cnn_rnn_inference.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "from torch.autograd import Variable\n",
    "from cnn_rnn_config import classes_, cdict_, icdict_, fixed_size\n",
    "import editdistance\n",
    "import argparse\n",
    "from skimage import io as img_io\n",
    "import numpy as np\n",
    "from data.iamdataset import IAMDataset, image_resize, centered\n",
    "import cv2"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### Helper functions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# only works for 2 heads\n",
    "# TODO: make it work only for RNN head\n",
    "def inference(model, inputs, gpu_id=0, two_heads=True):\n",
    "    tdecs_rnn = []\n",
    "    outputs = []\n",
    "    for img in inputs:\n",
    "        # add batch dimension\n",
    "        img = img.unsqueeze(0)\n",
    "        img = Variable(img.cuda(gpu_id))\n",
    "        with torch.no_grad():\n",
    "            if two_heads:\n",
    "                output, aux_output = model(img)\n",
    "            else:\n",
    "                output = model(img)\n",
    "        tdec = output.argmax(2).permute(1, 0).cpu().numpy().squeeze()\n",
    "        tdecs_rnn += [tdec]\n",
    "\n",
    "    for tdec in tdecs_rnn:\n",
    "        tt = [v for j, v in enumerate(tdec) if j == 0 or v != tdec[j - 1]]\n",
    "        dec_transcr = ''.join([icdict_[t] for t in tt]).replace('_', '')\n",
    "        dec_transcr = dec_transcr.strip()\n",
    "        outputs += [dec_transcr]\n",
    "    \n",
    "    return outputs"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### Calculate Character Error Rate (CER) and Word Error Rate (WER) \n",
    "\n",
    "<img src=\"info/cer.jpg\" alt=\"Character Error Rate\" height=\"100\"/>\n",
    "<img src=\"info/wer.jpg\" alt=\"Word Error Rate\" height=\"100\"/><br/>\n",
    "<img src=\"info/explain.jpg\" alt=\"Notation\" width=\"400\"/>\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "def cer_wer(outputs, targets):\n",
    "    cer, wer = [], []\n",
    "    cntc, cntw = 0, 0\n",
    "    for o, t in zip(outputs, targets):\n",
    "        # calculate CER and WER\n",
    "        cc = float(editdistance.eval(o, t))\n",
    "        ww = float(editdistance.eval(o.split(' '), t.split(' ')))\n",
    "        cntc += len(t)\n",
    "        cntw +=  len(t.split(' '))\n",
    "        cer += [cc]\n",
    "        wer += [ww]\n",
    "\n",
    "    cer = sum(cer) / cntc\n",
    "    wer = sum(wer) / cntw\n",
    "    return cer, wer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "def load_model(model_path='./saved_models/cnn_rnn/best_rnn_head.pth', gpu_id=0):\n",
    "    model = torch.load(model_path)\n",
    "    model = model.cuda(gpu_id)\n",
    "    model.eval()\n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "def preprocess_images(img_paths):\n",
    "    imgs = []\n",
    "    for img_path in img_paths:\n",
    "        img = img_io.imread(img_path)\n",
    "        img = 1 - img.astype(np.float32) / 255.0\n",
    "        img = image_resize(img, height=img.shape[0] // 2)\n",
    "        fheight, fwidth = fixed_size[0], fixed_size[1]\n",
    "        nheight, nwidth = img.shape[0], img.shape[1]\n",
    "\n",
    "        nheight, nwidth = max(4, min(fheight-16, nheight)), max(8, min(fwidth-32, nwidth))\n",
    "\n",
    "        img = image_resize(img, height=int(1.0 * nheight), width=int(1.0 * nwidth))\n",
    "        img = centered(img, (fheight, fwidth), border_value=0.0)\n",
    "        img = torch.Tensor(img).float().unsqueeze(0)\n",
    "        imgs += [img]\n",
    "    return imgs\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "def testing(dataset, n_images, model, gpu_id=0, two_heads=True):\n",
    "    tuples = []\n",
    "    for i in range(n_images):\n",
    "        img, target = dataset[i]\n",
    "        tuples += [(img, target)]\n",
    "    \n",
    "    for img, label in tuples:\n",
    "        label = label.strip()\n",
    "        img = img.unsqueeze(0)\n",
    "        outputs = inference(model, img, gpu_id=gpu_id, two_heads=two_heads)\n",
    "        cer, wer = cer_wer(outputs, [label])\n",
    "        print('YH: ', outputs[0])\n",
    "        print('GT: ', label)\n",
    "        print(f'CER: {cer}, WER: {wer}')\n",
    "        img = img.squeeze().numpy().astype(np.float32)\n",
    "        img = 1 - img\n",
    "        img = img_io.imshow(img)\n",
    "        img_io.show()\n",
    "        "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "def testing_custom_images(image_paths, model, gpu_id=0, two_heads=True):\n",
    "    imgs = preprocess_images(image_paths)\n",
    "    outputs = inference(model, imgs, gpu_id, two_heads=two_heads)\n",
    "    for i, img in enumerate(imgs):\n",
    "        print('Predicted: ', outputs[i])\n",
    "        img = img.squeeze().numpy().astype(np.float32)\n",
    "        img = 1 - img\n",
    "        img = img_io.imshow(img)\n",
    "        img_io.axis('off')\n",
    "        img_io.show()\n",
    "    "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Inference"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "model_path = './saved_models/cnn_rnn/best_rnn_head.pth'\n",
    "gpu_id = 0\n",
    "model = load_model(model_path, gpu_id)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 6482/6482 [00:00<00:00, 912668.39it/s]\n",
      "100%|██████████| 2915/2915 [00:00<00:00, 1061134.89it/s]\n"
     ]
    }
   ],
   "source": [
    "train_set = IAMDataset(subset='train', fixed_size=fixed_size, transforms=None)\n",
    "test_set = IAMDataset(subset='test', fixed_size=fixed_size, transforms=None)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### Performance on training data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "YH:  A MOVE to stop Mr. Gaitskell from\n",
      "GT:  A MOVE to stop Mr. Gaitskell from\n",
