenizer.decode(full_tokens[input_length:], skip_special_tokens=True)\n",
    "    return decoded_tokens"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "14d82b14",
   "metadata": {},
   "outputs": [],
   "source": [
    "run_model(\"Hello! Here's my example of an input you might test...\")"
   ]
  }
 ],
 "metadata": {
  "hide_input": false,
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.9.17"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}