bert-finetune-validation-97.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "df4c22d3-c023-4a4e-8fd2-1179a715e88c",
   "metadata": {},
   "source": [
    "# Testing Model over updated weights \n",
    "\n",
    "## Epochs = 3\n",
    "## Accuracy = 97\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "35b84b76-34c7-4ba6-9cbb-c993ebb55b0a",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/atharva/anaconda3/envs/venv/lib/python3.9/site-packages/tqdm/auto.py:21: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html\n",
      "  from .autonotebook import tqdm as notebook_tqdm\n",
      "Some weights of BertForSequenceClassification were not initialized from the model checkpoint at bert-base-uncased and are newly initialized: ['classifier.bias', 'classifier.weight']\n",
      "You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference.\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "BertForSequenceClassification(\n",
       "  (bert): BertModel(\n",
       "    (embeddings): BertEmbeddings(\n",
       "      (word_embeddings): Embedding(30522, 768, padding_idx=0)\n",
       "      (position_embeddings): Embedding(512, 768)\n",
       "      (token_type_embeddings): Embedding(2, 768)\n",
       "      (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n",
       "      (dropout): Dropout(p=0.1, inplace=False)\n",
       "    )\n",
       "    (encoder): BertEncoder(\n",
       "      (layer): ModuleList(\n",
       "        (0-11): 12 x BertLayer(\n",
       "          (attention): BertAttention(\n",
       "            (self): BertSdpaSelfAttention(\n",
       "              (query): Linear(in_features=768, out_features=768, bias=True)\n",
       "              (key): Linear(in_features=768, out_features=768, bias=True)\n",
       "              (value): Linear(in_features=768, out_features=768, bias=True)\n",
       "              (dropout): Dropout(p=0.1, inplace=False)\n",
       "            )\n",
       "            (output): BertSelfOutput(\n",
       "              (dense): Linear(in_features=768, out_features=768, bias=True)\n",
       "              (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n",
       "              (dropout): Dropout(p=0.1, inplace=False)\n",
       "            )\n",
       "          )\n",
       "          (intermediate): BertIntermediate(\n",
       "            (dense): Linear(in_features=768, out_features=3072, bias=True)\n",
       "            (intermediate_act_fn): GELUActivation()\n",
       "          )\n",
       "          (output): BertOutput(\n",
       "            (dense): Linear(in_features=3072, out_features=768, bias=True)\n",
       "            (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n",
       "            (dropout): Dropout(p=0.1, inplace=False)\n",
       "          )\n",
       "        )\n",
       "      )\n",
       "    )\n",
       "    (pooler): BertPooler(\n",
       "      (dense): Linear(in_features=768, out_features=768, bias=True)\n",
       "      (activation): Tanh()\n",
       "    )\n",
       "  )\n",
       "  (dropout): Dropout(p=0.1, inplace=False)\n",
       "  (classifier): Linear(in_features=768, out_features=2, bias=True)\n",
       ")"
      ]
     },
     "execution_count": 1,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import torch\n",
    "from transformers import BertTokenizer, BertForSequenceClassification\n",
    "from safetensors.torch import load_file\n",
    "\n",
    "# Define the path to the best checkpoint\n",
    "checkpoint_path = \"./bert-hate-speech2/checkpoint-20424\"  # Update to actual best checkpoint path\n",
    "\n",
    "# Load tokenizer\n",
    "tokenizer = BertTokenizer.from_pretrained(\"bert-base-uncased\")\n",
    "\n",
    "# Load model architecture\n",
    "model = BertForSequenceClassification.from_pretrained(\"bert-base-uncased\", num_labels=2)\n",
    "\n",
    "# Load fine-tuned weights\n",
    "weights_path = f\"{checkpoint_path}/model.safetensors\"\n",
    "model.load_state_dict(load_file(weights_path))\n",
    "model.eval()  # Set to evaluation mode\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "c8069474-eaa8-4f60-8c65-0f4f45a3c30f",
   "metadata": {},
   "outputs": [],
   "source": [
    "device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\n",
    "model.to(device)  # Move model to GPU\n",
    "def predict_hate_speech(texts, batch_size=32):\n",
    "    \"\"\"Efficiently tokenize and predict labels in batches.\"\"\"\n",
    "    all_preds = []\n",
    "    model.eval()  # Ensure model is in evaluation mode\n",
    "    \n",
    "    for i in range(0, len(texts), batch_size):\n",
    "        batch_texts = texts[i : i + batch_size]\n",
    "        \n",
    "        # Tokenize the batch\n",
    "        inputs = tokenizer(batch_texts, return_tensors=\"pt\", padding=True, truncation=True, max_length=128)\n",
    "        inputs = {key: val.to(device) for key, val in inputs.items()}  # Move to GPU\n",
    "\n",
    "        with torch.no_grad():\n",
    "            outputs = model(**inputs)\n",
    "        \n",
    "        # Get predicted labels\n",
    "        logits = outputs.logits\n",
    "        batch_preds = torch.argmax(logits, dim=-1).cpu().tolist()  # Move to CPU for processing\n",
    "        all_preds.extend(batch_preds)\n",
    "\n",
    "    return all_preds"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "54a0fbd6-ed4a-47e3-8b2a-0b0235912499",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Text:  i ve seen afro americans who look and undoubtedly act like him\n",
      "Prediction: [1, 1]\n",
      "\n",
      "Text: hyper text transfer protocol www occultopedia com f feral children http www long lands ac uk chimp la http www occultopedia com wild peter him aragon again every time i will see filthy comment about georgia without providing any facts i will respond\n",
      "Prediction: [0, 0, 0, 0, 1, 1, 0, 0]\n",
      "\n",
      "Text: i will respond until armenian scum will stop\n",
      "Prediction: [0, 0]\n",
      "\n",
      "Text: the reason large white families were encouraged those days what a slut to build up the population of white germans which were ravaged by the jewish attacks at them during and in hyperinflation\n",
      "Prediction: [1, 1, 0, 1, 1, 0]\n",
      "\n",
      "Text: i see a nation where i can walk out my front door and not feel like a minority in my own country\n",
      "Prediction: [0, 0, 0]\n",
      "\n",
      "Text: where a white woman can go about in public and not fear assault by mud,no i see that as giving up\n",
      "Prediction: [1, 0, 1, 1]\n",
      "\n",
      "Text: i hope one day to proudly say again that i am an american when whatever take this country back\n",
      "Prediction: [0, 1, 0]\n",
      "\n",
      "Text: okay i have been through some other threads on this section about nice and watched a few videos on you there\n",
      "Prediction: [0, 0, 0, 1]\n",
      "\n",
      "Text: it has obviously been misunderstood and i would have been keen to learn more\n",