      "CER: 0.0, WER: 0.0\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAnQAAAB3CAYAAACKYt1IAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjguMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8g+/7EAAAACXBIWXMAAA9hAAAPYQGoP6dpAABD/ElEQVR4nO3dd3hUVf4/8Pf0SU8mvScQSCC00EEQkdARsaAou4vAIijsF0GlKbrrruK6P3cVC6KuwipFsaAgxQBKJwRIgBBCKklIMpM6LdNnzu8PnnucoShoCiOf1/PkUWbuzNy55865n3PO55wrYowxEEIIIYQQryXu6B0ghBBCCCG/DQV0hBBCCCFejgI6QgghhBAvRwEdIYQQQoiXo4COEEIIIcTLUUBHCCGEEOLlKKAjhBBCCPFyFNARQgghhHg5CugIIYQQQrwcBXSEEEIIIV6uwwK6d955B0lJSVAqlRg0aBCOHz/eUbtCCCGEEOLVOiSg++yzz7B48WK8+OKLOHXqFHr37o2xY8eirq6uI3aHEEIIIcSriRhjrL0/dNCgQRgwYADefvttAIDL5UJ8fDz+8pe/YNmyZb/4epfLhZqaGgQEBEAkErX17hJCCCGEtDnGGAwGA2JiYiAW31yfm7SN9um6bDYbTp48ieXLl/PHxGIxMjMzcfTo0Wu+xmq1wmq18n9XV1eje/fubb6vhBBCCCHtraqqCnFxcTf1mnYP6BoaGuB0OhEZGenxeGRkJAoLC6/5mlWrVuFvf/vbVY9XVVUhMDCwTfaTEEIIIaQ96fV6xMfHIyAg4KZf2+4B3a+xfPlyLF68mP9b+MKBgYEU0BFCCCHkd+XXpJO1e0AXFhYGiUQCjUbj8bhGo0FUVNQ1X6NQKKBQKNpj9wghhBBCvE67z3KVy+Xo168f9u7dyx9zuVzYu3cvhgwZ0t67QwghhBDi9TpkyHXx4sWYMWMG+vfvj4EDB+KNN95AS0sLZs6c2RG7QwghhBDi1TokoHv44YdRX1+PF154AWq1Gn369MGuXbuumihBCCGEEEJ+WYesQ/db6fV6BAUFQafT0aQIQgghhPwu/Jb4hu7lSgghhBDi5SigI4QQQgjxchTQEUIIIYR4OQroCCGEEEK8HAV0hBBCCCFejgI6QgghhBAvRwEdIYQQQoiXo4COEEIIIcTLUUBHCCGEEOLlKKAjhBBCCPFyFNARQgghhHg5CugIIYQQQrwcBXSEEEIIIV6OAjpCCCFey+l0Yu7cuYiLi8P69ethNps7epcI6RDSjt4BQgghtzaXy3VD2xgMBpw/fx5arRalpaVwOBwQi8Xo1KkTEhISkJaWBplM1qr7ZrVaUVxcDMYY1Go1dDodfHx8WvUzrudGjktbfGZNTQ127dqF8ePHIzw8HHK5vNXeXyQSQSQStdr7kfZDAR0h5HfLZrOhqKgIkZGRCAkJgUQi6dD98dYLJWMMjLFrPud0OlFaWoqdO3fi0KFDUKvVkEgk/M/lckEsFsPPzw/33nsvHn300VYNQCQSCfr06YOWlhYMHjz4usGc8B3E4tYbmPq549IWGGMoLS3FSy+9hOLiYjQ0NOBPf/oTIiMjb+jcYoz94nZisdhrz9PbHQV0hHghxhi0Wi1++OEHZGZmwt/fv9U/4/fQUt++fTvWrl2LXr164YknnkBCQsINv9Zms0EsFkMqbZ1qUiQSdXhA2ZrsdjtOnz6NU6dOYcOGDRCLxVCpVPjTn/6EiIgIxMbGwtfXF3l5efj++++h1+uxdetWREVFYcyYMa0WWInFYqxcuRJOpxNBQUEe78sYg8vlAmMMGzduhMlkwp///OebPq9FIlGrBoK/hsPhQH19PZ5//nlUVFRAIpHg+PHjuOuuuxAWFvaz56ndbkddXR20Wi1CQ0OvGwC2V3DqHgg7nU7U19fjwIEDiIiIQJ8+fRAUFASRSMS3uV55/R7qqNZEAR0hXshms2Hfvn3YvHkzYmNj0a9fv5+t2H6pYrzea7w9ALl06RIsFguqqqpQUFCAuLi4G7owu1wunD9/Hv7+/ujcuXM77Kn3qaurw7p161BUVASHw4GxY8di5MiR6NGjB2QyGcRiMcRiMWJiYuDj44Pjx4/j3LlzOHz4MDIzM1stQBKJRAgICLhu7xNjDA6HA8ePH4dUKuU9hjf7GTfqRnrBfg2TyYTS0lI0NjZCIpFAJBKhpaUFlZWV6Nev33Vf53K5UFtbi82bN+P8+fMYOHAgZs6cCYVCcdV+tldw5B5o6/V6fP7559ixYwf8/f0xc+ZMjB49+oYaUhTMeaJJEeS2Y7fb23WYpC20tLRg165dUKvV2LVrF68cr/cn+Lltrvcab3bHHXfA5XJBrVbjo48+gk6nu6HX7d27Fy+//DK+/fbb382xaC02mw27du3CE088Aa1Wi3HjxuHjjz/GkiVLMGTIEAQEBMDHxwe+vr5QKpUIDg7GpEmTsHDhQkRGRkKpVP6mY3q9117v4s4Yg8ViQUNDA2w2G5xO56/+7F/ar+PHj+Nvf/sbmpqaWv29jx07hv/9738wm814+OGHERAQAJvNhm+++QYmk+m6x2X37t148skn8d1336GkpATffPMNjh07dkuc1xqNBitWrMCWLVvgcDig1+vxwQcfYOPGjbDZbHwff891VGu6bXvonE4nGGOtNpxys9oin6O9McZgt9t/Nh/G5XLB6XSipqYGFRUVsFgsCAgIgL+/P/z8/BASEgI/P79Wzam53r7q9Xrk5eVh06ZNmDRpEsaNG+dxEfCm3iiRSASpVAqHw4GKigrYbDYolcqO3q1bTqdOnRAfH4/i4mK0tLTg1KlTGDly5M/+7hhjKC4uhlarhUajabMeF2/DGIPVasW2bduwdu1adO/eHQsXLkR0dDTkcvnP1qUymQxhYWF49NFHkZ6e/psmRtxsWTidThgMBpjNZgQGBrZZnWsymfDee++hrq4ORqMRKpWq1d7barXiwIEDuHDhAiIiIjBx4kS4XC5s3boVGo0GFRUVSE1N9TiujDEYDAZ8/fXXEIvFmD59OnJzc5Gfn4+cnBwMHTq0zevd62GMwWw2480330RBQQEyMzMRGRmJ/Px8XLx4ETt27EDXrl0xcOBA+u3dhNsyoDMajVi9ejVqamrwj3/8A35+fvykEcbk2zLQstls+PDDD6HRaPDEE09ApVLxRNQrLx7urRD3nAL3f7u/1uFwwGq1QqfT8VabRCKBQqGAQqGAr68vfHx8PL7flT8Yp9MJq9UKqVR61XZCIGqxWPDf//4Xn332GTZu3Ijk5GSP9xC+R0NDA/Ly8vDiiy9Cr9cDAKRSKQ9IQkNDcc8992DBggWtc3CvgzGGpqYmfPPNNygpKcHu3bsxevRorwri3MlkMoSEhIAxhvLychQXFyM9Pd2rGwhtwc/PD126dEFxcTGkUinWr1+P9PR0REVF/ezrwsPDeSNAOOdv5wsLYwwtLS04cuQI1q5dix49emDFihUIDQ39xXNOGOKUSqW466672v04ms1m5Ofnw2w2Izg4uE0+3+FwoLm5GRqNBj4+PggODm7V929paUFBQQEAYOrUqQgNDcWQIUOwbds2OJ1OFBQUIC0tjR9r4Zy9ePEiysvL0a9fP0yePBnp6el4+eWX0djY2Kr7d7OsViuqqqpw8uRJhIaGYs6cOQgKCoLJZMKaNWtw6tQpfPjhh0hLS0NAQIDX1tPt7bYM6FpaWlBSUoKamhqcPn0aQ4YM4T/y9qi4bTYbzp49i5KSEsTHx+OBBx6An58frxiF/ALg8om/f/9+jBo1igdCAiG4czgcsNlsKC0txdtvv43y8nJoNBo4HA64XC64XC5IpVIolUpERERg/PjxvMteyKNwT9iura3Fm2++ifHjx2PYsGH8eeHzLBYL9uzZgw8++AA9evRAaGgo3yen04ny8nJ88cUXmDt3Lj788EP8+OOPsNvtyMjIQFRUFKxWK+x2O2pra+Hr64tRo0a16fEWqFQqTJkyBXl5eaioqEBLSwsCAgIAwOsCIblcjsGDByMrKwsmkwkffPABXnvtNeqlu4bo6GgAP+Ugbdu27RcT4+VyOVwuF3Q6HSVe43JQtn37duzatQsqlQorV65EcHDwVTMi7XY7bDabRw9yc3MzIiIiPOovIUXAvZEo1D+tfax1Oh12794Nu92OmJiYNilLg8GA/Px8WCwWJCYmtmrPl9PpRHNzMz8XhZ64mJgYKJVKOJ1O6HQ6Hsy5XC6IRCK4XC4cPnwYDocDSUlJ8PHxgY+PD6RSKex2u0cZAO1z7RMYDAYUFRXBZDJh8ODBCAgIgFwuh0wmw7x58/Daa6/h0qVLqKioQLdu3Sigu0G3ZUAnkUigVCphsVhw5swZDBgwADKZrF0rbYlEAqfTicLCQtTV1SEhIcEjoAMuB2pfffUVTp48iaFDh1438ZcxhubmZnzwwQc4ceIETCYTHA4H5HI5nE4nnE4nD/osFgu+/vprAMDIkSPRuXPnq5Jjc3NzUVxcjMTERNxxxx0APCtZnU6H8+fPw+VyYezYsR5BhMvlwr59+5CXlwedTgen04nY2FhMmTIFPXv2hEKh4PtsMBggkUgQFxfXBkf4amKxGBERERCLxTyovPK7eQuRSISgoCAAl4P+oqIi2Gy2ayY63+4CAwMhkUh4QnxBQQGsVut1g1+Xy8XzrYScq5sJ6pxOJ2+QCcnr3l4mFosF2dnZUKvV6NWrF2+AXvm9qqursXPnTkilUkRGRkKn0+Hs2bN48sknERsbC5lMBpfLBZPJhMrKSjDG0NjYCJVKhbi4OAQGBt7QsRJSOYTA8Zf2XRitCAkJafWyEIai9Xo9nE4ngoODIZFIftVEpOsRzke5XA5/f3/e4+nj4wOTycTzAq9M4xFyRv38/OByueBwOAAAPj4+vONA2P7KEaC2ZDabUVdXB8YYr5OF4+Tv74+4uDio1Wqqz27SbRnQWSwWGAwG2O127N+/H4888ki7dutKpVLEx8fj9OnTKCgowKZNm7Bo0SJeMQkncE1NDT755BMAgFqtho+PD79ACBhjPLG0sLAQ48ePR1BQEFJTUxEWFoaPPvoIWq0Wer2eL/RZU1ODNWvWYPfu3Xj22WcxePBgyGQy/v2lUilaWlpw6NAhPPTQQx6VrNPpxLFjx3DixAkEBgbi7rvv5vstDPnm5ORAo9FAIpHgvvvug7+/P59W73A4IJFIIBaL2/1iJ+TxhIaGor6+HvX19TeU5yJUfEKL1n3fO4pIJIKvry/i4+NRWloKjUaDI0eOoE+fPggLC7vh/fu9DyW69zzL5XLYbDacOHECO3fuxKRJk67q9Xa5XHxo0eFwwGAwoLGxEaGhodfM+3LvDbl06RLOnz+PnJwcVFZWQi6XIyIiAmFhYUhISECfPn2QlJTklcc7Pz8feXl5AICHHnromhdak8mEt956C4cPH4bRaOS/dQAYNmwY/P394XA4sGPHDuzatQsajQYA+GzYrl27YtasWdDr9UhJSUFCQsI1j5XL5cKPP/6IPXv2QKVS4amnnrpuUCdMiKmoqIBSqURaWlqbnPO1tbU4fvw4HA4HevbseUNrublcLpjNZigUiuvuv1D3XLx4EYwxyOVyHpz5+voiJCSEB21XjqQ4HA74+/tDKpXyDgv3NQGBq/OG2+vcbGxsRElJCcRiMRITE/k1QTB69GiEh4cjJiaGeuduwm0Z0JnNZhgMBjidTqjValRWVqJLly7ttrq40EoShnR++OEHzJo1C76+vnz4QehZMxqNUCgUqK2tRUJCgscPXwgytm7dinPnzmHhwoW49957eW+ESCRCly5dAABarRaLFi1CfHw8hg8fjn/+858oKSnBkiVL8Pbbb6NXr178gpWRkYHw8HDU1NTg3Llz6NevH+9Zs1gs+O6771BWVobBgwcjJCTE47tZLBZotVo4HA60tLRApVJBLpfz7yxULh3xI3W5XHzWXVFREUpLS9GlSxePnElhsgxwuZwqKyuxf/9+bNu2Dc3NzTCbzfDz88PkyZPx8MMPQ6VSwcfHp1UqQvdyN5vNsFqtEIlEPOFcCOgFDocDd999N6qrq9HY2Ijnn38eXbt2xR/+8Ad0794dYWFhUCqVvLfAZrNBq9WiubkZNpsNOp2O95CGhYXxCt/X19frhqCvRyQSeQwz+fr68hzazp07IyEhwWPNK7PZjMLCQpSUlMDlcsFisUCtViMgIMAjoLPb7WhsbMSGDRswceJE2O12LF26FFVVVR6pDu6Tn1JTU7F58+ZWz69qa4wxHD58GIwxpKenIy0tzeM59/zjhIQEVFRUoLq6mjeapVIpgoOD0dTUhFWrVqGgoACRkZHo1asX5HI5DAYDLl68iDNnzmDJkiWQyWQYOHAg/vrXv14z0DEajfj4449RVVWFXr16/ezdGoTfsNFo5MOgVquV/yakUinv3TKZTLBarbDZbAgKCvJIg/k5Qq9vYWEhlEolhg0bBrFYDIfDAbvdDq1Wi8jISJ7bBlzuOdu6dSu++uorDB48GNOmTUNycvI1lxERi8Worq7mCzW798K1tLTwHOkrA1XhOmCxWKDRaGA0GnHhwgXY7XYEBwd32IRAAGhqakJtbS1cLhcCAgI89lssFqNz586Ii4vz6Cy41siU8F/htwZcPj+qqqqQn58Pq9WK3r17IyUlBUql0qNjwj21SSCVSj16V90/y/3z3fPdhc+22WwAAIVC0ep3Q7lRt2VAJ3TXC7M0z507h/DwcL52Unu0UtxPJpvNhjNnziAiIsLjgu3r6wun0wmTyYSTJ0+ia9euiIyMBPBTS8put6OkpATJycmYOHEi70oX3icoKAgulws+Pj546KGHkJmZCalUCj8/P2zZsgU5OTl4/vnn8fHHH8PX1xfA5e754OBgVFRUICsry2P2VFNTE86fPw+r1Yr7778fcrncIwfj0qVLKC8vh9FoxJ///GdIpVIoFArExsZi6tSp6N+/f6vO/rpRTqcTDQ0NPEA2GAyorq6+apaucG5YrVasWrUK3333HUwmE3+OMYaGhgasXr0aZ86cwf3334977rnnN58zwmr77733HrRaLbRaLSwWC+8BkslkmDRpEu6//36e+9inTx8YDAYYjUbYbDYYjUbk5ubi7Nmz8PX1RVxcHObOnYvu3btDr9dj9erVOH36NCwWC1+6RZj4EhkZyYcKZ8+ejSlTpvBWvDcTKlqn0wmxWIzFixfj3XffRW1tLZYtW4Zx48Zh9uzZUCgUsNvt2LRpE7777jsAl4d+hMVwZ8+ezYNjACgpKcHXX3+NHTt2QCwWo7S0FNXV1QgODsYzzzyDkJAQHD9+HN988w3PI+vevTtvGHkTxhhyc3MhEomQkpLi8Zt3/+0rlUrMnTsXDz30ENRqNZYuXYqysjL069cPNTU1+PTTT5Gfn4+FCxdi1KhRCAgIgFgshtlsxpdffon333+fz9bWaDR8YtaV+1JVVcV7rBITE382oCstLcX7778PvV7Py1/IR1YoFLz3q76+Ho2NjbBarbBYLAgPD8eDDz6IyZMn/2JQp9Pp8M0338BoNGLSpEkIDw+Hy+VCZWUlPvvsM5w4cQJvvPEGYmJieCfC888/j/LycsjlcuzevRulpaV44403eBDpHlAI3889500YthYm7Qg9v+6vczgcKCoq4j3TxcXF+PHHHyGTyX5xpndbs1qtMBqNcDqdPKATbhMHXB5NuXJSnkD47mvXrsWcOXP45DDhmL/66qsoLi72GEqePHkyHnvsMTgcDqjVaqxduxYtLS08v9zpdCIsLAxPPPEEz7kVeo4dDgfMZjMOHTqEvn37Ijw8HMDla69MJoPdbofBYMBzzz0HlUqFf/zjHxTQtSfhxywSiWC327Fjxw4EBgZi9OjR7XKS22w2VFZWQiKR8Err888/R79+/XjrXVjqw263QyQS4cCBA/D19cWMGTP4RcHpdKKlpQUVFRU8kAN+6kYXuuCFH8bkyZPh6+sLh8OBESNGoEuXLvi///s/VFRU4IsvvsCiRYv460eNGoXTp0/j/PnzOHXqFO644w7Y7XZs3LgRLS0tyMjIQO/evT1aKjabDa+88goaGxshlUpRX1/PK6PKykrk5ORg7ty5mDt3brv30GVnZ2PRokWwWCzw9/eHSCRCbW0tb1VJpVI+JHzkyBFkZWVh586duPfeezF06FB8+OGHsFgsSEhIgF6vR3Z2Ng4dOgSDwYAxY8b85t7d2tpavPTSS1Cr1VAqlbwn0WKx8PPgrbfewrFjx/Dyyy8jKioKIpEIffv2RY8ePXDq1Cn+XmKxGCaTCUVFRVi6dCmUSiXMZjOSkpIwadIkHsAaDAYcOXKELycj5HWuWrUKubm5+Pvf/87vQHHlRVOoQIWh/Lq6OjidTqSmpiImJob3NndUxea+n0ajEQCgVCqRkpKCZ555Bs8//zxqa2uxceNGVFRUICMjA42Njdi6dSvsdjvmz5+P06dPY+fOnTh8+DBGjRqFqKgo3noXGoNCg/DcuXMIDg7GyJEjeRpC//798fjjj8Nms8FgMCAiIsJrJ60IDSGVSnXNAEqoB2QyGYKCgjx6T8rLy/Hmm2/CaDRi3rx5uPfee+Hj48MbKyKRCBaLhef6MsYQHx9/zTrC4XDg7NmzMBgMAID09HSP24tdi3DhdjgcqKqqwvr163nwIORVAj/lO0qlUly8eBGrV6/GhQsX8MADDyAlJeW6wcXBgwdRWVkJmUyGESNGQCKRoL6+Hq+88goqKip43qDVasXu3bvx/fffw2g04sUXX0ROTg62b9+OoqIi1NXV8dUCruwNEoZZTSYTGhsbERgYiOrqalitVjDGeA+gO4lEgujoaJw9exY6nQ579+5FXV0d0tPT22xyyI2SyWQe1zHhOud0OlFdXQ2j0cgbEFKplDciHA4H9uzZg//9738oLCxEv379MHLkSL4k1Ztvvgm73Y4nn3wSdrsdFy9exN69e/HDDz9g2LBhKC0txfr169HY2MjTMYTedLFYjKqqKqxevRr+/v6w2WxwOBz49ttvsX37djQ0NODpp5/GkCFDeB6j2WzGF198gW+++QaFhYXIyMiAyWTqsN/5bRnQGY1GGAwGyGQy6HQ65OfnY//+/bjzzjtb9VY/1yNUYMJFu7GxEcXFxfjyyy8xdepUiEQiaLVabNmyBf7+/jAYDCgtLcXWrVsxYcIEREREQC6X84uq0OMiLMTo3iUvVIpisRj+/v5wuVyQSCTQ6XTYvn07amtrAQDHjh3j++dyudCrVy8+1Ltx40b06dMHhw8fxoYNGzB+/HhMnjyZL/ciVAwSiYR35T/66KMYPXo0qqqqUF9fD6PRiOjoaNx9990dUpFcuHABDQ0NkEgkMBgMfNZeWFgYOnfujH79+iE6Oho1NTV47bXXUFdXh4cffhhz586FXC7H0KFDeS+PVqvF7NmzUVlZibKysla5GbhSqUTfvn3Rp08fJCYm8kR+k8mEnJwcnDt3Dl9++SWOHj2K9evXY+nSpRCLxVAqlXj44Ydx5swZ/l4REREYMmQIDh48CLVaDYvFgqlTp2Lu3LkeSeEWiwU+Pj746quv8OKLLyIqKgrr1q1DdnY29uzZg4kTJ2LUqFG8RwC4nK6g0+mQk5ODY8eO4cCBA3yInTEGhUKB1NRUTJkyBcOGDbvmMFJ7q6mpgVKphL+/P0JCQhAfH4+33noLL774IhobG3HgwAHs378fEokEI0aMQN++fTF+/HhIJBJkZWWhubkZBw8eRL9+/XjQL/zmXC4XcnNz4XK50Lt3b0yZMgUymQwymQyMMd7Q6ohe6dbk4+ODxsZG1NbWevRqu/fQicViOJ1O1NbWYt++fSgvL4e/vz8sFgtsNhtUKhUmTJjAg3z3SWBCb53QQ96pUyeekysEi8IF+qOPPoLVavWY5CT0al3ZSxUREQE/Pz9IJBL4+/ujb9++aGlp4a8XfmtdunRBYmIiHybdsmULDhw4gG3btuHChQt4/fXXERERcdVxEXKGASAqKgrx8fEwm83IysrChQsXeO9ZUFAQ9Ho9vvzyS9TW1mLevHkYMGAAamtr4efnh4CAgJ9dUkV4zuVyQavV8gl1EokEKSkpiIuL8xiZEY5FRkYGsrKycObMGZjNZkgkEkyaNKnDe9+FAFtIK9JqtcjOzkZRURF27NgBi8UCh8OBO++8E/3798fEiRMhl8tRWlqKV155BTqdDpmZmRgwYACsVitee+015OXlITo6Gk8//TQSEhLgdDphsVhgNBqRn5+PiooKxMbGIioqCiaTCfHx8bjzzjuhUCiwY8cOlJaWoqKiAmq1GiqVCj/++CPOnDmDo0eP8uu1cL4JKSuffPIJNmzYwEf8dDodn2zXEW7LgE5oGapUKjQ1NcFisaC0tBQtLS2/eRXzGyEEYgEBAejVqxcOHjwIk8mE48ePY8iQIbDb7bhw4QKys7ORlpaGc+fOwWKxoL6+HmVlZXyNLCE4UyqVfOHMKycZXNlyFSrdlpYWfsse4PJMQIFQ+QUEBKCpqQmVlZXQarU4evQopFIpMjMzeYv1ygkaVqsVANC/f3+kpKSgc+fOfF8VCkW7HN9rEfJPhHwZITD73//+h06dOgEAJk6ciPLycjQ2NkKpVGLkyJFX5UMIs9i6du2KyspKPnTwW/n7+2PixIk8+d49kX/o0KHo0qULcnJyUFJSgrq6Oo/vlZycDH9/f5jNZjidTvj4+ODBBx9EaWkp6uvrERAQgDFjxvBlJoRJHUK+3rBhwzBw4EAoFAr84Q9/QGVlJcrLy1FYWIiRI0fyXilheKqgoADr169HUVERH2oT8iSFJXkaGxuhUCh4wnNHERLPhTUPhWHTlJQUZGRkID8/HxqNBgqFAuHh4Zg0aRLv9enSpQtkMhmsVisKCgp4ION0OnHp0iU+S0/4HSYkJCAsLAwAPJaPaM/Zg21BJBIhKSkJpaWlKCkpQVNTk0fQCoDPOK2vr0d2djZ2797NZ7jr9Xo+lCoMcwI/1U0Oh4MHvEJ9EhYWxi/6Qj2j0+lw6NAhaDQafsyFHjbhHL2y/pNIJJDL5bBarQgPD8fMmTNhtVrR3NwMmUyGxMRE3iPu3nCZNGkSzp49C41Gw39z1wronE4nKioq4O/vj65du8LX1xculwt5eXlwOByQSqVITk6GWCxGbW0tamtreY6YsDqBUqnk5+a1csVEIpFHT7lwPPPy8qBSqXDHHXdclfoiDM8Ky+80NTVBIpFApVLx3rmObGgJvbHAT72uH330Ec/vFdZT3bdvH8rKytC/f39ERkYiOzsbzc3NkMvlPMizWCw4f/48dDodRo8ejdjYWH7NUygU6Nq1KyoqKhAUFISIiAh07doVZWVlSElJwfjx4wEAhYWFuHjxIhwOB5qamlBTU4PNmzdDp9NBoVDAZDJBLBbzlSZMJhPy8vKQlZWF6OhoSKVSlJSUQKfTwWKxdNhxvS0DOrVaDbPZjIEDByIxMRHff/89ioqKcOTIEYwbN44nRrYVi8UCnU6Hzp07Y8aMGejfvz9eeuklnDt3DnPnzuW5TKNGjcLSpUvx8ccfY926dbDZbFi+fDk+/fRTxMbGwul0QqlUIjk5GYcOHcJXX32FmTNn8orzyh+t+//HxcVh2bJlKC8vR21tLWbOnMmfY4zB19cXd955J9atW4fm5ma89dZbyM7OxsiRIzFw4EAolcqrlnoREpAVCgUfEhS4XC5UV1fj/PnzGD58eLstVSJITU2F0+nkw8/ChUKv1yM/Px979uzBhAkTkJWVBZFIhPj4eKSkpPC8SncKhQLTp0/HwYMH+eSP30oqlSImJgYA+DCAcJHy9fVFbGwsli5diieffJLfVkgYHkpNTcX69euxePFiXLp0CY2NjXzoIikpCQ8++CB69OgBhULBh5WE3rRZs2bxWbGMMfTp0wczZ87ECy+8gJycHMyZM4enBuTk5GDJkiUwGAy818PlcmHZsmXo2rUrjEYj3njjDRQVFUGj0eCVV17B4MGDkZiY+JuPz69lNptRXFwMmUyGbt26QalU8iGcJUuWwGQywWAwQKFQICgoiFfYjDHExsYiICAAZrMZ586dQ3l5OdLS0tDY2IhPP/0U5eXlAC6f2/7+/hg3bhzvoXb/7bkv9uqNRCIRxowZg3379iE/Px/z58/HE088ge7du/OA6uLFi/juu++QnZ2Nuro62O12+Pn5Yf78+Xj77beh0+mg1+tRWFiIXr168bxGvV6Po0eP4ssvv+TpD3a7Hbt27eITVpqbm1FWVoa1a9eipqaGD5EJua4/t3yJkCcqlUrBGENYWBgCAgJ4MCG4crJZz5498dBDD+H111+HzWbD8ePH0aNHj6vev6ysDIWFhUhJSeG92YWFhTh58iQkEgm6devG1zzcv38/bDYbgoODERoaioMHD+L7779HTEwMZs2axRuOVwZ1TqcT8fHxfGJPdnY2amtrkZubi0WLFmHQoEGQy+VXNdxFIhFiY2P5xAiJRIK0tDSec3ZlvfZr7nH7awUHByMiIgIXL17Ef/7zH2i1WkilUiQlJaFv3758qLOhoQEXLlzA888/jxUrVmDNmjUICwvDfffdh4EDB0IsFuPixYtoamqCj48PJk2a5NFpIJPJEBcXhwEDBiAjIwNmsxmDBg3Cvn37+IQ2xhhvSDudTqxZswbV1dWIiYnBlClT+GMSiYTnJf6///f/UFVVhZEjR2LWrFnQaDSYN28etFotSktLO6zOuy0Dus8++wwNDQ34/PPPERISwrv1N2zYgJ49eyIuLq5Nc390Oh1vMQYFBWHQoEF48MEHsW3bNrS0tGDAgAHIzMzEXXfdBZlMhkcffRQnTpzAuXPnoNVq8fbbb2PJkiU8327WrFkoLS3Fxo0bceTIEXTu3BlPPvkkn2EqfBf32Tg2mw1SqRSBgYEIDg5GUlIS3z+hB2f06NFYt24dXC4X9u/fj+TkZMybN4/PgrzyAtXS0oKGhgY4nU6+yG1DQwPsdjtaWlrQ3NwMiUSCp59+GrNmzWqz43st6enpyMjIwOnTp3ll3qtXL8ydOxf+/v7o1q0bxGIxDAYDvwAIgfWVC58KFxNhCFZoPf8WFosFubm5KC8v58P+I0aMgFKp5LkzBQUFHr0Zdrud9zzGx8fjtddeQ0lJCSIiIhAQEACLxQJfX180NTVd1ZMqBIzCJBvhe9rtdt6T2r17d/6dCwoKsGTJEgQFBfGJIAsWLEBVVRXWrl2LXr164dFHH8Xzzz+PDRs24Pvvv4dYLO7wHDqj0QiTyQS5XI6YmBh+bjPG4OfnBx8fH57kLARh7uthjRs3Dp988gkcDgfeffddxMXF4dixYwgLC8OgQYOQk5PDc76EWXnC64WLo3tydkcPP/9avXv3xvDhw3HmzBnU19fjr3/9K280CucncHnYUehtEoINPz8/LF++HGazGcuXL+dLuRiNRqjVathsNnTp0gUzZszAF198wRdTP3jwIORyOex2OxwOB0JCQvDiiy9i3759yMrKgkQiwaVLl/isW+Gcdl+qxr2OamxshE6nQ0BAAFwuF+rr6/Hyyy/DZrNhypQpfPhN+P13796dN/4qKiquOiaMMT7pRVig/NNPP0VWVhZaWlqQnJyMhx9+mPcQCg0xk8mEl156CRcvXkTv3r2xYMECREdHX3MZJ6GXNzIyEqGhoVCr1cjKykJERAQefvhhDB8+nKffXLlvIpEIMTExSEhI4LNk4+LiIJFIeM7alUFgexHSjRhjUKvVGDp0KJ5++mm+fJhIJMJ9992HBQsWoKKiAufPn8eyZctgsVjwzDPP8Ml9YrEYjY2NcDgcCAoKgkKh4NdzoZc8LS0NPXr0gFgsRl1dHQ4dOgSJRII+ffp4jD4IPcLNzc2YNm0apkyZAh8fH3zwwQcQiUQwGAyYM2cOrFYrgoODMW/ePIwaNYrnKEulUlitVpw6dQp33313ux1Ld7dlQCfcRcFms6G+vp6Pf+fn52PatGlYsGAB5syZ02aff+TIEZ4nInSVz5s3D/fddx9MJhPi4uIgl8v5MJYwc27fvn345JNPsGvXLhgMBkyfPh2BgYFITEzECy+8gH//+98oKChAWVkZcnJy0LlzZ6SkpCA8PByBgYH8pNPr9aiurkZNTQ18fX3x3HPPeSylIAQUCQkJmDJlCrZt2war1YoBAwbwlqJQ8QjHTuj1ERJcz549y4fjfHx8EBkZiZCQEDz66KOYPHlymx3b61EqlViyZAk2bNiAY8eOISgoCMuWLePDa0IlGhwcDIfDgfLycmzfvh2TJk3i+XFC5a7VarFz504EBQWhd+/evGftt9iyZQtWrlzJj6VEIsEdd9yB2NhYVFdXo6GhAeXl5ejWrRumT5/OlzNxb4mmpKQgJSWFJ2B3794dubm5+PLLL6FQKDBmzBiEhoby1rxQkbW0tECtVqO2thaHDh3C6dOnMWrUKDz22GO8Z2PHjh1wuVyYP38++vTpg+DgYPzrX//CokWLUFtbC41GgwMHDkAmk8Fms6Fv377IzMzkwVJHMZvNPH/LfShGqPCvtUSBEKRIpVL07dsX69atAwAcPXoUYWFhGDlyJB588EHo9XqcPn0aYrEYcXFxUCqVHnlhNxK8ecs6gAEBAXjuuedQUlLCRzSam5vh6+uLwMBApKWlYdCgQUhOTuaNH2ENtH79+mH69Ok4ffo0Ll26BIvFgtraWnTq1AnJyclITU3FhAkT4OPjg7S0NBw/fhynTp3iZdCpUyckJCRgwoQJCA8PR0NDA7KysgAAGzduhFqtRnBwMGQyGdRqNeRyOZ544gk+O1yhUPDcJqEXsL6+Hh988AEOHDgAxhhOnjyJtLQ0hIeH86FzoefdffbllS5dugTGGHJycpCbm8t74IQZl5s3b8bZs2f5vplMJhiNRjQ2NmLAgAGYNWsWVCoVb2S6T/AQ6lIhz1smk8FkMsHlciEsLAzjx4/nuYzut/sCfspRKyoqQmVlJRwOBxwOB7Zu3YpevXqhb9++Hks0AT81EK81KtHahDX0nE4nkpOTsXjxYkRERPDfpdC7+Oijj+Lbb79FYWEhysvL0bdvX4wYMYLPggXAJ1c0NjYiKysL48eP91hKSiqVQq1W4+DBgzh8+DAMBgPuu+8+pKam8uMtLMTfu3dvfr9hIT3FfRWEjIwMREdHY9y4cXwpFKGxERISAqPRiB07duCZZ55p0+N3PbdlQMfY5RXDhWBOIBKJYDKZPNZZam0ulwv5+fmQSCTo3Lkz78GQyWQei44KuRzC+kLp6elITEzkN1Y+dOgQjh07BplMhhdeeAHjxo3D66+/ji+++AJlZWWorKxEfn4+Tp8+zVt5QqsoICAAoaGhSElJwezZs69K2BYWeVQqlXj88ccRHx8Pl8uFqVOnQqFQXLNrH7h8ay0hWV8IlIRWkzDM5f5DbE9yuRyDBg3CwIEDYbfbedB65cX0nnvuwQ8//ACbzYb//ve/OHDgAM9rFIlEKCkpwdmzZxEQEIAVK1agT58+rTKjqbq6+qrb9hw8eJDfss3Pzw+jR4/G8uXLERUV5VF5A+CBF/BTr8Tjjz+OF154AXa7HR999BE2bdrEc2iSkpJQXV0NjUYDvV7vsaL8X/7yF9x11108r9LhcODMmTN8wWqVSgWlUonU1FS8//77+Ne//oXCwkIYjUb4+/vjvvvuwx//+Ec+vNmRLBYLxGIxgoKCEBoa6jEDXDiG1+o5E3oXe/fujYyMDDQ1NaFnz56YMWMGoqKiIJPJ0NLSgujoaFRVVfFbFwm/h18K0pxOJ7Kzs7F27Vq88847rdLL25YkEgkCAwPRp08f9OzZk/fyC78j93IW6g/hXPbx8cGf//xnPpnEPaVAeK1w3ISRCSEPyX2Go3BMMzMzsWHDBtTW1uLMmTM4e/Ysnwnao0cPjxu6y+VydOnSBVqtlg+1isViREdHY/78+YiLi8OWLVtQV1eH/Px8j+WkhDozJiYGY8aMueZxEYaOlUolQkJCMHPmTISHh2PlypUwm804e/Ysr++FNTiFuiQ3NxcrV65Eeno6zGYz0tPTeX6zsBj8xYsXec6gcA7rdDoUFxdj3bp1HsvpuK82YLFYUF1djRUrVsBisSAyMhIajQaXLl3CihUrMGvWLISHh/MRAGHY2m63o2/fvtdd1Lm1CDOm/fz8kJqayhuawnkhFotht9vRp08fFBcX4/z58xCJRJg2bRpfN1IoJ+HWk3a7HevXr8e2bdvQrVs3+Pr6wm6348SJE3wdQmGCRUZGhscSKdOmTcOECROQkJDAP1/4u/fee5GSkoJOnTrxXnihsSF0BgQHB2PAgAG8rDrKbRnQCUsK7Nq1yyMgESqatiQkwsbFxV13ar57K0Ugk8ng5+eHefPmoby8nAej4eHhSEpKgkQigZ+fHx555BH+Y16/fj3KyspgNBohFouhUqnQvXt3jBkzBvHx8fD39+dLeFyZSCwcp9jYWDz22GM8wfbnfuRSqRTTp0/n3+FWI3zPn1sLLCkpCcnJySgrK4PD4UBBQQHOnz8P4KdlDcRiMe6//34MGzbsmgt6/hoDBw5EVlYWtFotXC4Xr/TDwsIwYMAApKen4+677+aVl3sP6ZVBisPhgEKh4L15FRUV2LZtG8xmM0/MPnnyJK/QhArKx8cHAwcOREZGhsfiwsJ7Aj8NzQKXz8nY2FjMmzcPxcXFOHXqFAYPHozhw4fzZRY6ugdKq9XyiQ1NTU0e+Wy/NIEIuDy7c8WKFWhubkZsbCxUKhUfjpdKpVCpVKiqqvLo7biR72u1WvHPf/4TOTk50Gq1t3xAJxC+m9BIc7/wuQdCwE9DeMJ6XcJyUQA8jpXQGBEa10JP55W5iMKdJwIDA3HXXXdh586d0Ol0PE8uKSkJ8+fP50t/CPvQs2dP5OXlYdCgQYiOjua/4bCwMDz00ENISUnBxo0bcerUKb4MiPDaTp064ZFHHrluI19YvHvkyJEYMGAARowYAYvFgpSUFJSVlfG8LGGGbWNjIy5cuMAXtq+pqUFNTQ2AyyM3QlAjBIrCsVKpVHjggQdgNpuxadMmMMawf/9+JCUlIS0tjfcQu1yXb1tXVlaGgwcPQqfToWfPnujevTveffddvo7mm2++yctQSB8RLFq0CNOnT2/TPHK73Q6z2cxvyyZ8vlDmQpkKIzxC0NytWzee/yc0IlQqFYKCgtDY2IiWlhZ+WzGBcO0aOnQopk2bdtUSL0L+clRUlMekOWGkJDo6GuHh4byR5z4aBfx0Xbj33nvR2NiIoUOHttlx+yW3ZUDXtWtXzJkzB8nJyVizZo3HYo0ulwtFRUUYPnx4m3y2SCTC8OHDMXLkSN7SEE6Q601gEP5fqVSif//++Oyzz1BSUgLGGOLi4tC5c+erekLi4+MxaNAgjwuw8Fk3051+ZZfzjXw/bxYWFob//Oc/qK2txZ49e3Dp0iVYrVYoFAoEBgaiV69e6Ny5M58B2Vp5UWPGjMGoUaOuOfx3vcWur5cILuxXQEAAHnzwQZjNZtx1111obGzkU/CdTicCAwPRu3dvdO/enfcCCtPy3YehheVojEYjmpqa+IQX4YLes2dPpKenY/LkybzV6h40daROnTohKirKo9fBPVAFcM3fn/C4sHad8Lww7C6TySCXyzF27FhoNBoMHDjQo8Xu/p7Xem/hBvbPPvssnxl7K3PvAXZP3he4f2f3x4S16X7peAjvKzwnNB7dtxUeUygUeO6557Bo0SLo9Xo4HA4olUqoVKqrbkkmk8kwdepUjB07FiqVymMil7CszJgxY3D33XfDYDDw20KazWb4+PggISHhurfmEolEWL58ORYuXMiHl4Vza/Xq1aivrwdwOXgJCQmBn58fn8iRl5fHR1uEWfXCTMqIiAiEhoYiMDAQqampSElJQdeuXfnrk5OT8d///hf19fV47733eC+o0Psn9NBFRUVh3rx5GDZsGJRKJcLDw3HkyBGcPHkSJpMJLS0tHiNCwlCucI/otiQs7C6Mil0r51E4xsI1bPr06XzJLmF7iUSC0NBQrF69Gnq9HgUFBcjLy/NYUisyMhJ33nmnR1qE+6QQ9xQL95nb7uev0MN65fnt3jM6YMAA9OvXr11zEa8kYl449Uqv1yMoKAg6nc5juY0btX37dgwdOhQOhwPHjx/H9u3bodFoUF9fD5FIhDVr1lxzRhMhtyPGGF5//XXs3r0b06ZNw/Dhw5GYmMhb0cI27sOY7usgdnRQJ+QGtUVFK+TfXDnj+0YIw38deQEgv43NZvPo2XXPYQPgETC5B5LCWmVWqxV6vZ4HyULOtDDZSWhECMGEMKmtvLwcX3zxBfLz86HT6WA2m/nEgPT0dISHh2Py5MlISkriDTRhjT2j0Yj6+noetAo9TQ0NDYiPj0dqaupNNeJ/jT179uCTTz7BmTNnkJiYiPfee4/nHro3YOvq6rBnzx5cuHABs2fP5pNH3IdLf2+/n98S39yWPXTCoqHA5bXHJkyY4NGF2tYnMyHeRCS6vGL7rl27sHPnTvj6+l6VLiBUwK3VY9ma2nKm7c32YLvr6PxC8tu1xrXiWuvb/RylUonevXujV69efD03oYdNuO/zz/0OVSoVEhISfvN+/xZFRUWorq6Gr68vX05JaAi6/39ERASmTp3qkZfZketa3upuyxrlyguR+9ASIeRqaWlpkMvlaGhowFdffYVhw4YhKioKgOfQJSGkfQhpD97YMBBSFlwuF3x9fT3SR67s4Xfv/XRvnHnh4GKb+331VRJC2kRISAhf9qOlpQXl5eUew0oUzBFCbpRKpYKvry8AIDk5mU8ucw/Srsxlc39MeJx48r7QnhDS7gIDA/H444/j9ddfR0tLCzZt2oQuXbogLCyMKlZCyE0ZMWIE4uLioNFo+KSNKycruE+Iud4kK6p7PFFARwj5RcKts2bMmIHDhw9Do9HAaDR2+MLBhBDvExYW5hWzu73NbTnLlRBCCCHkVvNb4hvKoSOEEEII8XIU0BFCCCGEeLmbDugOHDiAe+65BzExMRCJRNi6davH84wxvPDCC4iOjoaPjw8yMzNRXFzssU1TUxO/sXxwcDBmz54No9H4m74IIYQQQsjt6qYDupaWFvTu3RvvvPPONZ9/7bXXsHr1arz33nvIzs6Gn58fxo4dy2+2DADTp0/HuXPnkJWVhe3bt+PAgQN4/PHHf/23IIQQQgi5jf2mSREikQhff/01pkyZAuBy71xMTAyefvppPPPMMwAAnU6HyMhIrFu3DtOmTcP58+fRvXt35OTkoH///gCAXbt2YcKECbh06RJiYmJ+8XNpUgQhhBBCfm9umUkR5eXlUKvVyMzM5I8FBQVh0KBBOHr0KADg6NGjCA4O5sEcAGRmZkIsFiM7O/ua7yvc7879jxBCCCGEXNaqAZ1arQYAREZGejweGRnJn1Or1Vfdu04qlUKlUvFtrrRq1SoEBQXxv/j4+NbcbUIIIYQQr+YVs1yXL18OnU7H/6qqqjp6lwghhBBCbhmtGtAJN+vWaDQej2s0Gv5cVFQU6urqPJ53OBxoamri21xJoVAgMDDQ448QQgghhFzWqgFdcnIyoqKisHfvXv6YXq9HdnY2hgwZAgAYMmQItFotTp48ybfZt28fXC4XBg0a1Jq7QwghhBByW7jpe7kajUaUlJTwf5eXlyMvLw8qlQoJCQl46qmn8I9//ANdunRBcnIyVq5ciZiYGD4Ttlu3bhg3bhzmzJmD9957D3a7HQsWLMC0adNuaIYrIYQQQgjxdNMB3YkTJzBy5Ej+78WLFwMAZsyYgXXr1mHJkiVoaWnB448/Dq1Wi2HDhmHXrl1QKpX8NRs2bMCCBQswatQoiMViPPDAA1i9enUrfB1CCCGEkNvPb1qHrqPQOnSEEEII+b25ZdahI4QQQggh7Y8COkIIIYQQL0cBHSGEEEKIl6OAjhBCCCHEy1FARwghhBDi5SigI4QQQgjxchTQEUIIIYR4uZteWPhWICydp9frO3hPCCGEEEJahxDX/Jolgr0yoGtsbAQAxMfHd/CeEEIIIYS0LoPBgKCgoJt6jVcGdCqVCgBQWVl501+YdCy9Xo/4+HhUVVXRXT68DJWd96Ky815Udt7r15QdYwwGg+FX3dveKwM6sfhy6l9QUBCd4F4qMDCQys5LUdl5Lyo770Vl571utux+bUcVTYoghBBCCPFyFNARQgghhHg5rwzoFAoFXnzxRSgUio7eFXKTqOy8F5Wd96Ky815Udt6rvctOxH7N3FhCCCGEEHLL8MoeOkIIIYQQ8hMK6AghhBBCvBwFdIQQQgghXo4COkIIIYQQL0cBHSGEEEKIl/PKgO6dd95BUlISlEolBg0ahOPHj3f0Lt3WVq1ahQEDBiAgIAARERGYMmUKLly44LGNxWLB/PnzERoaCn9/fzzwwAPQaDQe21RWVmLixInw9fVFREQEnn32WTgcjvb8Kre9V199FSKRCE899RR/jMru1lVdXY0//OEPCA0NhY+PD3r27IkTJ07w5xljeOGFFxAdHQ0fHx9kZmaiuLjY4z2ampowffp0BAYGIjg4GLNnz4bRaGzvr3JbcTqdWLlyJZKTk+Hj44POnTvj73//u8cN2ansbg0HDhzAPffcg5iYGIhEImzdutXj+dYqpzNnzmD48OFQKpWIj4/Ha6+9dvM7y7zM5s2bmVwuZx999BE7d+4cmzNnDgsODmYajaajd+22NXbsWPbxxx+z/Px8lpeXxyZMmMASEhKY0Wjk28ybN4/Fx8ezvXv3shMnTrDBgwezoUOH8ucdDgfr0aMHy8zMZLm5uWzHjh0sLCyMLV++vCO+0m3p+PHjLCkpifXq1YstXLiQP05ld2tqampiiYmJ7LHHHmPZ2dmsrKyM7d69m5WUlPBtXn31VRYUFMS2bt3KTp8+zSZPnsySk5OZ2Wzm24wbN4717t2bHTt2jB08eJClpKSwRx55pCO+0m3j5ZdfZqGhoWz79u2svLycbdmyhfn7+7M333yTb0Nld2vYsWMHe+6559hXX33FALCvv/7a4/nWKCedTsciIyPZ9OnTWX5+Ptu0aRPz8fFha9euval99bqAbuDAgWz+/Pn8306nk8XExLBVq1Z14F4Rd3V1dQwA279/P2OMMa1Wy2QyGduyZQvf5vz58wwAO3r0KGPs8o9GLBYztVrNt1mzZg0LDAxkVqu1fb/AbchgMLAuXbqwrKwsNmLECB7QUdndupYuXcqGDRt23eddLheLiopi//rXv/hjWq2WKRQKtmnTJsYYYwUFBQwAy8nJ4dvs3LmTiUQiVl1d3XY7f5ubOHEimzVrlsdj999/P5s+fTpjjMruVnVlQNda5fTuu++ykJAQj/py6dKlLDU19ab2z6uGXG02G06ePInMzEz+mFgsRmZmJo4ePdqBe0bc6XQ6AIBKpQIAnDx5Ena73aPc0tLSkJCQwMvt6NGj6NmzJyIjI/k2Y8eOhV6vx7lz59px729P8+fPx8SJEz3KCKCyu5V9++236N+/P6ZOnYqIiAhkZGTggw8+4M+Xl5dDrVZ7lF1QUBAGDRrkUXbBwcHo378/3yYzMxNisRjZ2dnt92VuM0OHDsXevXtRVFQEADh9+jQOHTqE8ePHA6Cy8xatVU5Hjx7FnXfeCblczrcZO3YsLly4gObm5hveH+lv/ULtqaGhAU6n0+PCAQCRkZEoLCzsoL0i7lwuF5566inccccd6NGjBwBArVZDLpcjODjYY9vIyEio1Wq+zbXKVXiOtJ3Nmzfj1KlTyMnJueo5KrtbV1lZGdasWYPFixdjxYoVyMnJwf/93/9BLpdjxowZ/Nhfq2zcyy4iIsLjealUCpVKRWXXhpYtWwa9Xo+0tDRIJBI4nU68/PLLmD59OgBQ2XmJ1iontVqN5OTkq95DeC4kJOSG9serAjpy65s/fz7y8/Nx6NChjt4VcgOqqqqwcOFCZGVlQalUdvTukJvgcrnQv39/vPLKKwCAjIwM5Ofn47333sOMGTM6eO/Iz/n888+xYcMGbNy4Eenp6cjLy8NTTz2FmJgYKjvyq3nVkGtYWBgkEslVM+w0Gg2ioqI6aK+IYMGCBdi+fTt++OEHxMXF8cejoqJgs9mg1Wo9tncvt6ioqGuWq/AcaRsnT55EXV0d+vbtC6lUCqlUiv3792P16tWQSqWIjIyksrtFRUdHo3v37h6PdevWDZWVlQB+OvY/V19GRUWhrq7O43mHw4GmpiYquzb07LPPYtmyZZg2bRp69uyJP/7xj1i0aBFWrVoFgMrOW7RWObVWHepVAZ1cLke/fv2wd+9e/pjL5cLevXsxZMiQDtyz2xtjDAsWLMDXX3+Nffv2XdV13K9fP8hkMo9yu3DhAiorK3m5DRkyBGfPnvU48bOyshAYGHjVRYu0nlGjRuHs2bPIy8vjf/3798f06dP5/1PZ3ZruuOOOq5YHKioqQmJiIgAgOTkZUVFRHmWn1+uRnZ3tUXZarRYnT57k2+zbtw8ulwuDBg1qh29xezKZTBCLPS+/EokELpcLAJWdt2itchoyZAgOHDgAu93Ot8nKykJqauoND7cC8M5lSxQKBVu3bh0rKChgjz/+OAsODvaYYUfa1xNPPMGCgoLYjz/+yGpra/mfyWTi28ybN48lJCSwffv2sRMnTrAhQ4awIUOG8OeFpS/GjBnD8vLy2K5du1h4eDgtfdEB3Ge5MkZld6s6fvw4k0ql7OWXX2bFxcVsw4YNzNfXl3366ad8m1dffZUFBwezb775hp05c4bde++911xSISMjg2VnZ7NDhw6xLl260NIXbWzGjBksNjaWL1vy1VdfsbCwMLZkyRK+DZXdrcFgMLDc3FyWm5vLALB///vfLDc3l1VUVDDGWqectFoti4yMZH/84x9Zfn4+27x5M/P19f39L1vCGGNvvfUWS0hIYHK5nA0cOJAdO3aso3fptgbgmn8ff/wx38ZsNrMnn3yShYSEMF9fX3bfffex2tpaj/e5ePEiGz9+PPPx8WFhYWHs6aefZna7vZ2/DbkyoKOyu3Vt27aN9ejRgykUCpaWlsbef/99j+ddLhdbuXIli4yMZAqFgo0aNYpduHDBY5vGxkb2yCOPMH9/fxYYGMhmzpzJDAZDe36N245er2cLFy5kCQkJTKlUsk6dOrHnnnvOY9kKKrtbww8//HDN69uMGTMYY61XTqdPn2bDhg1jCoWCxcbGsldfffWm91XEmNvS1IQQQgghxOt4VQ4dIYQQQgi5GgV0hBBCCCFejgI6QgghhBAvRwEdIYQQQoiXo4COEEIIIcTLUUBHCCGEEOLlKKAjhBBCCPFyFNARQgghhHg5CugIIYQQQrwcBXSEEEIIIV6OAjpCCCGEEC/3/wGLLhmpAMeVqAAAAABJRU5ErkJggg==",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "YH:  nominating any more Labour life Peers\n",
      "GT:  nominating any more Labour life Peers\n",
      "CER: 0.0, WER: 0.0\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "YH:  is to be made at a meeting of Labour\n",