      "Prediction: [0, 1, 0]\n",
      "\n",
      "Text: i retract always because an animal lacks the skills to properly raise a half human child\n",
      "Prediction: [1, 1, 1]\n",
      "\n",
      "Text: i apologize for my ignorance\n",
      "Prediction: [0]\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Example sentences\n",
    "samples = [\n",
    " \" i ve seen afro americans who look and undoubtedly act like him\",\n",
    "\"hyper text transfer protocol www occultopedia com f feral children http www long lands ac uk chimp la http www occultopedia com wild peter him aragon again every time i will see filthy comment about georgia without providing any facts i will respond\",\n",
    "\"i will respond until armenian scum will stop\",\n",
    "\"the reason large white families were encouraged those days what a slut to build up the population of white germans which were ravaged by the jewish attacks at them during and in hyperinflation\",\n",
    "\"i see a nation where i can walk out my front door and not feel like a minority in my own country\",\n",
    "\"where a white woman can go about in public and not fear assault by mud,\"\n",
    "\"no i see that as giving up\",\n",
    "\"i hope one day to proudly say again that i am an american when whatever take this country back\",\n",
    "\"okay i have been through some other threads on this section about nice and watched a few videos on you there\",\n",
    "\"it has obviously been misunderstood and i would have been keen to learn more\",\n",
    "\"i retract always because an animal lacks the skills to properly raise a half human child\",\n",
    "\"i apologize for my ignorance\"\n",
    "]\n",
    "\n",
    "# Run predictions\n",
    "for text in samples:\n",
    "    result = predict_hate_speech(text)\n",
    "    print(f\"Text: {text}\\nPrediction: {result}\\n\")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "774ee47c-dcfe-4b89-bdc9-8ccc77e8a6de",
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "df = pd.read_csv(\"HateSpeechDatasetBalanced.csv\")\n",
    "val_texts = df[\"Content\"].tolist()\n",
    "val_labels = df[\"Label\"].tolist()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "e8dcd51b-fa0b-4d5e-84be-821a02ad5fd4",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Classification Report:\n",
      "                  precision    recall  f1-score   support\n",
      "\n",
      "Not Hate Speech       0.96      0.98      0.97    361594\n",
      "    Hate Speech       0.98      0.96      0.97    364525\n",
      "\n",
      "       accuracy                           0.97    726119\n",
      "      macro avg       0.97      0.97      0.97    726119\n",
      "   weighted avg       0.97      0.97      0.97    726119\n",
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAhwAAAHWCAYAAAA8ZVAzAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjkuNCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8ekN5oAAAACXBIWXMAAA9hAAAPYQGoP6dpAABxvUlEQVR4nO3de1yO9/8H8Ndd6u6kEqUiFeUQkdMSE8aEGGOb4xRhLEY5xEbIIcvM+bAxwtgchs0pWs7KKXIaOUVDhagUlbqv3x9+XV+3iju77/vaul/PPe7Huj/X5/pc7+sWvfucLpkgCAKIiIiINEhP6gCIiIio/GPCQURERBrHhIOIiIg0jgkHERERaRwTDiIiItI4JhxERESkcUw4iIiISOOYcBAREZHGMeEgIiIijWPCQaSi69evo2PHjrCwsIBMJsOOHTvU2v7t27chk8kQGRmp1nb/y9q2bYu2bdtKHQYRqQETDvpPuXnzJr744gvUrFkTRkZGMDc3R6tWrbBw4UI8f/5co9f28/PDxYsXMWvWLKxfvx7NmjXT6PW0yd/fHzKZDObm5iV+jtevX4dMJoNMJsN3331X5vbv37+PadOmISEhQQ3REtF/UQWpAyBS1e7du/Hpp59CLpdj4MCBaNCgAfLz83Hs2DGMHz8ely9fxo8//qiRaz9//hxxcXH45ptvMHLkSI1cw9HREc+fP4eBgYFG2n+bChUq4NmzZ9i5cyc+++wzpWMbNmyAkZERcnNz36nt+/fvY/r06XBycoKHh4fK5+3fv/+drkdE/z5MOOg/ISkpCX369IGjoyMOHDgAOzs78VhgYCBu3LiB3bt3a+z6Dx8+BABYWlpq7BoymQxGRkYaa/9t5HI5WrVqhV9++aVYwrFx40b4+vrit99+00osz549g4mJCQwNDbVyPSLSPA6p0H9CREQEsrOz8dNPPyklG0VcXFwwevRo8X1BQQFmzJiBWrVqQS6Xw8nJCV9//TXy8vKUznNyckLXrl1x7NgxvPfeezAyMkLNmjWxbt06sc60adPg6OgIABg/fjxkMhmcnJwAvByKKPr6VdOmTYNMJlMqi46Oxvvvvw9LS0uYmZmhTp06+Prrr8Xjpc3hOHDgAFq3bg1TU1NYWlqie/fuuHLlSonXu3HjBvz9/WFpaQkLCwsMGjQIz549K/2DfU2/fv2wd+9eZGRkiGWnT5/G9evX0a9fv2L1Hz9+jHHjxsHd3R1mZmYwNzdH586dcf78ebHOoUOH0Lx5cwDAoEGDxKGZovts27YtGjRogPj4eHh7e8PExET8XF6fw+Hn5wcjI6Ni9+/j44NKlSrh/v37Kt8rEWkXEw76T9i5cydq1qyJli1bqlR/yJAhCA0NRZMmTTB//ny0adMG4eHh6NOnT7G6N27cwCeffIIPP/wQ8+bNQ6VKleDv74/Lly8DAHr27In58+cDAPr27Yv169djwYIFZYr/8uXL6Nq1K/Ly8hAWFoZ58+bho48+wvHjx9943p9//gkfHx88ePAA06ZNQ3BwMGJjY9GqVSvcvn27WP3PPvsMT58+RXh4OD777DNERkZi+vTpKsfZs2dPyGQybNu2TSzbuHEj6tatiyZNmhSrf+vWLezYsQNdu3bF999/j/Hjx+PixYto06aN+MO/Xr16CAsLAwAMGzYM69evx/r16+Ht7S22k56ejs6dO8PDwwMLFixAu3btSoxv4cKFsLa2hp+fHwoLCwEAP/zwA/bv34/FixfD3t5e5XslIi0TiP7lMjMzBQBC9+7dVaqfkJAgABCGDBmiVD5u3DgBgHDgwAGxzNHRUQAgHDlyRCx78OCBIJfLhbFjx4plSUlJAgBh7ty5Sm36+fkJjo6OxWKYOnWq8Opfr/nz5wsAhIcPH5Yad9E11qxZI5Z5eHgINjY2Qnp6ulh2/vx5QU9PTxg4cGCx6w0ePFipzY8//lioXLlyqdd89T5MTU0FQRCETz75RGjfvr0gCIJQWFgo2NraCtOnTy/xM8jNzRUKCwuL3YdcLhfCwsLEstOnTxe7tyJt2rQRAAgrVqwo8VibNm2Uyvbt2ycAEGbOnCncunVLMDMzE3r06PHWeyQiabGHg/71srKyAAAVK1ZUqf6ePXsAAMHBwUrlY8eOBYBicz3c3NzQunVr8b21tTXq1KmDW7duvXPMryua+/H7779DoVCodE5KSgoSEhLg7+8PKysrsbxhw4b48MMPxft81fDhw5Xet27dGunp6eJnqIp+/frh0KFDSE1NxYEDB5CamlricArwct6Hnt7Lf0YKCwuRnp4uDhedPXtW5WvK5XIMGjRIpbodO3bEF198gbCwMPTs2RNGRkb44YcfVL4WEUmDCQf965mbmwMAnj59qlL9O3fuQE9PDy4uLkrltra2sLS0xJ07d5TKa9SoUayNSpUq4cmTJ+8YcXG9e/dGq1atMGTIEFStWhV9+vTB5s2b35h8FMVZp06dYsfq1auHR48eIScnR6n89XupVKkSAJTpXrp06YKKFSti06ZN2LBhA5o3b17ssyyiUCgwf/58uLq6Qi6Xo0qVKrC2tsaFCxeQmZmp8jWrVatWpgmi3333HaysrJCQkIBFixbBxsZG5XOJSBpMOOhfz9zcHPb29rh06VKZznt90mZp9PX1SywXBOGdr1E0v6CIsbExjhw5gj///BOff/45Lly4gN69e+PDDz8sVvef+Cf3UkQul6Nnz55Yu3Yttm/fXmrvBgDMnj0bwcHB8Pb2xs8//4x9+/YhOjoa9evXV7knB3j5+ZTFuXPn8ODBAwDAxYsXy3QuEUmDCQf9J3Tt2hU3b95EXFzcW+s6OjpCoVDg+vXrSuVpaWnIyMgQV5yoQ6VKlZRWdBR5vRcFAPT09NC+fXt8//33+OuvvzBr1iwcOHAABw8eLLHtojgTExOLHbt69SqqVKkCU1PTf3YDpejXrx/OnTuHp0+fljjRtsjWrVvRrl07/PTTT+jTpw86duyIDh06FPtMVE3+VJGTk4NBgwbBzc0Nw4YNQ0REBE6fPq229olIM5hw0H/ChAkTYGpqiiFDhiAtLa3Y8Zs3b2LhwoUAXg4JACi2kuT7778HAPj6+qotrlq1aiEzMxMXLlwQy1JSUrB9+3aleo8fPy52btEGWK8v1S1iZ2cHDw8PrF27VukH+KVLl7B//37xPjWhXbt2mDFjBpYsWQJbW9tS6+nr6xfrPdmyZQvu3bunVFaUGJWUnJVVSEgIkpOTsXbtWnz//fdwcnKCn59fqZ8jEf07cOMv+k+oVasWNm7ciN69e6NevXpKO43GxsZiy5Yt8Pf3BwA0atQIfn5++PHHH5GRkYE2bdrg1KlTWLt2LXr06FHqkst30adPH4SEhODjjz/GV199hWfPnmH58uWoXbu20qTJsLAwHDlyBL6+vnB0dMSDBw+wbNkyVK9eHe+//36p7c+dOxedO3eGl5cXAgIC8Pz5cyxevBgWFhaYNm2a2u7jdXp6epg8efJb63Xt2hVhYWEYNGgQWrZsiYsXL2LDhg2oWbOmUr1atWrB0tISK1asQMWKFWFqagpPT084OzuXKa4DBw5g2bJlmDp1qrhMd82aNWjbti2mTJmCiIiIMrVHRFok8SoZojK5du2aMHToUMHJyUkwNDQUKlasKLRq1UpYvHixkJubK9Z78eKFMH36dMHZ2VkwMDAQHBwchEmTJinVEYSXy2J9fX2LXef15ZilLYsVBEHYv3+/0KBBA8HQ0FCoU6eO8PPPPxdbFhsTEyN0795dsLe3FwwNDQV7e3uhb9++wrVr14pd4/Wlo3/++afQqlUrwdjYWDA3Nxe6desm/PXXX0p1iq73+rLbNWvWCACEpKSkUj9TQVBeFlua0pbFjh07VrCzsxOMjY2FVq1aCXFxcSUuZ/39998FNzc3oUKFCkr32aZNG6F+/folXvPVdrKysgRHR0ehSZMmwosXL5TqBQUFCXp6ekJcXNwb74GIpCMThDLMJiMiIiJ6B5zDQURERBrHhIOIiIg0jgkHERERaRwTDiIiItI4JhxERESkcUw4iIiISOOYcBAREWnZ8uXL0bBhQ5ibm8Pc3BxeXl7Yu3eveLxt27aQyWRKr9efBp2cnAxfX1+YmJjAxsYG48ePR0FBgVKdQ4cOoUmTJpDL5XBxcUFkZGSxWJYuXQonJycYGRnB09MTp06dUjqem5uLwMBAVK5cGWZmZujVq1eJOz6/TbncadS48UipQyDSuCenl0gdApHGGWn4p5Q6f148P6f638nq1atjzpw5cHV1hSAIWLt2Lbp3745z586hfv36AIChQ4ciLCxMPMfExET8urCwEL6+vrC1tUVsbCxSUlIwcOBAGBgYYPbs2QCApKQk+Pr6Yvjw4diwYQNiYmIwZMgQ2NnZwcfHBwCwadMmBAcHY8WKFfD09MSCBQvg4+ODxMRE8SnMQUFB2L17N7Zs2QILCwuMHDkSPXv2xPHjx8v0+ZTLjb+YcJAuYMJBukDjCUeTr9TW1vOzi/7R+VZWVpg7dy4CAgLQtm1beHh4FHsmVJG9e/eia9euuH//PqpWrQoAWLFiBUJCQvDw4UMYGhoiJCQEu3fvVnrSdp8+fZCRkYGoqCgAgKenJ5o3b44lS17+e6JQKODg4IBRo0Zh4sSJyMzMhLW1NTZu3IhPPvkEwMuHR9arVw9xcXFo0aKFyvfHIRUiIiI1yMvLQ1ZWltJLlYcKFhYW4tdff0VOTg68vLzE8g0bNqBKlSpo0KABJk2ahGfPnonH4uLi4O7uLiYbAODj44OsrCxcvnxZrNOhQwela/n4+IhP3c7Pz0d8fLxSHT09PXTo0EGsEx8fjxcvXijVqVu3LmrUqKHS07tfxYSDiIh0l0ymtld4eDgsLCyUXuHh4aVe+uLFizAzM4NcLsfw4cOxfft2uLm5AQD69euHn3/+GQcPHsSkSZOwfv16DBgwQDw3NTVVKdkAIL5PTU19Y52srCw8f/4cjx49QmFhYYl1Xm3D0NAQlpaWpdZRVbmcw0FERKQSmfp+7540aRKCg4OVyuRyean169Spg4SEBGRmZmLr1q3w8/PD4cOH4ebmhmHDhon13N3dYWdnh/bt2+PmzZuoVauW2mLWJiYcREREaiCXy9+YYLzO0NAQLi4uAICmTZvi9OnTWLhwIX744YdidT09PQEAN27cQK1atWBra1tsNUnRyhFbW1vx/6+vJklLS4O5uTmMjY2hr68PfX39Euu82kZ+fj4yMjKUejleraMqDqkQEZHuUuOQyj+lUChKnfORkJAAALCzswMAeHl54eLFi3jw4IFYJzo6Gubm5uKwjJeXF2JiYpTaiY6OFueJGBoaomnTpkp1FAoFYmJixDpNmzaFgYGBUp3ExEQkJycrzTdRBXs4iIhId6lxSKUsJk2ahM6dO6NGjRp4+vQpNm7ciEOHDmHfvn24efMmNm7ciC5duqBy5cq4cOECgoKC4O3tjYYNGwIAOnbsCDc3N3z++eeIiIhAamoqJk+ejMDAQLGXZfjw4ViyZAkmTJiAwYMH48CBA9i8eTN2794txhEcHAw/Pz80a9YM7733HhYsWICcnBwMGjQIAGBhYYGAgAAEBwfDysoK5ubmGDVqFLy8vMq0QgVgwkFERKR1Dx48wMCBA5GSkgILCws0bNgQ+/btw4cffoi///4bf/75p/jD38HBAb169cLkyZPF8/X19bFr1y6MGDECXl5eMDU1hZ+fn9K+Hc7Ozti9ezeCgoKwcOFCVK9eHatWrRL34ACA3r174+HDhwgNDUVqaio8PDwQFRWlNJF0/vz50NPTQ69evZCXlwcfHx8sW7aszPfMfTiI/qO4DwfpAo3vw+E5Xm1tPT85V21tlUfs4SAiIt0l0ZCKLuInTURERBrHHg4iItJdalhdQqphwkFERLqLQypaw0+aiIiINI49HEREpLs4pKI1TDiIiEh3cUhFa/hJExERkcaxh4OIiHQXh1S0hgkHERHpLg6paA0/aSIiItI49nAQEZHuYg+H1jDhICIi3aXHORzawtSOiIiINI49HEREpLs4pKI1TDiIiEh3cVms1jC1IyIiIo1jDwcREekuDqloDRMOIiLSXRxS0RqmdkRERKRx7OEgIiLdxSEVrWHCQUREuotDKlrD1I6IiIg0jj0cRESkuzikojVMOIiISHdxSEVrmNoRERGRxrGHg4iIdBeHVLSGCQcREekuDqloDVM7IiIi0jj2cBARke7ikIrWSP5Jp6Wl4fPPP4e9vT0qVKgAfX19pRcREZHGyPTU96I3kryHw9/fH8nJyZgyZQrs7Owg43gaERFRuSN5wnHs2DEcPXoUHh4eUodCRES6hr/kao3kCYeDgwMEQZA6DCIi0kUcCtEayT/pBQsWYOLEibh9+7bUoRAREZGGSNLDUalSJaW5Gjk5OahVqxZMTExgYGCgVPfx48faDo+IiHQFh1S0RpKEY8GCBVJcloiISBmHVLRGkoTDz89PissSERGRRCSfNLpnzx7o6+vDx8dHqXz//v0oLCxE586dJYqMiIjKPQ6paI3kfUkTJ05EYWFhsXKFQoGJEydKEBEREekKmUymthe9meQJx/Xr1+Hm5lasvG7durhx44YEEREREZG6SZ5wWFhY4NatW8XKb9y4AVNTUwkiIiIiXcEeDu2RPOHo3r07xowZg5s3b4plN27cwNixY/HRRx9JGBkREZV7MjW+6I0kTzgiIiJgamqKunXrwtnZGc7OzqhXrx4qV66M7777TurwiIiISA0kX6ViYWGB2NhYREdH4/z58zA2NkbDhg3h7e0tdWhERFTOcShEeyTv4QBe/oF37NgRo0aNQmBgIJMNIiLSCqnmcCxfvhwNGzaEubk5zM3N4eXlhb1794rHc3NzERgYiMqVK8PMzAy9evVCWlqaUhvJycnw9fWFiYkJbGxsMH78eBQUFCjVOXToEJo0aQK5XA4XFxdERkYWi2Xp0qVwcnKCkZERPD09cerUKaXjqsSiCskTDoVCgRkzZqBatWowMzNDUlISAGDKlCn46aefJI6OiIhI/apXr445c+YgPj4eZ86cwQcffIDu3bvj8uXLAICgoCDs3LkTW7ZsweHDh3H//n307NlTPL+wsBC+vr7Iz89HbGws1q5di8jISISGhop1kpKS4Ovri3bt2iEhIQFjxozBkCFDsG/fPrHOpk2bEBwcjKlTp+Ls2bNo1KgRfHx88ODBA7HO22JRlUyQ+FGtYWFhWLt2LcLCwjB06FBcunQJNWvWxKZNm7BgwQLExcWVuU3jxiM1ECnRv8uT00ukDoFI44w0PPBv3med2trK+nXgPzrfysoKc+fOxSeffAJra2ts3LgRn3zyCQDg6tWrqFevHuLi4tCiRQvs3bsXXbt2xf3791G1alUAwIoVKxASEoKHDx/C0NAQISEh2L17Ny5duiReo0+fPsjIyEBUVBQAwNPTE82bN8eSJS//PVEoFHBwcMCoUaMwceJEZGZmvjUWVUnew7Fu3Tr8+OOP6N+/P/T19cXyRo0a4erVqxJGRkRE5Z06h1Ty8vKQlZWl9MrLy3trDIWFhfj111+Rk5MDLy8vxMfH48WLF+jQoYNYp27duqhRo4b4S3hcXBzc3d3FZAMAfHx8kJWVJfaSxMXFKbVRVKeojfz8fMTHxyvV0dPTQ4cOHcQ6qsSiKskTjnv37sHFxaVYuUKhwIsXLySIiIiIqOzCw8NhYWGh9AoPDy+1/sWLF2FmZga5XI7hw4dj+/btcHNzQ2pqKgwNDWFpaalUv2rVqkhNTQUApKamKiUbRceLjr2pTlZWFp4/f45Hjx6hsLCwxDqvtvG2WFQl+SoVNzc3HD16FI6OjkrlW7duRePGjSWKioiIdIIaF6lMmjQJwcHBSmVyubzU+nXq1EFCQgIyMzOxdetW+Pn54fDhw+oL6F9G8oQjNDQUfn5+uHfvHhQKBbZt24bExESsW7cOu3btkjo8IiIqx9S5LFYul78xwXidoaGh2MPftGlTnD59GgsXLkTv3r2Rn5+PjIwMpZ6FtLQ02NraAgBsbW2LrSYpWjnyap3XV5OkpaXB3NwcxsbG0NfXh76+fol1Xm3jbbGoSvIhle7du2Pnzp34888/YWpqitDQUFy5cgU7d+7Ehx9+KHV4REREWqFQKJCXl4emTZvCwMAAMTEx4rHExEQkJyfDy8sLAODl5YWLFy8qrSaJjo6Gubm5+HwyLy8vpTaK6hS1YWhoiKZNmyrVUSgUiImJEeuoEouqJO/hAIDWrVsjOjpa6jCIiEjHSLXx16RJk9C5c2fUqFEDT58+xcaNG3Ho0CHs27cPFhYWCAgIQHBwMKysrGBubo5Ro0bBy8tLXBXSsWNHuLm54fPPP0dERARSU1MxefJkBAYGir0sw4cPx5IlSzBhwgQMHjwYBw4cwObNm7F7924xjuDgYPj5+aFZs2Z47733sGDBAuTk5GDQoEEAoFIsqvpXJBwZGRnYunUrbt26hXHjxsHKygpnz55F1apVUa1aNanDIyKickqqhOPBgwcYOHAgUlJSYGFhgYYNG2Lfvn1iz/78+fOhp6eHXr16IS8vDz4+Pli2bJl4vr6+Pnbt2oURI0bAy8sLpqam8PPzQ1hYmFjH2dkZu3fvRlBQEBYuXIjq1atj1apV8PHxEev07t0bDx8+RGhoKFJTU+Hh4YGoqCiliaRvi0VVku/DceHCBXTo0AEWFha4ffs2EhMTUbNmTUyePBnJyclYt67sa6S5DwfpAu7DQbpA0/twWH2+UW1tPV7fT21tlUeSz+EIDg6Gv78/rl+/DiMjI7G8S5cuOHLkiISRERFRecfH02uP5EMqp0+fxg8//FCsvFq1amVe40tERFQmzBO0RvIeDrlcjqysrGLl165dg7W1tQQRERERkbpJnnB89NFHCAsLE3cVlclkSE5ORkhICHr16iVxdEREVJ5xSEV7JE845s2bh+zsbNjY2OD58+do06YNXFxcULFiRcyaNUvq8IiIqBxjwqE9ks/hsLCwQHR0NI4dO4YLFy4gOzsbTZo0KfbAGSIiIvrvkjzhKPL+++/j/ffflzoMIiLSIeyZ0B7Jh1QAICYmBl27dkWtWrVQq1YtdO3aFX/++afUYRERUXknU+OL3kjyhGPZsmXo1KkTKlasiNGjR2P06NEwNzdHly5dsHTpUqnDIyIiIjWQfEhl9uzZmD9/PkaO/N/uoF999RVatWqF2bNnIzAwUMLoiIioPOOQivZI3sORkZGBTp06FSvv2LEjMjMzJYiIiIh0BVepaI/kCcdHH32E7du3Fyv//fff0bVrVwkiIiIiInWTfEjFzc0Ns2bNwqFDh+Dl5QUAOHHiBI4fP46xY8di0aJFYt2vvvpKqjCJiKgcYs+E9kj+tFhnZ2eV6slkMty6dUulunxaLOkCPi2WdIGmnxZr/8U2tbV1/4eeamurPJK8hyMpKUnqEIiIiEjDJJ/D8bqCggJkZ2dLHQYREekC7sOhNZIlHDt37kRkZKRS2axZs2BmZgZLS0t07NgRT548kSY4IiLSCVyloj2SJRzff/89cnJyxPexsbEIDQ3FlClTsHnzZvz999+YMWOGVOERERGRGkk2h+Py5cv4/vvvxfdbt27Fhx9+iG+++QYAYGRkhNGjRyvVISIiUif2TGiPZD0cT58+ReXKlcX3x44dQ/v27cX39evXx/3796UIjYiIdASHVLRHsoSjWrVquHLlCgAgOzsb58+fR8uWLcXj6enpMDExkSo8IiIiUiPJhlQ+/fRTjBkzBl9//TX27NkDW1tbtGjRQjx+5swZ1KlTR6rwiIhIF7BjQmskSzhCQ0Nx7949fPXVV7C1tcXPP/8MfX198fgvv/yCbt26SRUeERHpAA6FaI9kCYexsTHWrVtX6vGDBw9qMRoiIiLSJMl3GiUiIpIKezi0hwmHDhj66fsY+klrONpbAQCu3ErF7B/3Yv/xvwAA+1aOhnczV6VzVm49hq9m/QoAsLIwxZpZfnCvXQ1WFiZ4+Dgbuw5dQOiSnXiakwsAsK1ijjnBPdHErQZqOVTBsl8OY/x3v5Ua06c+TbFuziDsPHgenwWvFMu7f9AIQz55H43r1UBlS1N49g7HhWv31Pp5kO4qLCzE8qWLsXvXH0h/9AjWNjb4qPvHGDb8yxJ/8MyYHoqtmzdhfMgkDBjoL5Z3/vAD3L+v/H351ZixCBg6DABwO+kWZkyfilu3biL76VNY29igS5eu+OLLkTAwMNDoPVLZMOHQHiYcOuBeWgamLP4dN5IfQgYZBnTzxJb5w9CizxxcuZUKAPjpt+OYsXyXeM6z3Bfi1wqFArsOX8D0Zbvw6MlT1HSwxoKJn2GxhSn8v44EABgaVMCjJ08xZ1UURvVv98Z4athZITyoB46dvVHsmImxIWITbuK36LNYHtpfDXdP9D9rflqJLZt+wYzZ36KWiwv+unQJoZMnwaxiRfQfMFCpbsyf0bh4/jysbWxKbOvLkV+h1yefie9NTE3FrytUMEC37j1Qr159VDSviGtXr2L6tClQCAK+GhOsmZsj+pdjwqED9hy5pPR+2tKdGPrp+3ivobOYcDzPzUda+tMSz894+hwrtxwT3yenPMGPW44iaGCHV8oeY9zclz0aft29So1FT0+GyNl+mLFiD1o1rgXLisZKx3/ZfRrAy6SESN0SEs6h7Qft4d2mLQCgWrXq2LtnNy5dvKBULy0tDXNmz8DyH3/CqBFflNiWqakpqlhbl3isuoMDqjs4iO/t7avh9OlTOBt/Rj03QmrDHg7t+Vc9vC03N1fqEMo9PT0ZPvVpClNjQ5y88L8n9fbu0gx/H5iDM1u+Rtioj2BsVHq3r521Bbp/4IGj8dfLfP2vh3XGw8fZWLsj7p3iJ/onPDwa49SJE7h9++X3fuLVqzh3Lh7vt/YW6ygUCnwzcTz8BwXAxcW1tKawetVKeLf0xGe9eiBy9SoUFBSUWjf5zh3EHjuKZs2aq+9mSD348DatkbyHQ6FQYNasWVixYgXS0tJw7do11KxZE1OmTIGTkxMCAgLeeH5eXh7y8vKUygRFIWR6+qWcoZvqu9jj0NqxMDKsgOzneeg9diWu/n/vxqa9Z5Cc8hgpDzPh7mqPmaO7o7ajDfqMW6XUxtpwf3Rt0xAmxobYdfgiRoRtLFMMLT1qwr+HFzz7zFHbfRGVxeAhw5CdnY0eXTtDX18fhYWFGDU6CL5dPxLrrPlpJfQrVEC/14ZYXtW3/+eo5+