      "GT:  is to be made at a meeting of Labour\n",
      "CER: 0.0, WER: 0.0\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "testing(train_set, 3, model, two_heads=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### Performance on test data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "YH:  Become a suceswith a dise and hey pres . You're a star. . Ralysmg with\n",
      "GT:  Become a success with a disc and hey presto ! You're a star ... . Rolly sings with\n",
      "CER: 0.1951219512195122, WER: 0.5\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "YH:  assurethess \" Belle Bela tane \" ( Parlophane ) a Wvely song thal chonges tempo mitway .\n",
      "GT:  assuredness \" Bella Bella Marie \" ( Parlophone ) , a lively song that changes tempo mid-way .\n",
      "CER: 0.17204301075268819, WER: 0.5555555555555556\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "YH:  I donit thinw he will slorm the charls wilth his on , pal i t's a good start .\n",
      "GT:  I don't think he will storm the charts with this one , but it's a good start .\n",
      "CER: 0.14102564102564102, WER: 0.5555555555555556\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAnQAAAB3CAYAAACKYt1IAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjguMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8g+/7EAAAACXBIWXMAAA9hAAAPYQGoP6dpAAAvgElEQVR4nO3deXxM9/4/8NeZPcskEdkjIV+UqKWKEEVbcm39WoqW27QUtUYljVpzFbcVWq2LospFqrYvvbba/WIJFUEIEgQVkohM9sxMMvt8fn/kMeeayqqRZJr38/GYBznnM2c+53zOnPM+n204xhgDIYQQQgixWYL6zgAhhBBCCPlzKKAjhBBCCLFxFNARQgghhNg4CugIIYQQQmwcBXSEEEIIITaOAjpCCCGEEBtHAR0hhBBCiI2jgI4QQgghxMZRQEcIIYQQYuMooCOEEEIIsXH1FtCtW7cOLVq0gEwmQ/fu3XH58uX6ygohhBBCiE2rl4Du//7v/xAZGYlFixbh2rVr6NSpEwYMGICcnJz6yA4hhBBCiE3jGGOsrj+0e/fu6NatG9auXQsAMJvN8PPzw6effop58+ZV+X6z2YysrCzI5XJwHPeys0sIIYQQ8tIxxqBSqeDj4wOBoGZ1bqKXlKcK6fV6JCYmYv78+fwygUCAkJAQxMfHl/senU4HnU7H//3kyRO0a9fupeeVEEIIIaSuZWRkoFmzZjV6T50HdHl5eTCZTPD09LRa7unpibt375b7nmXLlmHJkiXPLc/IyICTk9NLySchhBBCSF1SKpXw8/ODXC6v8XvrPKB7EfPnz0dkZCT/t2WHnZycKKAjhBBCyF/Ki3Qnq/OAzs3NDUKhEAqFwmq5QqGAl5dXue+RSqWQSqV1kT1CCCGEEJtT56NcJRIJunTpgtjYWH6Z2WxGbGwsgoOD6zo7hBBCCCE2r16aXCMjIzFu3Dh07doVQUFBWLVqFUpKSjB+/Pj6yA4hhBBCiE2rl4Bu9OjRyM3NxRdffIHs7Gy89tprOH78+HMDJQghhBBCSNXqZR66P0upVMLZ2RnFxcU0KIIQQgghfwl/Jr6h33IlhBBCCLFxFNARQgghhNg4CugIIYQQQmwcBXSEEEIIITaOAjpCCCGEEBtHAR0hhBBCiI2jgI4QQgghxMZRQEcIIYQQYuMooCOEEEIIsXEU0BFCCCGE2DgK6AghhBBCbBwFdIQQQgghNo4COkIIIYQQG0cBHSGEEEKIjaOAjhBCCCHExlFARwghhBBi4yiga2R69OgBrVZbaRqdTgetVov79+9jxYoVKCoqqnK7Fy5cwLJly7Bp06ZaymndM5lMMBgM9Z0NQsgzdDodYmJi8OOPP8JoNMJoNNZ3lghpkCiga2S0Wm21LogCgQDFxcV4+PAh9Hp9lend3Nzg6OiIO3fu1Dg/Fy9ehMlkgtlshtlsrtH7a9P27duxZs0aGAwGmEwmmEymessLIaSMSCRC8+bNcezYMfpOElIJCugamb59+4IxVmkag8EAo9EIuVwOhUKB0tLSKrfbsmVLjBkzBnfv3kVpaSnOnTtXrcCopKQEGzZsQElJCUpKSrBx40Zs27YNer0eRqOR30Zd6NevH/Lz86HVamEymao8ToSQmsvIyMCPP/5Y7fSMMTg4OEChUIAxRt9LQipAAV0jc+3aNeh0ukrTiMVicBwHJycnmM3matXQcRwHgUCA0tJSKJVKbNmyBTdu3OBr3ioiEAiQm5sLAJBKpRg8eDBOnjyJzMxM/sJdVxdwDw8PZGZmQiQSQSCov6/G06dPsX379nr7/Nr2+PFjaDQavga2Pmthq6M65/vLtH//fqsa6/LOf8aY1fE0GAw4dOgQMjIyGvwxzsjIqFY3DguO4+Du7g6hUFith0sLxhiGDBmC+fPnIycnx6plwmAwUGD4J+n1+gZ9njVGFNA1MoyxagUrHMfBzs4OSqUSOTk5VaY3m83gOA729vYQCAQIDAzE9evXq7y5GI1G/qlbr9fDzc0Nr7/+Os6cOVPnF1yO41BYWAiDwVDnF6rk5GTExcUBAFxdXZGamlqjPGg0GgQHB6NHjx4oLS2tt35GRqMR27dvR1FREV/Tm5SUhGPHjr3UgG7evHk4f/4838eqoldVHj9+jKlTp6KwsLDOjqFareb/bzKZsHPnThQWFlYZ0KnVaixatIj/Dnl7e+OHH36o8iGqvmVlZSEgIKDa6TmOg5eXF0wmEwQCATiOq/Z7W7ZsiatXr/LfLYvRo0fjs88+oz6zL8hgMGDevHnYtGkT9WtsQCiga2SEQiEKCwsrTWNpbhQIBJBKpdV+KhaLxZBIJOA4DkKhEFeuXIHBYKj0Amy5oHIcB5FIBI7jEBgYiLNnz/K1EHXV5GppHq7LZl4Lg8EApVLJ/33r1q0Kb8omkwnFxcWIjY3lAyeTyYTS0lKoVCrk5+fXW+2DwWDAvn37+Fous9mM06dPIyMj46XmKSwsDOvWreMfDl70s5ycnBAUFITs7OxazmGZn3/+mf+/0WhEXl4etm7dyge/QqEQQUFBSElJqXJbAoEA9+/f57+rgYGBuHbt2nP7zhhDRkZGre/LixKLxWjVqlW105vNZpSUlEAkEtWoXDmOw7x58yAQCJ4LOL777jtoNBokJCRUe3t1xXIdKu9lNBr5f+Pj43H9+vUXCkqraqWpCmMMnTp1QlJSUoN+eGhsKKBrZBwcHKwCh/JYghpL7UBV6YGyi6flxmIymSCRSKDT6cBxXKUBneVia2kCMRgM8PX1hVKphFqtrlbtYEVq2nTGcRxkMhlKS0vrPCAqKSlBVlYWgLKLZUlJSYVpTSYTTpw4gcWLF2Pt2rVWAahEIoFcLq+TPJeHMQaNRgOZTMbnVa/Xw8HBoUY1KzVlMpmg0WgAwOqcq2mz/aNHj3Djxg3k5+e/lHz6+vry/7ecb3q93ioIzcjIgEQiqXQ7lrTFxcX8A5jRaERJSclzDyOMMWRmZtbynrw4FxeXGn2/BALBC3eBEAqF0Ol0UKlUVsudnZ3x6quv4vLlyy+03Zfp2YeSP74s6wEgNTUVN2/erPbDpyUYzM7OxowZM/h+ypUFkBUxmUwoKCjAkydP6r2LAvkvCugaGUdHxypr6AwGAx4/fgyVSgU7OzukpaXxN+qKnsYswVhoaCjs7OwgEAhgZ2cHkUhU6WdZapgsFyyBQAC1Wg2DwYDIyEhERUXh5s2bAICYmBirp1GNRoNVq1Zh0qRJ+OCDD7Bnzx5+3eXLlzFr1iyo1WosWLAA+/btq9bx0Wq1+OmnnxAeHm5VS2IymaDVanHhwgUolcoXqsWzBMeWQPnZfTGbzUhOTrbKR0U3vczMTPz8888YPnw4vL29YTKZwHEcevfuDUdHxxo9MTPGKq2BNRqNfKBUHU+ePIFarUZSUhLfiV2n00EqlcJoNPLBS21TqVQoKirim0qfbaq0/L86n3vmzBncuXMHgYGBfPpbt25hxIgRtRLk9e3b1+pvs9mMuLg4qxokmUyGJ0+eVNjcCpQFg0ajESqVCmazGUajERzHwWw28wN7LAQCAYKDg2E2m1FYWIgVK1bg3XffLXe7JpMJKpUKt2/fxs8//4w7d+7AYDBAp9PBZDJBp9M9V3tp6S6RmZmJ8ePH4+zZs5UGkM2aNbMKbKtiaUJWq9U1rn0ViUTlflclEgk6deqEq1evVntb9am8fQ4ICICdnV2N+hWaTCakpaXBzs7O6ntRWQBZHrFYjN9//x0KhYJq6BoQCugaGZ1OBw8Pj0rTfPnll3j//ffx+eefw2g0QiaTITk5Gf3798cnn3yCgoICJCYmYt++fVi5ciVGjBiB5ORkuLq6Yvjw4fjll19w4cIFNGnShH8CLI/ZbEZ4eDjUajUyMzP5zt35+flQqVSYNm0axo4dy89td/PmTSgUCv79Go0GMTExmDJlCv7xj39YBXSxsbF4/PgxtFotSkpK4OTkVOWxsVzA3NzcMHPmTHz55Zf8uoyMDCxbtgxTp07F8OHDUVRUVOmFTKfT4fvvv4dWq8X58+dx4cIFqNVqHDx4EMOHD0daWhr+8Y9/8MGSXq/nA21Ln6iKarRSU1PRqlUrjB49GqNHjwbHcRCLxfDy8gJjDGKxmB/MsmDBAgwdOhTvvvuu1fY3bdqE3NxcHD9+HFu2bIFarcaaNWuQmZmJjIwMaDQa7NmzB2PHjsWIESOqHbxGRUVBrVZDqVTy/QDd3d1x//59XLt2DQUFBfxxNhqNuHXrFiZPnoz27dtj9+7dlTYf6fV66PV6XLx4EcHBwRg/fjw0Gg0MBgMKCgqg0+nw6aefYvbs2VAqlVCpVPjb3/6G6dOn47fffsPDhw8BANnZ2Th48CAiIyPxwQcfWO1bZmYm3/RpcezYMeTk5ODMmTNV3rwYY1ZN4REREQgJCcGePXv44PaXX37hy1ggEECpVPI13AAQGBiIo0eP4sMPP8TEiRPLnTfS0ofO0nQmFAohFouh1+sRFRWFMWPG8HkdP348kpOTwRjD3LlzsWXLFnh7e5eb/8LCQvTp0wfjx4/Hxo0bMWvWLJw7dw7z589HQkIC5s6di+nTp1sdM5PJhO+++w47d+7ExIkT8csvv2D69OkVHqOAgADodDp8++23iIyMxO7du/myXbx4MUaNGoXU1FRERkaCMYY7d+5g1apVMBqNVT4g/vEYFRcXlzuwSyAQQKVSwdnZGXq9Hvfv30dISAg++uijam+/LpV3LWjRogVOnjxZ4XfTEvC/9dZbiI2N5a8prVq1Qm5u7gv3e2OMwWQyIScnB87OzjSVTANCAV0jo1arq6xxSUlJwcaNG/Hw4UMIBAIUFhbi+PHjeP/995GQkIB///vfkMvl+PrrryEUCjFt2jSrm5TZbIadnR18fX0rbS4RCASIiYmBo6MjvL29YTAYIBKJkJycDKPRiNatW8PPzw9paWkAymphnr25ZWdnQ6fTwc/PD46OjlY1B7dv34ZQKITRaEROTg7c3NyqPDZ6vR5ZWVno1asX3Nzc+JF4paWl+Pzzz9GqVSts27YNUVFRWLt2baVNiFlZWdBoNCguLkZERAS+/fZbGAwGDB06FBqNBjt37kRqaiofwNy9exc+Pj78+8VicYXHzdXVFWlpaXz/KaFQiPPnzyMpKQlGo5GvTTGZTHj06BEcHR3RsWNHjBw50qpZPDo6Gnfu3MHEiRMBAE2aNEFKSgrCw8Nx4MABHD9+HOHh4dBoNNWuBVi6dCnc3Nzw1ltv4Y033oBIJEKXLl2wd+9eXLx4EWPHjkV0dDQAIC8vD0uWLEHLli2xfv16bN269bmmsWelp6dj1qxZ2LdvH0JDQ5GcnIy0tDQYjUZkZGRAp9MhOjoaN2/eRGxsLLZu3QoXFxdMmDABc+bMQXh4OICyc+Pbb7/FgAEDsGrVKqvg7bfffkNubi5OnjzJ3/Du3r2LgIAA/O///m+VTX+MMaxbtw7vvfcezGYzFi1ahPXr1+Obb77huxCsX78earUajx49wsOHD6HVavmpcgBALpfj2rVrmDhxIkQiUbn96eLi4jB//ny+X5ler4dSqURBQQGmTJkCV1dXvqvEyJEjsXv3bj5d06ZNrR5WnrV48WKEhYXh5MmT2Lt3L7KysuDs7Ixbt25hy5YtGDVqFGQymVXg/eDBAxw+fBghISF49OgRsrKy4O/vX+ExEggEWLRoEY4cOYIePXpg3bp1+O2331BYWIjExERMnDgRubm5yMvLg16vR0BAAHr16gW9Xo/Y2NhKv3dfffUV7t27BwDIz8/H6NGj+ZaFZ2uc9Ho9fv/9d4waNYrPT1FRUaUd/I1GI9avX48OHTqgS5cuNW7GtowKtdR+WbqaxMTEYPbs2Th79ix/XC21zFqt1irvln8tTZ4tWrSAVCqt8POioqKwYMECxMfH89vR6XRIS0vj+yxbmuyLiopQUFDA1/halJaWIiIiAgsXLsTs2bP5ZXl5efxgONIwUEDXyDg6OlaZRqfToXXr1oiOjsbDhw+h0Whw7tw5DBo0CD179kR6ejpcXFyg1WoxZMgQvokVKGvieP/998FxHJo1a4YtW7Zg3rx5FX6WUCjkm1QsTTqWwMRsNkMul/MXuaKiIqv8P3nyhK+VsvTTssjPz4dIJMKRI0dw//79anXCtvQpMRqNKCoq4oPA4uJiKBQKcBwHiUSCVatWoVu3bpXW1uTm5vK1CTqdDgqFgu9buGPHDhQVFUEul/MX44sXL8LPzw9AWXOrUCis8ELZtm1bKJVKfPHFF9i7dy8KCwthMpmgVCqh1Wr5912/fh3u7u6Ijo7GzJkzUVhYyK/v2LEjkpOTMXbsWHAcB4VCgWPHjmH9+vVo2rQpfv75Z3z11Vfw9fXl810djo6O/BO8paN/y5YtwRjDyJEjERMTg1OnTgEApk2bhvz8fJw4cQL/+te/IJVKYWdnV+G2o6OjcffuXURFRWHcuHG4ePEiWrRoAZPJhKysLEilUjg4OECn0+Hu3bs4ceIE1q9fj7Zt20IikaC4uBhA2TlaUFCA7t27w8XFxeoz2rRpAycnJ/Ts2RMlJSVgjKFbt27Iycnha/gqYzAYoNfrkZubC4PBAHt7ezRv3hxDhw5FXl4e8vLy0KtXL9jb26NJkybQarWQy+VWDz7/7//9P8hkMrRu3RqOjo7lHvsuXbogMDAQXl5e4DgOOp0O9+7dg0wmQ6tWraDX6/nzQKvV4vHjxwDK+mp6eHhUeB14+PAh3nvvPYjFYjg4OMDDwwP5+fnIyclBeHg42rdvz3eJsCgtLYWzszMWL14MtVqN7777DsuXL6/wGDHG8PjxY+zatQuDBw/Gv/71L/z666/Ys2cPhg4dii5duqB58+Z8wCAWi9GiRQvI5XL07t270qbA5ORkHDlyBADQtGlT7N27FxzHPfegEBcXh8TERHTv3h0GgwGFhYXYsWMHNmzYUGEtYElJCQ4fPozQ0FBIJBJ8/fXXFebDwtLNYNu2bRgwYAAiIiKwZcsWjBw5EhqNBkuXLsW2bdtw9OhRHDp0iO/icOrUKURGRmLmzJmIj48H8N+uA0eOHMGUKVOQmJiIZs2aVfiQwXEcHjx4gG7duiEiIgLZ2dkoLCyEUqnkm9YtD+AKhQIREREYPHgwBg4ciFWrVvHXty1btiA1NRX5+fmYO3cusrOzMWnSJDx9+hR5eXmIjIzE3bt3qzwW5OWjgK6R0Wq1cHd3rzRNmzZtIBKJ0KpVK7i5ufH9vYRCIYYMGcI3i7q4uPA3nKysLP4CIBQK4e3tjWPHjuE///lPpR2PLZ3njx49isWLFyM+Ph4qlQoLFy6ETCaDWCzmn/ZVKhXs7e35965duxYrV66Evb39cyPgVCoVVCoV1q5dC39//2p13M3KyoK7uzt8fHxQVFTEBxfOzs4wGo3YsmUL5syZg8mTJ6N3796VNlmoVCqkpaVBKBRi4MCBkMlkfO3YvXv34ObmxtdoAmW1jUFBQQDKaq4swV1Fx+w///kPNm7ciOvXr2PixIno0aMHHB0dIZfLkZOTgwkTJuCXX37BoEGD+IDB0dERQqEQjDEkJSUhICCADyiTk5ORnp6OSZMmobCwELNnz4aTkxMKCwv50cvVYW9vj5KSEqSkpCAqKgqbN2/mawst27HURJWWlmLz5s3Ytm0boqOj8dNPP+HcuXMVbvvRo0do1qwZpFIpJBIJ34QkFAphNpv52gpLgNqsWTM4OjpCpVJZNdfFxcVBJpOVe+Pu378/5HI5srOzMXz4cHAch48//hjLli1DRERElc1UliZvoCyotzSl5ufn88emRYsWEAqFcHZ2hpOTE18uaWlpMJvNSEhIgJeXF18L4unp+dznODg4QCAQwN3dHXl5eXjzzTexe/duuLi4QCqV8scbKJuKxfI9GzVqFAoLC3H27NkK90Gj0UCn0yEhIQGFhYXQ6/WQyWRo3rw5pFIpVCqV1cOGXq/HmDFj+CZ6SxlVxGAwwMHBgd/v33//Hb6+vjh16hRCQkIgFotx8OBBZGRkQCAQQCQSwdvbGyKRCEqlEqtXr66wme/x48f8vJaWz2rSpAn69+9vtbxv377YvHkzHB0dIRKJ4ODgAIlEUukDxdy5c/H1119jzJgx8PT0RHp6epX9+UwmExISErB9+3Zs2LAB8+bNA8dxcHFxQX5+PhITE7F//35cunQJarUaR44cwW+//YY7d+5g5syZeO+99zBnzhycPn0a+fn5WLp0KUpLSxEdHQ2O47Bt2zYcPny43M+2fOfPnTuH7OxsLF26FN9//z0WLFiA0tJSCAQCaLVa6HQ6rFmzBv7+/ti/fz+ePn2KkydP8vs2cOBAZGVlQS6XQy6Xw8vLCxs3boSjoyO6du2KqKioKn9OktQNCugaGbPZXGUtnUKhgNFoRHFxMRwdHWFvbw+xWAyhUIiYmBi4uLhAp9Pxc0JpNBr+RgqUXUiCgoLw6NEjiMXiSvulSKVS+Pv7Y/fu3dBqtejatSu0Wi06dOgAoOwJ15JfgUDA36SAsotlQEAAJBIJhEIhvLy8+HV2dnYoKSlBjx49+A7hVSkuLoa3tzffl8/Sz0gsFsPOzg5vvvkmvvnmGwQFBUGhUCArK6vCC7parYa7uzvfn0ytVvOd1a9fv44HDx6gZcuWfFBhMpn40alKpdKqtvGPLE3TPj4+mDRpEh+k+fj4wGw2IykpCf7+/igsLOQDnuTkZDRr1oyvEc3MzMRrr73GHzu9Xg+DwYBXXnkFdnZ2fC3P3bt3a9RvSSwWQyqV4ttvv8WtW7fQo0cPKJVKfpBMbm4unJ2dAZSdiykpKRAKhfD19UVJSQkuXbpU4bZFIhGePn2K0tJSfuqWnJwcnDp1CsnJyQgKCuKntxCJRHBxcYFer0dcXBwEAgF/fjg4OMDX19eqqdWiTZs20Ol0WLduHbp168Y3j3l7e1e76Vmn08FgMODSpUvQ6/UoKSmBwWCAXC6HXq/nP9cysEEkEiE/Px+LFi2CTqeDSCRCx44dAYB/b3m0Wi30ej2Ki4shk8lw7949yOVy/nywlJvRaETXrl3BGMOrr76Kjh074qeffip3m15eXrh//z4uXLiAJUuWwMfHB7dv34aHhwcf1DPG+AARKKuVffDgAcxmMz8I4dm59f7IZDLx3SEs/Sg9PDygVqv50dA7duywKh/LROf37t2rcqoMy8Mbx3H8uaLT6fDJJ59Y5dlyLWGMwcnJCRkZGZUG7AqFAt7e3rh9+zbS0tIgk8mqbG5kjOHQoUPo06cPvLy8IBaL+drP+Ph4/joHAN26dYNKpcLhw4cxbNgw+Pj4ICgoiB9o8tNPPyE1NRWDBw9GkyZN0Lt3b0il0goHdjDGMHr0aGzevBmMMSxZsgQRERGYMWMG5HI5v6+ZmZlITU3FhAkTYGdnB5lMBoFAwPfPtDT7Wq7Dlhphxhjc3d1RXFyMJUuW0OCIBqD6V2ryl6DX66t8mpJKpYiOjkZ8fDyioqLQuXNnMMbw8ccfw8nJCTNmzOADMaFQiLy8PLz11lv8BdLOzg7Dhg1D79694ezsXO6N89nP2rlzJ/83YwwHDhyAXC4Hx3FwdHTE6tWrAZTNoP/sk/+uXbv4G4hEIsGaNWv4db/++ivf5FXdPh4dOnTAP//5Tzg5OaFdu3b84BGhUIhevXrh7NmzfC2KVCpFZGRkhduy3OQiIyPRp08feHt7IzIyEgaDAa+//jr69OmDgoICqyC4adOmAAB3d3eEhIRUmO8DBw7A1dUVT548wa5duzBixAgAZYM5zGYz9uzZgx9//BH37t1DdHQ0ZDIZhEIh/vnPf/I3ew8PDwwcOJBv6g4MDES/fv3g4+ODdevWQSaTIS8vD7t27UJYWFi1jh9QFnTt3bsXSUlJ6NChA5ycnPDVV1/B1dUVGo0G9+7dw6BBgwAAO3fuxNixY1FaWgoHBwcYjUasXbu2wm3/8MMP2LZtG95//32IRCK4ubmhT58+GDBgALZv345XX32VP46hoaGYNWsWhg8fDn9/f4SFheHRo0cAgISEBL4m6I86dOiATZs24fTp05g4cSIMBgPCw8ORnp6OVq1aVRpoA2Xnr1arxaFDhzBt2jRs27YNRqMRK1euhFQqhUKhQKdOnfh+jHZ2dtBqtdixYwffLCoSifDJJ59AKpWiX79+5dZ2cRyHFi1a4MSJE/jss8/w6aef4uDBgxg8eDCAsgcKS0B37tw5fPfddygoKMDChQuhVqsrrIn6+uuvERYWBoVCgVatWmHZsmUQCoWYMmUKhEIhX+P17He6TZs2uHnzJkaNGgVXV1d4eHjglVdewdSpU62Oi+V8lslkUKlU+OCDDyAUCtG0aVPMnTsXO3fuxPr165GWlgZPT08MGzYMBoOBf0hYsWIFpk6dinHjxpVbdpZj98orr/B/y+VylJaWYufOnVa1+380e/ZszJw5EyNHjrTK97OKi4sxc+ZMpKWlQavVYvz48RVuz4LjOKSkpGDhwoVITU3FxYsXcfLkSUydOhX//ve/sXTpUhiNRhQUFKCwsBA9e/bEyZMn4enpCaFQCAcHBwwePBgrV65ESkoKZs2axX+HmzdvjgEDBqCgoKDCz/7b3/4Gk8mEyZMn49ChQxCLxfD19YVUKoVWq0Xfvn0hl8sxc+ZMuLq6wmQy4e2338aVK1eQnp6O9u3bIyUlBX5+fhg/fjzfvFtaWgqxWIzOnTvDzc0Njx49QmlpabW69JCXh2P1NQPpn6BUKuHs7Izi4uJqjV4k/zVhwgSsXbu20oub2WyGTqfja+WA50dZWdKIRCLcvHkTfn5+VY6ebUxyc3OxYsUKfPLJJ/Dw8OCfeIVCIR9kCoXCGtV+WVhGOJrNZshkMn4y59pm6ZtY2QCN6li7di08PT3Rv39/ZGVlIT8/H7169arFnJbVQln6oVl+UcBSu2Y5hy1BxenTp9G1a1c+OKvsgeNlsszRN23aNHTu3BkjRoyAr68v5syZgy+//BIikQgajYbvUvBHllGcDg4OVus5joNareZvrs/WfNU1jUaDvXv34tq1a/zDEmA9b6Alb5aRxE2aNEFQUBAkEkml16nyqNVq/tdqLJ//7rvvokWLFlixYkW5czT+/vvv2Lx5M+Li4tCkSRP8+uuv5W67tLQUKSkpaNasGTQaDQ4fPoywsLBKzx+j0Yjjx49j6dKlcHBwwGuvvYbz589j06ZNuHbtGmJiYvha5WnTpmHYsGGYNm0a7O3t0alTJ+h0OhQUFOCjjz7Crl27kJycjL59+yI4OBhqtRoXL15ESEgIunXr9txnf/HFF8jNzUV2dja6deuGmTNnQiQS4c6dO5g8eTICAwOhUCig1WrRrl07SKVSaDQaTJgwgR+Y8+GHH2LevHno06cPevbsyTeTP3nyBMOGDUObNm1gNpuhVCpx5MgRcByHW7duYc6cOTh27FiNyo6U+TPxDQV0hBBCXoqCggKMHj0aO3fuRNOmTev1N5IrYhkMJRKJ+N+xri2WZk3LACGFQoHPP/8c4eHhePvtt8t9j1qtxoMHD5CdnQ2JRILg4GC+JcJsNkOlUiEvLw+Ojo7w9PTkm3Grs5+WX5T5+OOP0b9/f4SGhkKr1eLSpUvw9/dHmzZtIJVK+ebzih449Xo9du3ahePHj8PX1xcLFy7ku1IAZQPY/jjgiFQPBXSEEEIanOjoaHh6emLcuHGVjtz+q7JMCWIZTVpUVIQvvvgC3t7eWLx4cX1njzRAfya+aXiPS4QQQv4SLl++jHfeeafRTz7LcRw/qMvd3Z2f45KQ2kQBHSGEkJdi7dq18PDw4CexbWyenZqI4zhotVooFAq0adOmnnNG/ooooCOEEPJSWCa+ra+BJ/VJr9dj/vz5ePr0KT9Fy6VLl3D37t3nftOXkNpAAR0hhBBSyyQSCaZPn44ZM2bg0qVLuHnzJlavXs1PXk5IbaNBEYQQQshLYDQakZeXh9WrV+Px48eYMmUKXnvtNUilUqvJmQmxoFGuhBBCCCE2jka5EkIIIYQ0YhTQEUIIIYTYOAroCCGEEEJsHAV0hBBCCCE2jgI6QgghhBAbV+OALi4uDkOGDIGPjw84jsOBAwes1jPG+N+qs7OzQ0hICO7fv2+VpqCgAKGhoXBycoKLiwsmTpwItVr9p3aEEEIIIaSxqnFAV1JSgk6dOmHdunXlrv/mm2+wZs0abNiwAQkJCXBwcMCAAQOg1Wr5NKGhoUhJScGpU6dw+PBhxMXFYfLkyS++F4QQQgghjdifmoeO4zjs378fw4cPB1BWO+fj44NZs2bh888/BwAUFxfD09MTMTExGDNmDO7cuYN27drhypUr6Nq1KwDg+PHjGDx4MDIzM6s1gzbNQ0cIIYSQv5oGMw9dWloasrOzERISwi9zdnZG9+7dER8fDwCIj4+Hi4sLH8wBQEhICAQCARISEsrdrk6ng1KptHoRQgghhJAytRrQZWdnAwA8PT2tlnt6evLrsrOz4eHhYbVeJBLB1dWVT/NHy5Ytg7OzM//y8/OrzWwTQgghhNg0mxjlOn/+fBQXF/OvjIyM+s4SIYQQQkiDUasBnZeXFwBAoVBYLVcoFPw6Ly8v5OTkWK03Go0oKCjg0/yRVCqFk5OT1YsQQgghhJSp1YAuICAAXl5eiI2N5ZcplUokJCQgODgYABAcHIyioiIkJibyaU6fPg2z2Yzu3bvXZnYIIYQQQhoFUU3foFar8eDBA/7vtLQ0JCUlwdXVFf7+/oiIiMBXX32F1q1bIyAgAAsXLoSPjw8/EjYwMBADBw7EpEmTsGHDBhgMBsyYMQNjxoyp1ghXQgghhBBircYB3dWrV/H222/zf0dGRgIAxo0bh5iYGMyZMwclJSWYPHkyioqK0KtXLxw/fhwymYx/z44dOzBjxgz069cPAoEAI0eOxJo1a2phdwghhBBCGp8/NQ9dfaF56AghhBDyV9Ng5qEjhBBCCCF1jwI6QgghhBAbRwEdIYQQQoiNo4COEEIIIcTGUUBHCCGEEGLjKKAjhBBCCLFxFNARQgghhNi4Gk8s3BBYps5TKpX1nBNCCCGEkNphiWteZIpgmwzo8vPzAQB+fn71nBNCCCGEkNqlUqng7Oxco/fYZEDn6uoKAEhPT6/xDpP6pVQq4efnh4yMDPqVDxtDZWe7qOxsF5Wd7XqRsmOMQaVSvdBv29tkQCcQlHX9c3Z2phPcRjk5OVHZ2SgqO9tFZWe7qOxsV03L7kUrqmhQBCGEEEKIjaOAjhBCCCHExtlkQCeVSrFo0SJIpdL6zgqpISo720VlZ7uo7GwXlZ3tquuy49iLjI0lhBBCCCENhk3W0BFCCCGEkP+igI4QQgghxMZRQEcIIYQQYuMooCOEEEIIsXEU0BFCCCGE2DibDOjWrVuHFi1aQCaToXv37rh8+XJ9Z6lRW7ZsGbp16wa5XA4PDw8MHz4cqampVmm0Wi3CwsLQtGlTODo6YuTIkVAoFFZp0tPT8c4778De3h4eHh6YPXs2jEZjXe5Ko7d8+XJwHIeIiAh+GZVdw/XkyRN8+OGHaNq0Kezs7NChQwdcvXqVX88YwxdffAFvb2/Y2dkhJCQE9+/ft9pGQUEBQkND4eTkBBcXF0ycOBFqtbqud6VRMZlMWLhwIQICAmBnZ4eWLVviyy+/tPpBdiq7hiEuLg5DhgyBj48POI7DgQMHrNbXVjndvHkTvXv3hkwmg5+fH7755puaZ5bZmN27dzOJRMK2bNnCUlJS2KRJk5iLiwtTKBT1nbVGa8CAAWzr1q0sOTmZJSUlscGDBzN/f3+mVqv5NFOnTmV+fn4sNjaWXb16lfXo0YP17NmTX280Gln79u1ZSEgIu379Ojt69Chzc3Nj8+fPr49dapQuX77MWrRowTp27MjCw8P55VR2DVNBQQFr3rw5+/jjj1lCQgJ7+PAhO3HiBHvw4AGfZvny5czZ2ZkdOHCA3bhxgw0dOpQFBAQwjUbDpxk4cCDr1KkTu3TpEjt//jxr1aoV+/vf/14fu9RoLF26lDVt2pQdPnyYpaWlsb179zJHR0e2evVqPg2VXcNw9OhRFhUVxfbt28cAsP3791utr41yKi4uZp6eniw0NJQlJyezXbt2MTs7O/bjjz/WKK82F9AFBQWxsLAw/m+TycR8fHzYsmXL6jFX5Fk5OTkMADt37hxjjLGioiImFovZ3r17+TR37txhAFh8fDxjrOxLIxAIWHZ2Np/mhx9+YE5OTkyn09XtDjRCKpWKtW7dmp06dYq9+eabfEBHZddwzZ07l/Xq1avC9WazmXl5ebEVK1bwy4qKiphUKmW7du1ijDF2+/ZtBoBduXKFT3Ps2DHGcRx78uTJy8t8I/fOO++wCRMmWC0bMWIECw0NZYxR2TVUfwzoaquc1q9fz5o0aWJ1vZw7dy5r06ZNjfJnU02uer0eiYmJCAkJ4ZcJBAKEhIQgPj6+HnNGnlVcXAwAcHV1BQAkJibCYDBYlVvbtm3h7+/Pl1t8fDw6dOgAT09PPs2AAQOgVCqRkpJSh7lvnMLCwvDOO+9YlRFAZdeQHTp0CF27dsV7770HDw8PdO7cGZs2beLXp6WlITs726rsnJ2d0b17d6uyc3FxQdeuXfk0ISEhEAgESEhIqLudaWR69uyJ2NhY3Lt3DwBw48YNXLhwAYMGDQJAZWcraquc4uPj0adPH0gkEj7NgAEDkJqaisLCwmrnR/Rnd6gu5eXlwWQyWd04AMDT0xN3796tp1yRZ5nNZkREROCNN95A+/btAQDZ2dmQSCRwcXGxSuvp6Yns7Gw+TXnlallHXp7du3fj2rVruHLlynPrqOwarocPH+KHH35AZGQkFixYgCtXrmDmzJmQSCQYN24cf+zLK5tny87Dw8NqvUgkgqurK5XdSzRv3jwolUq0bdsWQqEQJpMJS5cuRWhoKABQ2dmI2iqn7OxsBAQEPLcNy7omTZpUKz82FdCRhi8sLAzJycm4cOFCfWeFVENGRgbCw8Nx6tQpyGSy+s4OqQGz2YyuXbsiOjoaANC5c2ckJydjw4YNGDduXD3njlRmz5492LFjB3bu3IlXX30VSUlJiIiIgI+PD5UdeWE21eTq5uYGoVD43Ag7hUIBLy+vesoVsZgxYwYOHz6MM2fOoFmzZvxyLy8v6PV6FBUVWaV/tty8vLzKLVfLOvJyJCYmIicnB6+//jpEIhFEIhHOnTuHNWvWQCQSwdPTk8qugfL29ka7du2slgUGBiI9PR3Af499ZddLLy8v5OTkWK03Go0oKCigsnuJZs+ejXnz5mHMmDHo0KEDPvroI3z22WdYtmwZACo7W1Fb5VRb11CbCugkEgm6dOmC2NhYfpnZbEZsbCyCg4PrMWeNG2MMM2bMwP79+3H69Onnqo67dOkCsVhsVW6pqalIT0/nyy04OBi3bt2yOvFPnToFJyen525apPb069cPt27dQlJSEv/q2rUrQkND+f9T2TVMb7zxxnPTA927dw/NmzcHAAQEBMDLy8uq7JRKJRISEqzKrqioCImJiXya06dPw2w2o3v37nWwF41TaWkpBALr269QKITZbAZAZWcraqucgoODERcXB4PBwKc5deoU2rRpU+3mVgC2OW2JVCplMTEx7Pbt22zy5MnMxcXFaoQdqVvTpk1jzs7O7OzZs+zp06f8q7S0lE8zdepU5u/vz06fPs2uXr3KgoODWXBwML/eMvVF//79WVJSEjt+/Dhzd3enqS/qwbOjXBmjsmuoLl++zEQiEVu6dCm7f/8+27FjB7O3t2fbt2/n0yxfvpy5uLiwgwcPsps3b7Jhw4aVO6VC586dWUJCArtw4QJr3bo1TX3xko0bN475+vry05bs27ePubm5sTlz5vBpqOwaBpVKxa5fv86uX7/OALCVK1ey69evs8ePHzPGaqecioqKmKenJ/voo49YcnIy2717N7O3t//rT1vCGGPff/898/f3ZxKJhAUFBbFLly7Vd5YaNQDlvrZu3cqn0Wg0bPr06axJkybM3t6evfvuu+zp06dW23n06BEbNGgQs7OzY25ubmzWrFnMYDDU8d6QPwZ0VHYN16+//srat2/PpFIpa9u2Ldu4caPVerPZzBYuXMg8PT2ZVCpl/fr1Y6mpqVZp8vPz2d///nfm6OjInJyc2Pjx45lKparL3Wh0lEolCw8PZ/7+/kwmk7H/+Z//YVFRUVbTVlDZNQxnzpwp9/42btw4xljtldONGzdYr169mFQqZb6+vmz58uU1zivH2DNTUxNCCCGEEJtjU33oCCGEEELI8yigI4QQQgixcRTQEUIIIYTYOAroCCGEEEJsHAV0hBBCCCE2jgI6QgghhBAbRwEdIYQQQoiNo4COEEIIIcTGUUBHCCGEEGLjKKAjhBBCCLFxFNARQgghhNi4/w+ZMrwwKatwGgAAAABJRU5ErkJggg==",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "testing(test_set, 3, model, two_heads=False)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "venv",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}