YGCwsLJCScw6IF3+Phw4cYHzJJqd7A/n1w5a/LyM/PR69Pe+PLUaM1dm9E/3aS93DMnDkTkZGRiIiIgKGhoVjeoEEDrFq16g1nvhQeHg4LCwulV0FavCZD/k+6djsNnn3C4T3wO6zccgwrwz5H3Zq2AIDV247jz7gruHzjPn7dewYBU9aje3sPOFevotTGhO9+g1e/b/HJmB9Qs3oVfDu2p8rXNzOR46eZA/HljF+QnpHz9hOINGBf1F7s2b0T4RHz8OuWbZgxew7WrlmNP3ZsBwD8dfkSNqxfhxmzwt/Y1T7QfxCav+eJ2nXq4rPefTF2fAh+3fgz8vPzlepFfDcfv27ZjjkR83D0yCGsXfOTRu+Pyo5bm2uP5D0c69atw48//oj27dtj+PDhYnmjRo1w9erVt54/adIkBAcrT8KyaR2i9jj/614UFOLW348AAOeu/I2m9WsgsG9bjPr/lSivOn3xNgCgloM1ku4+EsvT0p8iLf0prt1Ow5PMHMSsCcaclVFIfZT11uvXrF4FTtWq4LcF/xsP19N7+Rf06emFaPjxDKVrEWnC/HkRGBwwDJ27+AIAXGvXQcr9+/hp1Q/4qMfHOBt/Bo8fp6NTh/9NfC4sLMS8ud9iw/p12Bt9oMR23Rs2QkFBAe7fuwsn55piua2dHQCglosLChWFmDEtFAP9ByttckjSYqKgPZInHPfu3YOLi0uxcoVCgRcvXpRwhjK5XA65XK5UxuGUt9OTySA3LPmPv1Gd6gCA1EeZpZ4v+/9kwdBAtW+hxNtpaPrJLKWyaYFdYWZihHFzt+Ju6hOV2iH6J3Kf54qJbhF9fX0oFAIAoOtH3eHp1VLp+IhhAejarTt6fFx6j17i1SvQ09ODlVXlUusICgEFBQVQKBRMOEgnSZ5wuLm54ejRo3B0dFQq37p1Kxo3bixRVOVL2KiPsO/4Zfyd8gQVTY3Qu3MzeDdzRbcvl8G5ehX07twM+45dRnpGDtxrV0PE2J44Gn8dl66/fFqvz/tusLEyR/zlO8h+lge3WnaYHdQDseduIjnlsXidhrWrAQBMTeSoUskMDWtXQ35BIa7eSkVefgH+upmiFFfG0+cAoFReydwEDraVYGdjAQCo7VQVAJCWnlXqKhoiVbVp2w4rf1wBWzt71HJxwdUrV7B+7Rp0/7gXAMDSshIsLSspnWNQwQBVqlQRey7OJ5zDxQvn0fy9FjA1NcX58+cw99tw+Hb9COYWL79vd+/6AxUqVICrax0YGhri8uWLWLhgHjp26sx9OP5l2MGhPZInHKGhofDz88O9e/egUCiwbds2JCYmYt26ddi1a9fbG6C3srYyw08zBsK2ijkys3Nx6fo9dPtyGQ6cvIrqVS3xgWcdjOzXDqbGhrib9gQ7YhIwZ9U+8fznuS8wuGdLRIzrCblBBdxNy8DvBxLw3epopeuc3PS/CXNN3WqgT5fmuHM/HXV9p6ocq28bd6wM+1x8v/7bwQCAmSv2YNYPe971IyACAEz8ZjKWLlqI2TOm4/HjdFjb2OCTT3vjixGBKrdhaGiIqL17sGLZEuTn56Nater4fKA/PvcbJNbR16+ANT+twp3bSRAEwM7eHn37DVDaPIz+HTikoj0yQRAEqYM4evQowsLCcP78eWRnZ6NJkyYIDQ1Fx44d36k948Yj1Rwh0b/Pk9NLpA6BSOOMNPxrsev4KLW1dX1uJ7W1VR5J3sMBAK1bt0Z0dPTbKxIREakROzi0R/JlsTVr1kR6enqx8oyMDNSsWbOEM4iIiNSDy2K1R/KE4/bt2ygsLCxWnpeXh3v3+NAuIiKi8kCyIZU//vhD/Hrfvn2w+P/Z3cDLde8xMTFwcnKSIDIiItIV7JjQHskSjh49egB42Z3l5+endMzAwABOTk6YN2+eBJEREZGueH1fFtIcyRIOhUIBAHB2dsbp06dRpUqVt5xBRERE/1WSr1JJSkp6eyUiIiIN4JCK9kiecABATk4ODh8+jOTk5GIPP/rqq68kioqIiIjURfKE49y5c+jSpQuePXuGnJwcWFlZ4dGjRzAxMYGNjQ0TDiIi0hguZ9UeyZfFBgUFoVu3bnjy5AmMjY1x4sQJ3LlzB02bNsV3330ndXhERFSOyWTqe5VFeHg4mjdvjooVK8LGxgY9evRAYmKiUp22bdsW2+vj1aeqA0BycjJ8fX3FX9LHjx+PgoICpTqHDh1CkyZNIJfL4eLigsjIyGLxLF26FE5OTjAyMoKnpydOnTqldDw3NxeBgYGoXLkyzMzM0KtXL6SlpZXpniVPOBISEjB27Fjo6elBX18feXl5cHBwQEREBL7++mupwyMiIlK7w4cPIzAwECdOnEB0dDRevHiBjh07IicnR6ne0KFDkZKSIr4iIiLEY4WFhfD19UV+fj5iY2Oxdu1aREZGIjQ0VKyTlJQEX19ftGvXDgkJCRgzZgyGDBmCffv+97ysTZs2ITg4GFOnTsXZs2fRqFEj+Pj44MGDB2KdoKAg7Ny5E1u2bMHhw4dx//599OxZ+hOUSyL5s1Ssra0RGxsLV1dX1K5dG4sXL4aPjw+uXr2Kpk2bFvvwVcFnqZAu4LNUSBdo+lkqDUP/VFtbF8I6vPO5Dx8+hI2NDQ4fPgxvb28AL3s4PDw8sGDBghLP2bt3L7p27Yr79++jatWXT9ZesWIFQkJC8PDhQxgaGiIkJAS7d+/GpUuXxPP69OmDjIwMREW9fI6Mp6cnmjdvjiVLXv6bolAo4ODggFGjRmHixInIzMyEtbU1Nm7ciE8++QQAcPXqVdSrVw9xcXFo0aKFSvcoeQ9H48aNcfr0aQBAmzZtEBoaig0bNmDMmDFo0KCBxNEREVF5ps6tzfPy8pCVlaX0ysvLUymOzMxMAICVlZVS+YYNG1ClShU0aNAAkyZNwrNnz8RjcXFxcHd3F5MNAPDx8UFWVhYuX74s1unQQTkR8vHxQVxcHAAgPz8f8fHxSnX09PTQoUMHsU58fDxevHihVKdu3bqoUaOGWEcVkiccs2fPhp2dHQBg1qxZqFSpEkaMGIGHDx/ixx9/lDg6IiIi1YSHh8PCwkLpFR4e/tbzFAoFxowZg1atWin9ot2vXz/8/PPPOHjwICZNmoT169djwIAB4vHU1FSlZAOA+D41NfWNdbKysvD8+XM8evQIhYWFJdZ5tQ1DQ0NYWlqWWkcVkq9Sadasmfi1jY2N2MVDRESkaepcpDJp0iQEBwcrlcnl8reeFxgYiEuXLuHYsWNK5cOGDRO/dnd3h52dHdq3b4+bN2+iVq1a6glaiyRPOIiIiKSizmWxcrlcpQTjVSNHjsSuXbtw5MgRVK9e/Y11PT09AQA3btxArVq1YGtrW2w1SdHKEVtbW/H/r68mSUtLg7m5OYyNjaGvrw99ff0S67zaRn5+PjIyMpR6OV6towrJEo7GjRur9Ad99uxZLURDRESkPYIgYNSoUdi+fTsOHToEZ2fnt56TkJAAAOI0BC8vL8yaNQsPHjyAjY0NACA6Ohrm5uZwc3MT6+zZs0epnejoaHh5eQEADA0N0bRpU8TExIjPOFMoFIiJicHIkS8XYDRt2hQGBgaIiYlBr169AACJiYlITk4W21GF5A9vA15+8OHh4Rg+fHixCTNERESaItW+X4GBgdi4cSN+//13VKxYUZwLYWFhAWNjY9y8eRMbN25Ely5dULlyZVy4cAFBQUHw9vZGw4YNAQAdO3aEm5sbPv/8c0RERCA1NRWTJ09GYGCg2NMyfPhwLFmyBBMmTMDgwYNx4MABbN68Gbt37xZjCQ4Ohp+fH5o1a4b33nsPCxYsQE5ODgYNGiTGFBAQgODgYFhZWcHc3ByjRo2Cl5eXyitUgH/BstgiFStWxPnz51GzZs1/3BaXxZIu4LJY0gWaXhbbdMZBtbUVP6WdynVL6+Ffs2YN/P398ffff2PAgAG4dOkScnJy4ODggI8//hiTJ0+Gubm5WP/OnTsYMWIEDh06BFNTU/j5+WHOnDmoUOF/H9yhQ4cQFBSEv/76C9WrV8eUKVPg7++vdN0lS5Zg7ty5SE1NhYeHBxYtWiQO4QAvN/4aO3YsfvnlF+Tl5cHHxwfLli0r05AKEw6i/ygmHKQLymvCoYs4aZSIiHQWH6WiPUw4iIhIZ/HhbdojWcKxaNEipfcFBQWIjIxElSpVlMr5tFgiIqL/PskSjvnz5yu9t7W1xfr165XKZDIZEw4iItIYdnBoj2QJR1JSklSXJiIiAsAhFW2S/FkqREREVP5x0igREeksdnBoDxMOIiLSWRxS0R4OqRAREZHGsYeDiIh0Fjs4tEfyHg59fX08ePCgWHl6ejr09fUliIiIiHSFTCZT24veTPKEo7RHueTl5cHQ0FDL0RAREZEmSL7TqEwmw6pVq2BmZiYeKywsxJEjR1C3bl2pwiMiIh3AjgntkXynUUEQsGLFCqXhE0NDQzg5OWHFihVShUdERDqAQyHaI/lOo+3atcO2bdtQqVIlqUIhIiIiDZN8lcrBgwfFr4vmczDjJCIibeDPG+2RfNIoAKxbtw7u7u4wNjaGsbExGjZsWOxBbkREROomk6nvRW8meQ/H999/jylTpmDkyJFo1aoVAODYsWMYPnw4Hj16hKCgIIkjJCIion9K8oRj8eLFWL58OQYOHCiWffTRR6hfvz6mTZvGhIOIiDSGQyraI3nCkZKSgpYtWxYrb9myJVJSUiSIiIiIdAXzDe2RfA6Hi4sLNm/eXKx806ZNcHV1lSAiIiIiUjfJezimT5+O3r1748iRI+IcjuPHjyMmJqbERISIiEhdOKSiPZInHL169cLJkycxf/587NixAwBQr149nDp1Co0bN5Y2OCIiKteYb2iP5AkHADRt2hQ///yz1GEQERGRhvwrEg4iIiIp6LGLQ2skSzj09PTeOnYmk8lQUFCgpYiIiEjXMN/QHskSju3bt5d6LC4uDosWLYJCodBiRERERKQpkiUc3bt3L1aWmJiIiRMnYufOnejfvz/CwsIkiIyIiHQFV6loj+T7cADA/fv3MXToULi7u6OgoAAJCQlYu3YtHB0dpQ6NiIjKMT2Z+l70ZpImHJmZmQgJCYGLiwsuX76MmJgY7Ny5Ew0aNJAyLCIiIlIzyYZUIiIi8O2338LW1ha//PJLiUMsREREmsQhFe2RLOGYOHEijI2N4eLigrVr12Lt2rUl1tu2bZuWIyMiIl3BfEN7JEs4Bg4cyMySiIhIR0iWcERGRkp1aSIiIgCADPzFV1u40ygREeksri7Rnn/FslgiIiIq39jDQUREOotzCbVHpYTjwoULKjfYsGHDdw6GiIhIm5hvaI9KCYeHhwdkMhkEQSjxeNExmUyGwsJCtQZIRERE/30qJRxJSUmajoOIiEjr+Hh67VEp4eAzTYiIqDxivqE977RKZf369WjVqhXs7e1x584dAMCCBQvw+++/qzU4IiIiKh/KnHAsX74cwcHB6NKlCzIyMsQ5G5aWlliwYIG64yMiItIYmUymthe9WZkTjsWLF2PlypX45ptvoK+vL5Y3a9YMFy9eVGtwREREmiSTqe9Fb1bmhCMpKQmNGzcuVi6Xy5GTk6OWoIiIiKh8KXPC4ezsjISEhGLlUVFRqFevnjpiIiIi0go9mUxtr7IIDw9H8+bNUbFiRdjY2KBHjx5ITExUqpObm4vAwEBUrlwZZmZm6NWrF9LS0pTqJCcnw9fXFyYmJrCxscH48eNRUFCgVOfQoUNo0qQJ5HI5XFxcSnyW2dKlS+Hk5AQjIyN4enri1KlTZY7lbcqccAQHByMwMBCbNm2CIAg4deoUZs2ahUmTJmHChAllbY6IiEgyMjW+yuLw4cMIDAzEiRMnEB0djRcvXqBjx45KIwVBQUHYuXMntmzZgsOHD+P+/fvo2bOneLywsBC+vr7Iz89HbGws1q5di8jISISGhop1kpKS4Ovri3bt2iEhIQFjxozBkCFDsG/fPrHOpk2bEBwcjKlTp+Ls2bNo1KgRfHx88ODBA5VjUYVMKG03rzfYsGEDpk2bhps3bwIA7O3tMX36dAQEBJS1KY0wbjxS6hCINO7J6SVSh0CkcUYafgBHn7Xn1NbWr37Fpxuo6uHDh7CxscHhw4fh7e2NzMxMWFtbY+PGjfjkk08AAFevXkW9evUQFxeHFi1aYO/evejatSvu37+PqlWrAgBWrFiBkJAQPHz4EIaGhggJCcHu3btx6dIl8Vp9+vRBRkYGoqKiAACenp5o3rw5lix5+W+KQqGAg4MDRo0ahYkTJ6oUiyreaVls//79cf36dWRnZyM1NRV379791yQbREREqlLnKpW8vDxkZWUpvfLy8lSKIzMzEwBgZWUFAIiPj8eLFy/QoUMHsU7dunVRo0YNxMXFAQDi4uLg7u4uJhsA4OPjg6ysLFy+fFms82obRXWK2sjPz0d8fLxSHT09PXTo0EGso0osqnjnp8U+ePAA8fHxSExMxMOHD9+1GSIiIsnoydT3Cg8Ph4WFhdIrPDz8rTEoFAqMGTMGrVq1QoMGDQAAqampMDQ0hKWlpVLdqlWrIjU1VazzarJRdLzo2JvqZGVl4fnz53j06BEKCwtLrPNqG2+LRRVl7qx6+vQpvvzyS/zyyy9QKBQAAH19ffTu3RtLly6FhYVFWZskIiL6z5s0aRKCg4OVyuRy+VvPCwwMxKVLl3Ds2DFNhfavUOYejiFDhuDkyZPYvXs3MjIykJGRgV27duHMmTP44osvNBEjERGRRqhzSEUul8Pc3Fzp9baEY+TIkdi1axcOHjyI6tWri+W2trbIz89HRkaGUv20tDTY2tqKdV5fKVL0/m11zM3NYWxsjCpVqkBfX7/EOq+28bZYVFHmhGPXrl1YvXo1fHx8xA/Ux8cHK1euxM6dO8vaHBERkWSk2vhLEASMHDkS27dvx4EDB+Ds7Kx0vGnTpjAwMEBMTIxYlpiYiOTkZHh5eQEAvLy8cPHiRaXVJNHR0TA3N4ebm5tY59U2iuoUtWFoaIimTZsq1VEoFIiJiRHrqBKLKso8pFK5cuUSh00sLCxQqVKlsjZHRESkcwIDA7Fx40b8/vvvqFixojgXwsLCAsbGxrCwsEBAQACCg4NhZWUFc3NzjBo1Cl5eXuKqkI4dO8LNzQ2ff/45IiIikJqaismTJyMwMFDsWRk+fDiWLFmCCRMmYPDgwThw4AA2b96M3bt3i7EEBwfDz88PzZo1w3vvvYcFCxYgJycHgwYNEmN6WyyqKHPCMXnyZAQHB2P9+vViV0pqairGjx+PKVOmlLU5IiIiyUj1DJTly5cDANq2batUvmbNGvj7+wMA5s+fDz09PfTq1Qt5eXnw8fHBsmXLxLr6+vrYtWsXRowYAS8vL5iamsLPzw9hYWFiHWdnZ+zevRtBQUFYuHAhqlevjlWrVsHHx0es07t3bzx8+BChoaFITU2Fh4cHoqKilCaSvi0WVai0D0fjxo2V/lCuX7+OvLw81KhRA8DLnc7kcjlcXV1x9uzZMgWgCdyHg3QB9+EgXaDpfTj8f7mgtrYi+zZUW1vlkUp/lD169NBwGERERFSeqZRwTJ06VdNxEBERaR0fK689Gu6sIiIi+vdiuqE9ZU44CgsLMX/+fGzevBnJycnIz89XOv748WO1BUdERETlQ5n34Zg+fTq+//579O7dG5mZmQgODkbPnj2hp6eHadOmaSBEIiIizZDq8fS6qMwJx4YNG7By5UqMHTsWFSpUQN++fbFq1SqEhobixIkTmoiRiIhII6Ta+EsXlTnhSE1Nhbu7OwDAzMxMfMJd165dlTYSISIiIipS5oSjevXqSElJAQDUqlUL+/fvBwCcPn1apYfUEBER/Vuo81kq9GZlTjg+/vhjcT/1UaNGYcqUKXB1dcXAgQMxePBgtQdIRESkKRxS0Z4yr1KZM2eO+HXv3r3h6OiI2NhYuLq6olu3bmoNjoiIiMqHMvdwvK5FixYIDg6Gp6cnZs+erY6YiIiItIKrVLTnHyccRVJSUvjwNiIi+k/hkIr2qC3hICIiIioNtzYnIiKdxdUl2lMuE470k4ulDoFI4yq995XUIRBp3POzizTaPrv5tUflhCM4OPiNxx8+fPiPgyEiIqLySeWE49y5c2+t4+3t/Y+CISIi0iYOqWiPygnHwYMHNRkHERGR1ukx39AaDl8RERGRxpXLSaNERESqYA+H9jDhICIincU5HNrDIRUiIiLSOPZwEBGRzuKQiva8Uw/H0aNHMWDAAHh5eeHevXsAgPXr1+PYsWNqDY6IiEiT+CwV7SlzwvHbb7/Bx8cHxsbGOHfuHPLy8gAAmZmZfFosERERlajMCcfMmTOxYsUKrFy5EgYGBmJ5q1atcPbsWbUGR0REpEl8PL32lHkOR2JiYok7ilpYWCAjI0MdMREREWkFV05oT5k/a1tbW9y4caNY+bFjx1CzZk21BEVERETlS5kTjqFDh2L06NE4efIkZDIZ7t+/jw0bNmDcuHEYMWKEJmIkIiLSCE4a1Z4yD6lMnDgRCoUC7du3x7Nnz+Dt7Q25XI5x48Zh1KhRmoiRiIhIIzj3QnvKnHDIZDJ88803GD9+PG7cuIHs7Gy4ubnBzMxME/ERERFROfDOG38ZGhrCzc1NnbEQERFpFTs4tKfMCUe7du3euPf8gQMH/lFARERE2sKdRrWnzAmHh4eH0vsXL14gISEBly5dgp+fn7riIiIionKkzAnH/PnzSyyfNm0asrOz/3FARERE2sJJo9qjtj1PBgwYgNWrV6urOSIiIo3jsljtUVvCERcXByMjI3U1R0REROVImYdUevbsqfReEASkpKTgzJkzmDJlitoCIyIi0jROGtWeMiccFhYWSu/19PRQp04dhIWFoWPHjmoLjIiISNNkYMahLWVKOAoLCzFo0CC4u7ujUqVKmoqJiIiIypkyzeHQ19dHx44d+VRYIiIqF/Rk6nvRm5V50miDBg1w69YtTcRCRESkVUw4tKfMCcfMmTMxbtw47Nq1CykpKcjKylJ6EREREb1O5TkcYWFhGDt2LLp06QIA+Oijj5S2OBcEATKZDIWFheqPkoiISAPe9KgOUi+VE47p06dj+PDhOHjwoCbjISIi0hoOhWiPykMqgiAAANq0afPGFxEREb3ZkSNH0K1bN9jb20Mmk2HHjh1Kx/39/SGTyZRenTp1Uqrz+PFj9O/fH+bm5rC0tERAQECxR4xcuHABrVu3hpGRERwcHBAREVEsli1btqBu3bowMjKCu7s79uzZo3RcEASEhobCzs4OxsbG6NChA65fv17mey7THA52PRERUXki1dbmOTk5aNSoEZYuXVpqnU6dOiElJUV8/fLLL0rH+/fvj8uXLyM6Ohq7du3CkSNHMGzYMPF4VlYWOnbsCEdHR8THx2Pu3LmYNm0afvzxR7FObGws+vbti4CAAJw7dw49evRAjx49cOnSJbFOREQEFi1ahBUrVuDkyZMwNTWFj48PcnNzy3TPMqGo6+It9PT0YGFh8dak4/Hjx2UKQBOe5at0S0T/aZVbjJY6BCKNe352kUbbX3A0SW1tjWnt/E7nyWQybN++HT169BDL/P39kZGRUazno8iVK1fg5uaG06dPo1mzZgCAqKgodOnSBXfv3oW9vT2WL1+Ob775BqmpqTA0NAQATJw4ETt27MDVq1cBAL1790ZOTg527doltt2iRQt4eHhgxYoVEAQB9vb2GDt2LMaNGwcAyMzMRNWqVREZGYk+ffqofJ9l2vhr+vTpxXYaJSIiIiAvLw95eXlKZXK5HHK5/J3aO3ToEGxsbFCpUiV88MEHmDlzJipXrgzg5fPLLC0txWQDADp06AA9PT2cPHkSH3/8MeLi4uDt7S0mGwDg4+ODb7/9Fk+ePEGlSpUQFxeH4OBgpev6+PiIiU5SUhJSU1PRoUMH8biFhQU8PT0RFxenuYSjT58+sLGxKcspRERE/1rqnDQaHh6O6dOnK5VNnToV06ZNK3NbnTp1Qs+ePeHs7IybN2/i66+/RufOnREXFwd9fX2kpqYW+3lcoUIFWFlZITU1FQCQmpoKZ2flXpeqVauKxypVqoTU1FSx7NU6r7bx6nkl1VGVygkH528QEVF5o84fbZMmTSrWW/CuvRuv9hy4u7ujYcOGqFWrFg4dOoT27dv/ozilUuZVKkRERFScXC6Hubm50utdE47X1axZE1WqVMGNGzcAALa2tnjw4IFSnYKCAjx+/Bi2trZinbS0NKU6Re/fVufV46+eV1IdVamccCgUCg6nEBFRuaIHmdpemnT37l2kp6fDzs4OAODl5YWMjAzEx8eLdQ4cOACFQgFPT0+xzpEjR/DixQuxTnR0NOrUqSM+gNXLywsxMTFK14qOjoaXlxcAwNnZGba2tkp1srKycPLkSbGOqsq8tTkREVF5IdWy2OzsbCQkJCAhIQHAy8mZCQkJSE5ORnZ2NsaPH48TJ07g9u3biImJQffu3eHi4gIfHx8AQL169dCpUycMHToUp06dwvHjxzFy5Ej06dMH9vb2AIB+/frB0NAQAQEBuHz5MjZt2oSFCxcqDfuMHj0aUVFRmDdvHq5evYpp06bhzJkzGDly5P9/PjKMGTMGM2fOxB9//IGLFy9i4MCBsLe3V1pVo4oyTRolIiKif+7MmTNo166d+L4oCfDz88Py5ctx4cIFrF27FhkZGbC3t0fHjh0xY8YMpSGaDRs2YOTIkWjfvj309PTQq1cvLFr0v2XEFhYW2L9/PwIDA9G0aVNUqVIFoaGhSnt1tGzZEhs3bsTkyZPx9ddfw9XVFTt27ECDBg3EOhMmTEBOTg6GDRuGjIwMvP/++4iKioKRkVGZ7lnlfTj+S7gPB+kC7sNBukDT+3CsiLuttraGezmpra3yiD0cRESks/S4AlNrOIeDiIiINI49HEREpLPYwaE9TDiIiEhncUhFezikQkRERBrHHg4iItJZ7ODQHiYcRESks9jNrz38rImIiEjj2MNBREQ6i09C1x4mHEREpLOYbmgPh1SIiIhI49jDQUREOov7cGjPvyLhUCgUuHHjBh48eACFQqF0zNvbW6KoiIiovGO6oT2SJxwnTpxAv379cOfOHbz+4FqZTIbCwkKJIiMiIiJ1kTzhGD58OJo1a4bdu3fDzs6OM4aJiEhr+CNHeyRPOK5fv46tW7fCxcVF6lCIiEjH8Jdc7ZF8lYqnpydu3LghdRhERESkQZL0cFy4cEH8etSoURg7dixSU1Ph7u4OAwMDpboNGzbUdnhERKQjJP+tW4dIknB4eHhAJpMpTRIdPHiw+HXRMU4aJSIiTeKQivZIknAkJSVJcVkiIiKSiCQJh6OjoxSXJSIiUsL+De2RfPgqPDwcq1evLla+evVqfPvttxJEREREukImk6ntRW8mecLxww8/oG7dusXK69evjxUrVkgQEREREamb5PtwpKamws7Orli5tbU1UlJSJIiIiIh0heS/desQyT9rBwcHHD9+vFj58ePHYW9vL0FERESkKzikoj2S93AMHToUY8aMwYsXL/DBBx8AAGJiYjBhwgSMHTtW4uiIiIhIHSRPOMaPH4/09HR8+eWXyM/PBwAYGRkhJCQEkyZNkjg6IiIqz9gvoT0y4fVHtEokOzsbV65cgbGxMVxdXSGXy9+5rWf5/4pbItKoyi1GSx0CkcY9P7tIo+3/fjFVbW11d7dVW1vlkeRzOIqkpqbi8ePHqFWrFuRyebFH1RMREdF/l+QJR3p6Otq3b4/atWujS5cu4sqUgIAAzuEgIiKN0oNMbS96M8kTjqCgIBgYGCA5ORkmJiZiee/evREVFSVhZEREVN7JZOp70ZtJPml0//792LdvH6pXr65U7urqijt37kgUFREREamT5AlHTk6OUs9GkcePH/+jiaNERERvI+NQiNZIPqTSunVrrFu3Tnwvk8mgUCgQERGBdu3aSRgZERGVdxxS0R7JezgiIiLQvn17nDlzBvn5+ZgwYQIuX76Mx48fl7gDKREREf33SN7D0aBBA1y7dg2tWrVC9+7dkZOTg549e+LcuXOoVauW1OEREVE5xlUq2iN5DwcAWFhYYPLkyVKHQUREOoZDIdojeQ8HABw9ehQDBgxAy5Ytce/ePQDA+vXrcezYMYkjIyIiInWQPOH47bff4OPjA2NjY5w9exZ5eXkAgMzMTMyePVvi6IiIqDzjpFHtkTzhmDlzJlasWIGVK1fCwMBALG/VqhXOnj0rYWRERFTeydT4H72Z5AlHYmIivL29i5VbWFggIyND+wERERGR2kmecNja2uLGjRvFyo8dO4aaNWtKEBEREekKPZn6XvRmkiccQ4cOxejRo3Hy5EnIZDLcv38fGzZswLhx4zBixAipwyMionKMQyraI/my2IkTJ0KhUKB9+/Z49uwZvL29IZfLMW7cOIwaNUrq8IiIiEgNZIIgCFIHAQD5+fm4ceMGsrOz4ebmBjMzs3du61n+v+KWiDSqcovRUodApHHPzy7SaPsHE9PV1la7OpXV1lZ5JPmQShFDQ0NUrFgRdnZ2/yjZICIiUpVUQypHjhxBt27dYG9vD5lMhh07digdFwQBoaGhsLOzg7GxMTp06IDr168r1Xn8+DH69+8Pc3NzWFpaIiAgANnZ2Up1Lly4gNatW8PIyAgODg6IiIgoFsuWLVtQt25dGBkZwd3dHXv27ClzLKqQPOEoKCjAlClTYGFhAScnJzg5OYk7j7548ULq8IiIiNQuJycHjRo1wtKlS0s8HhERgUWLFmHFihU4efIkTE1N4ePjg9zcXLFO//79cfnyZURHR2PXrl04cuQIhg0bJh7PyspCx44d4ejoiPj4eMydOxfTpk3Djz/+KNaJjY1F3759ERAQgHPnzqFHjx7o0aMHLl26VKZYVCH5kMqIESOwbds2hIWFwcvLCwAQFxeHadOmoUePHli+fHmZ2+SQCukCDqmQLtD0kMqRa4/V1pano6m4eWURuVwOuVz+xvNkMhm2b9+OHj16AHjZo2Bvb4+xY8di3LhxAF5uhlm1alVERkaiT58+uHLlCtzc3HD69Gk0a9YMABAVFYUuXbrg7t27sLe3x/Lly/HNN98gNTUVhoaGAF7Om9yxYweuXr0KAOjduzdycnKwa9cuMZ4WLVrAw8MDK1asUCkWVUnew7Fx40ZERkbiiy++QMOGDdGwYUN88cUX+Omnn7Bx40apwyMionJMnUMq4eHhsLCwUHqFh4eXOaakpCSkpqaiQ4cOYpmFhQU8PT0RFxcH4OUv5paWlmKyAQAdOnSAnp4eTp48Kdbx9vYWkw0A8PHxQWJiIp48eSLWefU6RXWKrqNKLKqSfJWKXC6Hk5NTsXJnZ2elD4nUK/7MaayL/Al//XUZjx4+xPcLlqBd+/99Q61Ythj79u5BaloqDCoYoJ5bfYz8agzcGzYS62RmZuDb2TNx5PBByPT00L5DR0yY+DVMTEzFNn5YXry70MjYGHGnzgEAhgz6HPFnTher837rNli87Ad13zaVY0M/eR9DP20FR7uXE/eu3ErB7B+jsD/2CgBg34+j4N3MVemclVuP4avZm8X388b3QguPmqhfyw5Xk1LRom/x8e4iNR2q4MTGCShUKGDXZmKJdT7t2ATr5vhj58EL+GzsKrHc1NgQM7/6CN3aNoSVhQlu33+MZb8cxqrfjr/z/ZP0Jk2ahODgYKWyt/VulCQ1NRUAULVqVaXyqlWrisdSU1NhY2OjdLxChQqwsrJSquPs7FysjaJjlSpVQmpq6luv87ZYVCV5wjFy5EjMmDEDa9asEf9g8vLyMGvWLIwcOVLi6Mqv58+fo3btuuj+cS+MHVN8+bGjoxNCvp6C6tUdkJeXi5/Xr8WXXwTg9937YWVlBQD4OmQ8Hj16iOU/rkZBQQGmTvkaM6aFIjxiHgBgoP9gfPKZcnfbF0MGoX79BuL7eQsWK83VyczIQO9PeuDDjj6auG0qx+49yMCURTtxI/khZDJgQLf3sGX+ULToG4Ert17+w/jTtuOYsfx/E+Ke5RafJ7bu9xNo3sARDVztS71WhQp6WDfbD8fP3USLRs4l1qlhZ4XwoB44drb4xobfjv0YbZvXxqDJ63Dn/mN08KqLhRM/RcrDTOw+cqmE1khT1PkMFFWGT3SZ5AnHuXPnEBMTg+rVq6NRo5e/PZ8/fx75+flo3749evbsKdbdtm2bVGGWO++39sb7rYtvKV+ks283pfdjx0/Ejm1bcf1aIjxbeOHWrZuIPX4UP/+6BfXruwMAQiZNxqgvhyFo3ATY2FSFiYmp2NsBAImJV3Hr5g18M2WaWGZhYal0nX1798DIyAgfduz0z2+SdMqe135QT1u6G0M/eR/vuTuJCcfz3BdIS39aahtj5/4GAKhSyeyNCce0L7si8fYDHDyVWGLCoacnQ+SsgZixYg9aNa4Fy4rGSsdbNHTGzztP4Wj8y2Rk9bZYBPRqhWYNHJlwaNm/cbsuW1tbAEBaWhrs7OzE8rS0NHh4eIh1Hjx4oHReQUEBHj9+LJ5va2uLtLQ0pTpF799W59Xjb4tFVZLP4bC0tESvXr3QtWtXODg4wMHBAV27dkXPnj2LjYWRNF68yMe2rZtgVrEiatepCwC4cD4BFSuai8kGAHi28IKenh4uXbxQYjvbf9sCRycnNGnarMTjALBj21b4dOoCYxMT9d4E6RQ9PRk+7dgEpsZynLxwWyzv3bkZ/o6ZjTObJyJsZDcYGxmU3kgp2jR3Rc8OHhgzZ0updb4e1gkPHz/F2t9PlHj8xIUkdG3TAPbWL/9d827mCtca1vjzxNUyx0Plj7OzM2xtbRETEyOWZWVl4eTJk+LiCi8vL2RkZCA+Pl6sc+DAASgUCnh6eop1jhw5otSLHB0djTp16qBSpUpinVevU1Sn6DqqxKIqyXs41qxZ84/Oz8vLKzYruFBmyG4tNThy+CAmjh+L3NznqGJtjRU/rha/SdMfPYRVZSul+hUqVIC5hQUePXpUrK28vDzs3b0LgwKGlnq9Sxcv4MaN65gaNku9N0I6o76LHQ5FBsPIsAKyn+eh99hVuJr0sndjU1Q8klMeI+VhJtxdq2HmVx+htpMN+oz7SeX2rSxMsHJafwyash5Pc0peEtjSoyb8u3vBs++3pbYT/O1vWDq5N27um4EXLwqhEAR8OeMXHD97s2w3TP+YnkTPlc/OzlZ6jlhSUhISEhJgZWWFGjVqYMyYMZg5cyZcXV3h7OyMKVOmwN7eXlzJUq9ePXTq1AlDhw7FihUr8OLFC4wcORJ9+vSBvf3L3rl+/fph+vTpCAgIQEhICC5duoSFCxdi/vz54nVHjx6NNm3aYN68efD19cWvv/6KM2fOiEtnZTLZW2NRleQJx+sOHz6MnJwceHl5iT/c3iQ8PBzTp09XKvt6cqhStz29m+bNPfHr1u3IePIE237bggnjxmD9hs2wqlz23fQOxETj2bMcdPuoR6l1dmzbClfX2mjg3vAfRE267NrtB/Ds+y0szIzxcXsPrAwbgI5DFuFqUipWb4sV612+kYKUR5mI+mEUnKtXQdLd4klySZZN6YtNUfGlJgZmJnL8NONzfDnjF6Rn5JTazpd9vPGeuxN6jfkRySmP8X6TWljw/3M4Dp66Vrabpn9EqiGVM2fOoF27duL7osmmfn5+iIyMxIQJE5CTk4Nhw4YhIyMD77//PqKiomBkZCSes2HDBowcORLt27eHnp4eevXqhUWL/reM2MLCAvv370dgYCCaNm2KKlWqIDQ0VGmvjpYtW2Ljxo2YPHkyvv76a7i6umLHjh1o0OB/c+1UiUUVku3D8e233yI7OxszZswA8HLdcefOnbF//34AgI2NDWJiYlC/fv03tsMejn+usXvdYqtUSvKRrw+6f9wTAUO+wI7tv+H7ud/iSOwp8XhBQQFaNGuEiHkL8EH7D5XO/WKIP0xNzfD9wiUltv382TN0bO+NEYFfod+Agf/8pnQA9+F4u93LA3Hr7iOMmrWp2DETI0Okx36HboHL8Gec8lDGN190Rre27sVWqaQcngMz4//92yKTyaCvr4eCgkIEztqEhCt/4+SvISgoKBTr6P3/Y0QVCgENe85CysNMpB35Fr3HrkLUsb/Eesum9EW1qpboPrLsew+VZ5reh+PEjQy1tdXCxVJtbZVHkvVwbNq0CSEhIeL7rVu34siRIzh69Cjq1auHgQMHYvr06di8efMbWil5VjA3/tIMQaHAi/x8AEDDRh54+jQLf12+BLf/X3Vy+tQJKBSKYj0U9+7exelTJ7Fg8bJS247eH4X8/Hx06dqt1DpEZaWnJ4PcoOR/5hrVqQYASH2UpXJ7bf3nQ1/vf1PfurZ1x1i/Dmg3aD7uP8jA87wXaPqp8r4L0770hZmpHOPmbsPd1CcwkhvA0KACFArlf6cKFQrJuvd1Gj9yrZEs4UhKSkLDhv/7wbRnzx588sknaNWqFQBg8uTJ+PTTT6UKr9x79iwHfycni+/v3buLxKtXYG5hAUsLS6xauQJt2n6AKtbWyHjyBJt/3YgHD9LE1SM1a9ZCy1atMWP6y+GrgoICzJk9Az6dusDGRnm99o7tv6GKtTVavV/6qpgd239D2w86wNLy7cNoRCUJG9kN+2L/wt8pT1DRVI7enZrBu6kLugUuh3P1KujdqSn2Hf8L6Rk5cHe1R8TYnjgafwOXrt8X26jpUAVmxnJUrVwRxnIDNKz9Mim5cisVLwoKkZikPJu/iZsDFIICf91MEcte/RoAMp4+Vyp/UVCII2euY/aY7nie9wLJKY/RuqkL+vs2R8j3OzTx0dAb8LHy2iNZwlFQUKDUMxEXF4cxY8aI7+3t7UucfEjq8dflSxg62E98P2/uHABAt4964JvQ6bidlISdf3yFjCdPYGFpifr13bF67QbUcvnfxkmzv52LObNm4Ish/tAr2vhr0jdK11EoFNj5+3Z81P1j6OvrlxjL7aRbOHc2Hst/UH3yHtHrrK3M8FPYANhWsUBm9nNcun4f3QKX48DJRFSvaokPPOtgZL+2MDU2xN20J9hxIAFzVu1XamP5lL5Km4Od/PVlL2wd32lITlHfFtgDJ0UibFQ3RM4aiErmJkhOeYJpS3dj5dZjarsG0b+NZHM4PDw8MGbMGPj7+yM5ORlOTk64dOkS3NzcALx8oMxnn32Gu3fvlrltDqmQLuAcDtIFmp7DcepWptraeq8mt294E8l6OAIDAzFy5EgcPXoUJ06cgJeXl5hsAC/XEzdu3Fiq8IiISAdwQEV7JEs4hg4dCn19fezcuRPe3t6YOnWq0vH79+9j8ODBEkVHRERE6iT54+k1gUMqpAs4pEK6QNNDKqeT1Dek0tyZQypv8q/b+IuIiEhbuEpFeyR/lgoRERGVf+zhICIincW91rSHPRxERESkcf+ahOPGjRvYt28fnj9/uStfOZzLSkRE/zIyNb7ozSRPONLT09GhQwfUrl0bXbp0QUrKy+1/AwICMHbsWImjIyKico0Zh9ZInnAEBQWhQoUKSE5OhomJiVjeu3dvREVFSRgZERERqYvkk0b379+Pffv2oXr16krlrq6uuHPnjkRRERGRLuCyWO2RPOHIyclR6tko8vjx42KPnSciIlInrlLRHsmHVFq3bo1169aJ72UyGRQKBSIiItCuXTsJIyMiIiJ1kbyHIyIiAu3bt8eZM2eQn5+PCRMm4PLly3j8+DGOHz8udXhERFSOsYNDeyTv4WjQoAGuXbuG999/H927d0dOTg569uyJc+fOoVatWlKHR0RE5RlXqWiN5D0cycnJcHBwwDfffFPisRo1akgQFREREamT5D0czs7OePjwYbHy9PR0ODs7SxARERHpCpka/6M3k7yHQxAEyEqYJpydnQ0jIyMJIiIiIl3BVSraI1nCERwcDODlqpQpU6YoLY0tLCzEyZMn4eHhIVF0REREpE6SJRznzp0D8LKH4+LFizA0NBSPGRoaolGjRhg3bpxU4RERkQ5gB4f2SJZwHDx4EAAwaNAgLFy4EObm5lKFQkREuooZh9ZIPodjzZo1UodAREREGiZ5wgEAZ86cwebNm5GcnIz8/HylY9u2bZMoKiIiKu+4ukR7JF8W++uvv6Jly5a4cuUKtm/fjhcvXuDy5cs4cOAALCwspA6PiIjKMZlMfS96M8kTjtmzZ2P+/PnYuXMnDA0NsXDhQly9ehWfffYZN/0iIiIqJyRPOG7evAlfX18AL1en5OTkQCaTISgoCD/++KPE0RERUXnGnc21R/KEo1KlSnj69CkAoFq1arh06RIAICMjA8+ePZMyNCIiKu+YcWiN5JNGvb29ER0dDXd3d3z66acYPXo0Dhw4gOjoaLRv317q8IiIiEgNJE84lixZgtzcXADAN998AwMDA8TGxqJXr16YPHmyxNEREVF5xlUq2iN5wmFlZSV+raenh4kTJ0oYDRER6RKuLtEeyRKOrKwslepxB1IiIqL/PskSDktLyxKfEluk6CmyhYWFWoyKiIh0CTs4tEfyZ6kAL5OLLl26YNWqVahWrZpUIRERka5hxqE1kiUcbdq0UXqvr6+PFi1aoGbNmhJFRERERJoi+aRRIiIiqXCVivYw4SAiIp3FVSraI/lOo6960yRSIiIi+u+SrIejZ8+eSu9zc3MxfPhwmJqaKpXz8fRERKQp/DVXeyRLOF5/9PyAAQMkioSIiHQWMw6tkSzhWLNmjVSXJiIiIi37V83hICIi0iaZGv8ri2nTpkEmkym96tatKx7Pzc1FYGAgKleuDDMzM/Tq1QtpaWlKbSQnJ8PX1xcmJiawsbHB+PHjUVBQoFTn0KFDaNKkCeRyOVxcXBAZGVkslqVLl8LJyQlGRkbw9PTEqVOnynQvqmLCQUREOksmU9+rrOrXr4+UlBTxdezYMfFYUFAQdu7ciS1btuDw4cO4f/++0tzHwsJC+Pr6Ij8/H7GxsVi7di0iIyMRGhoq1klKSoKvry/atWuHhIQEjBkzBkOGDMG+ffvEOps2bUJwcDCmTp2Ks2fPolGjRvDx8cGDBw/e7QN9A5kgCILaW5XYs/xyd0tExVRuMVrqEIg07vnZRRptP+lRrtracq5ipHLdadOmYceOHUhISCh2LDMzE9bW1ti4cSM++eQTAMDVq1dRr149xMXFoUWLFti7dy+6du2K+/fvo2rVqgCAFStWICQkBA8fPoShoSFCQkKwe/duXLp0SWy7T58+yMjIQFRUFADA09MTzZs3x5IlSwAACoUCDg4OGDVqlNofpsoeDiIi0lkyNb7y8vKQlZWl9MrLyyv12tevX4e9vT1q1qyJ/v37Izk5GQAQHx+PFy9eoEOHDmLdunXrokaNGoiLiwMAxMXFwd3dXUw2AMDHxwdZWVm4fPmyWOfVNorqFLWRn5+P+Ph4pTp6enro0KGDWEedmHAQEZHuUmPGER4eDgsLC6VXeHh4iZf19PREZGQkoqKisHz5ciQlJaF169Z4+vQpUlNTYWhoCEtLS6VzqlatitTUVABAamqqUrJRdLzo2JvqZGVl4fnz53j06BEKCwtLrFPUhjpxp1EiIiI1mDRpEoKDg5XK5HJ5iXU7d+4sft2wYUN4enrC0dERmzdvhrGxsUbjlAp7OIiISGepc5WKXC6Hubm50qu0hON1lpaWqF27Nm7cuAFbW1vk5+cjIyNDqU5aWhpsbW0BALa2tsVWrRS9f1sdc3NzGBsbo0qVKtDX1y+xTlEb6sSEg4iIdJaUq1RelZ2djZs3b8LOzg5NmzaFgYEBYmJixOOJiYlITk6Gl5cXAMDLywsXL15UWk0SHR0Nc3NzuLm5iXVebaOoTlEbhoaGaNq0qVIdhUKBmJgYsY46MeEgIiLSsnHjxuHw4cO4ffs2YmNj8fHHH0NfXx99+/aFhYUFAgICEBwcjIMHDyI+Ph6DBg2Cl5cXWrRoAQDo2LEj3Nzc8Pnnn+P8+fPYt28fJk+ejMDAQLFXZfjw4bh16xYmTJiAq1evYtmyZdi8eTOCgoLEOIKDg7Fy5UqsXbsWV65cwYgRI5CTk4NBgwap/Z45h4OIiHSWVDub3717F3379kV6ejqsra3x/vvv48SJE7C2tgYAzJ8/H3p6eujVqxfy8vLg4+ODZcuWiefr6+tj165dGDFiBLy8vGBqago/Pz+EhYWJdZydnbF7924EBQVh4cKFqF69OlatWgUfHx+xTu/evfHw4UOEhoYiNTUVHh4eiIqKKjaRVB24DwfRfxT34SBdoOl9OO4+KX3ZallVr6TafA1dxSEVIiIi0jgOqRARkQ7j42K1hQkHERHprH+6uoRUxyEVIiIi0jj2cBARkc5iB4f2MOEgIiKdxSEV7eGQChEREWkceziIiEhnyTioojVMOIiISHcx39AaDqkQERGRxrGHg4iIdBY7OLSHCQcREeksrlLRHg6pEBERkcaxh4OIiHQWV6loDxMOIiLSXcw3tIZDKkRERKRx7OEgIiKdxQ4O7WHCQUREOourVLSHQypERESkcezhICIincVVKtrDhIOIiHQWh1S0h0MqREREpHFMOIiIiEjjOKRCREQ6i0Mq2sMeDiIiItI49nAQEZHO4ioV7WHCQUREOotDKtrDIRUiIiLSOPZwEBGRzmIHh/Yw4SAiIt3FjENrOKRCREREGsceDiIi0llcpaI9TDiIiEhncZWK9nBIhYiIiDSOPRxERKSz2MGhPUw4iIhIdzHj0BoOqRAREZHGsYeDiIh0FlepaA8TDiIi0llcpaI9HFIhIiIijZMJgiBIHQT9t+Xl5SE8PByTJk2CXC6XOhwijeD3OdE/w4SD/rGsrCxYWFggMzMT5ubmUodDpBH8Pif6ZzikQkRERBrHhIOIiIg0jgkHERERaRwTDvrH5HI5pk6dyol0VK7x+5zon+GkUSIiItI49nAQERGRxjHhICIiIo1jwkFEREQax4SD6DWRkZGwtLSUOgyit7p9+zZkMhkSEhKkDoXorZhwaIG/vz9kMhnmzJmjVL5jxw7IyvjkICcnJyxYsOCd602bNg0eHh5luqZMJsOOHTvKdE5Jnj17hkmTJqFWrVowMjKCtbU12rRpg99///0ft03/Xf7+/ujRo0ex8kOHDkEmkyEjI0Plttq2bYsxY8aoJa6VK1eiUaNGMDMzg6WlJRo3bozw8HC1tE2ki/i0WC0xMjLCt99+iy+++AKVKlWSOhxJDB8+HCdPnsTixYvh5uaG9PR0xMbGIj09XerQiJSsXr0aY8aMwaJFi9CmTRvk5eXhwoULuHTpktShEf1nsYdDSzp06ABbW9u3/ob022+/oX79+pDL5XBycsK8efPEY23btsWdO3cQFBQEmUxW5t6Rkpw+fRoffvghqlSpAgsLC7Rp0wZnz54Vjzs5OQEAPv74Y8hkMvE9APz+++9o0qQJjIyMULNmTUyfPh0FBQWlXuuPP/7A119/jS5dusDJyQlNmzbFqFGjMHjwYKXrzZgxA3379oWpqSmqVauGpUuXKrWTkZGBIUOGwNraGubm5vjggw9w/vx5pTpviy0jIwNffPEFqlatCiMjIzRo0AC7du1SamPfvn2oV68ezMzM0KlTJ6SkpKj8uZL6paeno2/fvqhWrRpMTEzg7u6OX375RTzu7++Pw4cPY+HCheLfj9u3bwMALl26hM6dO8PMzAxVq1bF559/jkePHpV6rT/++AOfffYZAgIC4OLigvr166Nv376YNWuW0vV69OiB6dOni9+Lw4cPR35+vlhHoVAgPDwczs7OMDY2RqNGjbB161ala70tNoVCgYiICLi4uEAul6NGjRpKcQDArVu30K5dO5iYmKBRo0aIi4t7p8+YSJOYcGiJvr4+Zs+ejcWLF+Pu3bsl1omPj8dnn32GPn364OLFi5g2bRqmTJmCyMhIAMC2bdtQvXp1hIWFISUlRS0/AJ8+fQo/Pz8cO3YMJ06cgKurK7p06YKnT58CeJmQAMCaNWuQkpIivj969CgGDhyI0aNH46+//sIPP/yAyMjIYv8QvsrW1hZ79uwR2y7N3Llz0ahRI5w7dw4TJ07E6NGjER0dLR7/9NNP8eDBA+zduxfx8fFo0qQJ2rdvj8ePH6sUm0KhQOfOnXH8+HH8/PPP+OuvvzBnzhzo6+uL13j27Bm+++47rF+/HkeOHEFycjLGjRv3Dp8wqUtubi6aNm2K3bt349KlSxg2bBg+//xznDp1CgCwcOFCeHl5YejQoeLfDwcHB2RkZOCDDz5A48aNcebMGURFRSEtLQ2fffZZqdeytbXFiRMncOfOnTfGFBMTgytXruDQoUP45ZdfsG3bNkyfPl08Hh4ejnXr1mHFihW4fPkygoKCMGDAABw+fBgAVIpt0qRJmDNnDqZMmYK//voLGzduRNWqVZXi+OabbzBu3DgkJCSgdu3a6Nu37xuTfyJJCKRxfn5+Qvfu3QVBEIQWLVoIgwcPFgRBELZv3y68+kfQr18/4cMPP1Q6d/z48YKbm5v43tHRUZg/f/5br+no6CgYGhoKpqamSi8DAwOhUaNGpZ5XWFgoVKxYUdi5c6dYBkDYvn27Ur327dsLs2fPVipbv369YGdnV2rbhw8fFqpXry4YGBgIzZo1E8aMGSMcO3asWNydOnVSKuvdu7fQuXNnQRAE4ejRo4K5ubmQm5urVKdWrVrCDz/8oFJs+/btE/T09ITExMQS41yzZo0AQLhx44ZYtnTpUqFq1aql3hu9Oz8/P0FfX7/Y96qRkZEAQHjy5Emp5/r6+gpjx44V37dp00YYPXq0Up0ZM2YIHTt2VCr7+++/BQClfg/cv39faNGihQBAqF27tuDn5yds2rRJKCwsVIrbyspKyMnJEcuWL18umJmZCYWFhUJubq5gYmIixMbGKrUdEBAg9O3bV6XYsrKyBLlcLqxcubLEOJOSkgQAwqpVq8Syy5cvCwCEK1eulHgOkVQ4h0PLvv32W3zwwQcl/rZ85coVdO/eXamsVatWWLBgAQoLC5V+A1fF+PHj4e/vr1S2aNEiHDlyRHyflpaGyZMn49ChQ3jw4AEKCwvx7NkzJCcnv7Ht8+fP4/jx40o9GoWFhcjNzcWzZ89gYmJS7Bxvb2/cunULJ06cQGxsLGJiYrBw4UJMnz4dU6ZMEet5eXkpnefl5SVOgD1//jyys7NRuXJlpTrPnz/HzZs3VYotISEB1atXR+3atUu9PxMTE9SqVUt8b2dnhwcPHrzxM6F3165dOyxfvlyp7OTJkxgwYID4vrCwELNnz8bmzZtx79495OfnIy8vr8TvtVedP38eBw8ehJmZWbFjN2/eLPH7wM7ODnFxcbh06RKOHDmC2NhY+Pn5YdWqVYiKioKe3svO4UaNGild38vLC9nZ2fj777+RnZ2NZ8+e4cMPP1RqOz8/H40bN1YptoyMDOTl5aF9+/ZvvMeGDRsqxQ4ADx48QN26dd94HpE2MeHQMm9vb/j4+GDSpEnFkgF1q1KlClxcXJTKrKyslN77+fkhPT0dCxcuhKOjI+RyOby8vJTGoUuSnZ2N6dOno2fPnsWOGRkZlXqegYEBWrdujdatWyMkJAQzZ85EWFgYQkJCYGho+NZ7ys7Ohp2dHQ4dOlTsWNFS1rfFZmxs/NbrGBgYKL2XyWQQ+BQAjTE1NS32vfr60OPcuXOxcOFCLFiwAO7u7jA1NcWYMWNU+l7t1q0bvv3222LHin44l6ZBgwZo0KABvvzySwwfPhytW7fG4cOH0a5du7feU3Z2NgBg9+7dqFatmtKxouexvC22W7duvfU6gPL3a9HcLoVCodK5RNrChEMCc+bMgYeHB+rUqaNUXq9ePRw/flyp7Pjx46hdu7bYu2FoaIjCwkK1xXL8+HEsW7YMXbp0AQD8/fffxSbTGRgYFLtmkyZNkJiYWOyHRFm5ubmhoKAAubm5YsJx4sQJpTonTpxAvXr1xOumpqaiQoUKShNYyxJbw4YNcffuXVy7du2NvRz073L8+HF0795d7PVQKBS4du0a3NzcxDol/f1o0qQJfvvtNzg5OaFChXf/J6/oOjk5OWLZ+fPn8fz5czGJPXHiBMzMzODg4AArKyvI5XIkJyejTZs2Jbb5tthcXV1hbGyMmJgYDBky5J1jJ/o34KRRCbi7u6N///5YtGiRUvnYsWMRExODGTNm4Nq1a1i7di2WLFmiNPzi5OSEI0eO4N69e2+cZa8qV1dXrF+/HleuXMHJkyfRv3//Yj0ATk5OiImJQWpqKp48eQIACA0Nxbp16zB9+nRcvnwZV65cwa+//orJkyeXeq22bdvihx9+QHx8PG7fvo09e/bg66+/Rrt27WBubi7WO378OCIiInDt2jUsXboUW7ZswejRowG8XO3j5eWFHj16YP/+/bh9+zZiY2PxzTff4MyZMyrF1qZNG3h7e6NXr16Ijo5GUlIS9u7di6ioqH/8eZLmuLq6Ijo6GrGxsbhy5Qq++OILpKWlKdVxcnLCyZMncfv2bTx69AgKhQKBgYF4/Pgx+vbti9OnT+PmzZvYt28fBg0aVGryPmLECMyYMQPHjx/HnTt3cOLECQwcOBDW1tZKQ375+fkICAjAX3/9hT179mDq1KkYOXIk9PT0ULFiRYwbNw5BQUFYu3Ytbt68ibNnz2Lx4sVYu3YtALw1NiMjI4SEhGDChAlYt24dbt68iRMnTuCnn37S3AdNpClSTyLRBa9OGi2SlJQkGBoaCq//EWzdulVwc3MTDAwMhBo1aghz585VOh4XFyc0bNhQkMvlxc59VWmTS6dOnao0afTs2bNCs2bNBCMjI8HV1VXYsmVLsXP/+OMPwcXFRahQoYLg6OgolkdFRQktW7YUjI2NBXNzc+G9994Tfvzxx1Jjmj17tuDl5SVYWVkJRkZGQs2aNYWvvvpKePTokVLc06dPFz799FPBxMREsLW1FRYuXKjUTlZWljBq1CjB3t5eMDAwEBwcHIT+/fsLycnJKseWnp4uDBo0SKhcubJgZGQkNGjQQNi1a5cgCC8njVpYWChd8/UJvqQ+Jf39EARBOHjwoNKk0fT0dKF79+6CmZmZYGNjI0yePFkYOHCg0rmJiYlCixYtBGNjYwGAkJSUJAiCIFy7dk34+OOPBUtLS8HY2FioW7euMGbMGEGhUJQY09atW4UuXboIdnZ2gqGhoWBvby/06tVLuHDhQrG4Q0NDhcqVKwtmZmbC0KFDlSY0KxQKYcGCBUKdOnUEAwMDwdraWvDx8REOHz4s1nlbbIWFhcLMmTMFR0dH8d+FoknRRZNGz507J7b35MkTAYBw8ODBMvwpEGkeH09P/ypOTk4YM2aM2naLJNIUf39/ZGRkqGUXXiJdwCEVIiIi0jgmHERERKRxHFIhIiIijWMPBxEREWkcEw4iIiLSOCYcREREpHFMOIiIiEjjmHAQERGRxjHhINIAf39/9OjRQ3zftm1bSTYzO3ToEGQyGTIyMjR2jdfv9V1oI04ikhYTDtIZ/v7+kMlkkMlkMDQ0hIuLC8LCwlBQUKDxa2/btg0zZsxQqa62f/g6OTlhwYIFWrkWEekuPi2WdEqnTp2wZs0a5OXlYc+ePQgMDISBgQEmTZpUrG5+fr74BNt/ysrKSi3tEBH9V7GHg3SKXC6Hra0tHB0dMWLECHTo0AF//PEHgP8NDcyaNQv29vaoU6cOAODvv//GZ599BktLS1hZWaF79+64ffu22GZhYSGCg4NhaWmJypUrY8KECXh9P73Xh1Ty8vIQEhICBwcHyOVyuLi44KeffsLt27fRrl07AEClSpUgk8ng7+8P4OXj2MPDw+Hs7AxjY2M0atQIW7duVbrOnj17ULt2bRgbG6Ndu3ZKcb6LwsJCBAQEiNesU6cOFi5cWGLd6dOnw9raGubm5hg+fDjy8/PFY6rETkTlG3s4SKcZGxsjPT1dfB8TEwNzc3NER0cDAF68eAEfHx94eXnh6NGjqFChAmbOnIlOnTrhwoULMDQ0xLx58xAZGYnVq1ejXr16mDdvHrZv344PPvig1OsOHDgQcXFxWLRoERo1aoSkpCQ8evQIDg4O+O2339CrVy8kJibC3NwcxsbGAIDw8HD8/PPPWLFiBVxdXXHkyBEMGDAA1tbWaNOmDf7++2/07NkTgYGBGDZsGM6cOYOxY8f+o89HoVCgevXq2LJlCypXrozY2FgMGzYMdnZ2+Oyzz5Q+NyMjIxw6dAi3b9/GoEGDULlyZcyaNUul2IlIB0j6rFoiLXr1MegKhUKIjo4W5HK5MG7cOPF41apVhby8PPGc9evXC3Xq1FF6jHleXp5gbGws7Nu3TxAEQbCzsxMiIiLE4y9evBCqV6+u9Nj0Nm3aCKNHjxYE4eUj1AEI0dHRJcb5+mPZBUEQcnNzBRMTEyE2NlapbkBAgNC3b19BEARh0qRJgpubm9LxkJCQYm29ztHRUZg/f36px18XGBgo9OrVS3zv5+cnWFlZCTk5OWLZ8uXLBTMzM6GwsFCl2Eu6ZyIqX9jDQTpl165dMDMzw4sXL6BQKNCvXz9MmzZNPO7u7q40b+P8+fO4ceMGKlasqNRObm4ubt68iczMTKSkpMDT01M8VqFCBTRr1qzYsEqRhIQE6Ovrl+k3+xs3buDZs2f48MMPlcrz8/PRuHFjAMCVK1eU4gAALy8vla9RmqVLl2L16tVITk7G8+fPkZ+fDw8PD6U6jRo1gomJidJ1s7Oz8ffffyM7O/utsRNR+ceEg3RKu3btsHz5chgaGsLe3h4VKij/FTA1NVV6n52djaZNm2LDhg3F2rK2tn6nGIqGSMoiOzsbALB7925Uq1ZN6ZhcLn+nOFTx66+/Yty4cZg3bx68vLxQsWJFzJ07FydPnlS5DaliJ6J/FyYcpFNMTU3h4uKicv0mTZpg06ZNsLGxgbm5eYl17OzscPLkSXh7ewMACgoKEB8fjyZNmpRY393dHQqFAocPH0aHDh2KHS/qYSksLBTL3NzcIJfLkZycXGrPSL169cQJsEVOnDjx9pt8g+PHj6Nly5b48ssvxbKbN28Wq3f+/Hk8f/5cTKZOnDgBMzMzODg4wMrK6q2xE1H5x1UqRG/Qv39/VKlSBd27d8fRo0eRlJSEQ4cO4auvvsLdu3cBAKNHj8acOXOwY8cOXL16FV9++eUb99BwcnKCn58fBg8ejB07dohtbt68GQDg6OgImUyGXbt24eHDh8jOzkbFihUxbtw4BAUFYe3atbh58ybOnj2LxYsXY+3atQCA4cOH4/r16xg/fjwSExOxceNGREZGqnSf9+7dQ0JCgtLryZMncHV1xZkzZ7Bv3z5cu3YNU6ZMwenTp4udn5+fj4CAAPz111/Ys2cPpk6dipEjR0JPT0+l2IlIB0g9iYRIW16dNFqW4ykpKcLAgQOFKlWqCHK5XKhZs6YwdOhQITMzUxCEl5NER48eLZibmwuWlpZCcHCwMHDgwFInjQqCIDx//lwICgoS7OzsBENDQ8HFxUVYvXq1eDwsLEywtbUVZDKZ4OfnJwjCy4muCxYsEOrUqSMYGBgI1tbWgo+Pj3D48GHxvJ07dwouLi6CXC4XWrduLaxevVqlSaMAir3Wr18v5ObmCv7+/oKFhYVgaWkpjBgxQpg4caLQqFGjYp9baGioULlyZcHMzEwYOnSokJubK9Z5W+ycNEpU/skEoZSZbURERERqwiEVIiIi0jgmHERERKRxTDiIiIhI45hwEBERkcYx4SAiIiKNY8JBREREGseEg4iIiDSOCQcRERFpHBMOIiIi0jgmHERERKRxTDiIiIhI4/4PlNZeDYKzrcgAAAAASUVORK5CYII=",
      "text/plain": [
       "<Figure size 600x500 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.metrics import confusion_matrix, classification_report\n",
    "import torch\n",
    "import numpy as np\n",
    "import seaborn as sns\n",
    "import matplotlib.pyplot as plt\n",
    "from transformers import BertTokenizer, BertForSequenceClassification\n",
    "from safetensors.torch import load_file\n",
    "\n",
    "\n",
    "predicted_labels = predict_hate_speech(val_texts)\n",
    "\n",
    "# Generate confusion matrix\n",
    "cm = confusion_matrix(val_labels, predicted_labels)\n",
    "\n",
    "# Print classification report\n",
    "print(\"Classification Report:\\n\", classification_report(val_labels, predicted_labels, target_names=[\"Not Hate Speech\", \"Hate Speech\"]))\n",
    "\n",
    "# Plot confusion matrix\n",
    "plt.figure(figsize=(6, 5))\n",
    "sns.heatmap(cm, annot=True, fmt=\"d\", cmap=\"Blues\", xticklabels=[\"Not Hate Speech\", \"Hate Speech\"], yticklabels=[\"Not Hate Speech\", \"Hate Speech\"])\n",
    "plt.xlabel(\"Predicted Label\")\n",
    "plt.ylabel(\"True Label\")\n",
    "plt.title(\"Confusion Matrix\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "8c7ad38e-0899-4b8f-845d-1c16ef4af821",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 800x500 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.metrics import precision_recall_fscore_support\n",
    "\n",
    "# Compute precision, recall, and f1-score\n",
    "precision, recall, f1, _ = precision_recall_fscore_support(val_labels, predicted_labels, average=None)\n",
    "\n",
    "# Create DataFrame for visualization\n",
    "metrics_df = pd.DataFrame({\n",
    "    \"Class\": [\"Not Hate Speech\", \"Hate Speech\"],\n",
    "    \"Precision\": precision,\n",
    "    \"Recall\": recall,\n",
    "    \"F1-score\": f1\n",
    "})\n",
    "\n",
    "# Plot\n",
    "metrics_df.set_index(\"Class\").plot(kind=\"bar\", figsize=(8, 5), colormap=\"coolwarm\", edgecolor=\"black\")\n",
    "plt.title(\"Precision, Recall, and F1-score\")\n",
    "plt.ylabel(\"Score\")\n",
    "plt.xticks(rotation=0)\n",
    "plt.ylim(0, 1)\n",
    "plt.grid(axis=\"y\", linestyle=\"--\", alpha=0.7)\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "4cee6336-2290-4469-842f-008b0a945841",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Text:  i ve seen afro americans who look and undoubtedly act like him\n",
      "Prediction: Hate Speech\n",
      "\n",
      "Text: hyper text transfer protocol www occultopedia com f feral children http www long lands ac uk chimp la http www occultopedia com wild peter him aragon again every time i will see filthy comment about georgia without providing any facts i will respond\n",
      "Prediction: Not Hate Speech\n",
      "\n",
      "Text: i will respond until armenian scum will stop\n",
      "Prediction: Hate Speech\n",
      "\n",
      "Text: the reason large white families were encouraged those days what a slut to build up the population of white germans which were ravaged by the jewish attacks at them during and in hyperinflation\n",
      "Prediction: Not Hate Speech\n",
      "\n",
      "Text: i see a nation where i can walk out my front door and not feel like a minority in my own country\n",
      "Prediction: Not Hate Speech\n",
      "\n",
      "Text: where a white woman can go about in public and not fear assault by mud,no i see that as giving up\n",
      "Prediction: Hate Speech\n",
      "\n",
      "Text: i hope one day to proudly say again that i am an american when whatever take this country back\n",
      "Prediction: Not Hate Speech\n",
      "\n",
      "Text: okay i have been through some other threads on this section about nice and watched a few videos on you there\n",
      "Prediction: Not Hate Speech\n",
      "\n",
      "Text: it has obviously been misunderstood and i would have been keen to learn more\n",
      "Prediction: Not Hate Speech\n",
      "\n",
      "Text: i retract always because an animal lacks the skills to properly raise a half human child\n",
      "Prediction: Hate Speech\n",
      "\n",
      "Text: i apologize for my ignorance\n",
      "Prediction: Not Hate Speech\n",
      "\n"
     ]
    }
   ],
   "source": [
    "import torch\n",
    "\n",
    "# Set device (GPU if available)\n",
    "device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\n",
    "model.to(device)  # Move model to GPU\n",
    "\n",
    "def predict_hate_speech(texts, batch_size=32):\n",
    "    \"\"\"Efficiently tokenize and predict labels in batches.\"\"\"\n",
    "    all_preds = []\n",
    "    model.eval()  # Ensure model is in evaluation mode\n",
    "\n",
    "    for i in range(0, len(texts), batch_size):\n",
    "        batch_texts = texts[i : i + batch_size]\n",
    "\n",
    "        # Tokenize the batch\n",
    "        inputs = tokenizer(batch_texts, return_tensors=\"pt\", padding=True, truncation=True, max_length=128)\n",
    "        inputs = {key: val.to(device) for key, val in inputs.items()}  # Move to GPU\n",
    "\n",
    "        with torch.no_grad():\n",
    "            outputs = model(**inputs)\n",
    "\n",
    "        # Get predicted labels (0 = Not Hate Speech, 1 = Hate Speech)\n",
    "        logits = outputs.logits\n",
    "        batch_preds = torch.argmax(logits, dim=-1).cpu().tolist()  # Move to CPU\n",
    "        all_preds.extend(batch_preds)\n",
    "\n",
    "    return all_preds\n",
    "\n",
    "# Example sentences\n",
    "samples = [\n",
    " \" i ve seen afro americans who look and undoubtedly act like him\",\n",
    "\"hyper text transfer protocol www occultopedia com f feral children http www long lands ac uk chimp la http www occultopedia com wild peter him aragon again every time i will see filthy comment about georgia without providing any facts i will respond\",\n",
    "\"i will respond until armenian scum will stop\",\n",
    "\"the reason large white families were encouraged those days what a slut to build up the population of white germans which were ravaged by the jewish attacks at them during and in hyperinflation\",\n",
    "\"i see a nation where i can walk out my front door and not feel like a minority in my own country\",\n",
    "\"where a white woman can go about in public and not fear assault by mud,\"\n",
    "\"no i see that as giving up\",\n",
    "\"i hope one day to proudly say again that i am an american when whatever take this country back\",\n",
    "\"okay i have been through some other threads on this section about nice and watched a few videos on you there\",\n",
    "\"it has obviously been misunderstood and i would have been keen to learn more\",\n",
    "\"i retract always because an animal lacks the skills to properly raise a half human child\",\n",
    "\"i apologize for my ignorance\"\n",
    "]\n",
    "\n",
    "# Run predictions\n",
    "predictions = predict_hate_speech(samples)\n",
    "\n",
    "# Print results\n",
    "for text, pred in zip(samples, predictions):\n",
    "    label = \"Hate Speech\" if pred == 1 else \"Not Hate Speech\"\n",
    "    print(f\"Text: {text}\\nPrediction: {label}\\n\")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "4622aa93-fa01-4955-97eb-3a67de6483b4",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "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": "3.9.21"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}