CLIP-32-Evaluation.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "e7bd18d3",
   "metadata": {},
   "source": [
    "# Multi-modal Embedding using CLIP-32"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a7e3f228",
   "metadata": {},
   "source": [
    "## Imports and Utilities"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "78762744",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sentence_transformers import SentenceTransformer, util, models, InputExample, losses\n",
    "from PIL import Image\n",
    "import glob\n",
    "import torch\n",
    "from torch.utils.data import DataLoader\n",
    "import pickle\n",
    "import zipfile\n",
    "from IPython.display import display\n",
    "from IPython.display import Image as IPImage\n",
    "import pandas as pd\n",
    "import os\n",
    "from tqdm.autonotebook import tqdm, trange\n",
    "import torch\n",
    "from matplotlib import pyplot as plt\n",
    "from matplotlib import image as mpimg\n",
    "import numpy as np\n",
    "from sklearn import metrics\n",
    "from sklearn.metrics import classification_report, confusion_matrix, f1_score, precision_score, recall_score"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "9056b848",
   "metadata": {},
   "outputs": [],
   "source": [
    "image_folder = 'images/'\n",
    "def show_image(img_id, img_folder=image_folder):\n",
    "    try:\n",
    "        image_link = img_folder + img_id + '.jpg'\n",
    "        plt.title(\"Image\")\n",
    "        plt.xlabel(\"X pixel scaling\")\n",
    "        plt.ylabel(\"Y pixels scaling\")\n",
    "\n",
    "        image = mpimg.imread(image_link)\n",
    "    except:\n",
    "        image_link = img_folder + img_id + '.png'\n",
    "        plt.title(\"Image\")\n",
    "        plt.xlabel(\"X pixel scaling\")\n",
    "        plt.ylabel(\"Y pixels scaling\")\n",
    "\n",
    "        image = mpimg.imread(image_link)\n",
    "    plt.imshow(image)\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "0415eab5",
   "metadata": {},
   "outputs": [],
   "source": [
    "import random\n",
    "def set_seed_everywhere(seed, cuda):\n",
    "    np.random.seed(seed)\n",
    "    torch.manual_seed(seed)\n",
    "    random.seed(seed)\n",
    "    if cuda:\n",
    "        torch.cuda.manual_seed_all(seed)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "366ad8a4",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>index</th>\n",
       "      <th>id</th>\n",
       "      <th>skepticism</th>\n",
       "      <th>mainstream</th>\n",
       "      <th>label</th>\n",
       "      <th>text_NL</th>\n",
       "      <th>text_EN</th>\n",
       "      <th>split</th>\n",
       "      <th>available</th>\n",
       "      <th>target</th>\n",
       "      <th>unseen_images</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>956</th>\n",
       "      <td>956</td>\n",
       "      <td>327027443732983808</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>1</td>\n",
       "      <td>RT @USER : 3D waterkaart , hoe stroomt het wat...</td>\n",
       "      <td>RT @USER : 3D water map , how the water flows ...</td>\n",
       "      <td>test</td>\n",
       "      <td>True</td>\n",
       "      <td>3_327026285966667776</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1211</th>\n",
       "      <td>1211</td>\n",
       "      <td>375896131127218176</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>1</td>\n",
       "      <td>Kom naar de Wilhelminakade R'dam en klim boven...</td>\n",
       "      <td>Come to the Wilhelminakade R'dam and climb abo...</td>\n",
       "      <td>val</td>\n",
       "      <td>True</td>\n",
       "      <td>3_375896131013971968</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1212</th>\n",
       "      <td>1212</td>\n",
       "      <td>375899134915772416</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>1</td>\n",
       "      <td>RT @USER : Kom naar de Wilhelminakade R'dam en...</td>\n",
       "      <td>RT @USER : Come to the Wilhelminakade R'dam an...</td>\n",
       "      <td>val</td>\n",
       "      <td>True</td>\n",
       "      <td>3_375896131013971968</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1214</th>\n",
       "      <td>1214</td>\n",
       "      <td>375946829881278464</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>1</td>\n",
       "      <td>Scholieren klimmen met Jeugdwaterschap naar ni...</td>\n",
       "      <td>Students climb to the water level of their hou...</td>\n",
       "      <td>val</td>\n",
       "      <td>True</td>\n",
       "      <td>3_375946829747089408</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1215</th>\n",
       "      <td>1215</td>\n",
       "      <td>375946961951539200</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>1</td>\n",
       "      <td>RT @USER : Scholieren klimmen met Jeugdwatersc...</td>\n",
       "      <td>RT @USER : Students climb to the water level o...</td>\n",
       "      <td>train</td>\n",
       "      <td>True</td>\n",
       "      <td>3_375946829747089408</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      index                  id  skepticism  mainstream  label  \\\n",
       "956     956  327027443732983808       False        True      1   \n",
       "1211   1211  375896131127218176       False        True      1   \n",
       "1212   1212  375899134915772416       False        True      1   \n",
       "1214   1214  375946829881278464       False        True      1   \n",
       "1215   1215  375946961951539200       False        True      1   \n",
       "\n",
       "                                                text_NL  \\\n",
       "956   RT @USER : 3D waterkaart , hoe stroomt het wat...   \n",
       "1211  Kom naar de Wilhelminakade R'dam en klim boven...   \n",
       "1212  RT @USER : Kom naar de Wilhelminakade R'dam en...   \n",
       "1214  Scholieren klimmen met Jeugdwaterschap naar ni...   \n",
       "1215  RT @USER : Scholieren klimmen met Jeugdwatersc...   \n",
       "\n",
       "                                                text_EN  split  available  \\\n",
       "956   RT @USER : 3D water map , how the water flows ...   test       True   \n",
       "1211  Come to the Wilhelminakade R'dam and climb abo...    val       True   \n",
       "1212  RT @USER : Come to the Wilhelminakade R'dam an...    val       True   \n",
       "1214  Students climb to the water level of their hou...    val       True   \n",
       "1215  RT @USER : Students climb to the water level o...  train       True   \n",
       "\n",
       "                    target  unseen_images  \n",
       "956   3_327026285966667776              1  \n",
       "1211  3_375896131013971968              1  \n",
       "1212  3_375896131013971968              1  \n",
       "1214  3_375946829747089408              0  \n",
       "1215  3_375946829747089408              0  "
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = pd.read_csv('data/tweets_subset.csv', index_col='Unnamed: 0').reset_index()\n",
    "df[df.available].head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "44baebe8",
   "metadata": {},
   "outputs": [],
   "source": [
    "# We use the original CLIP model for computing image embeddings and English text embeddings\n",
    "en_model = SentenceTransformer('clip-ViT-B-32')\n",
    "tokenizer = en_model._first_module().processor.tokenizer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "c10f520a",
   "metadata": {},
   "outputs": [],
   "source": [
    "def embed_documents(documents, embedding_model=en_model, convert_to_tensor = True, verbose = False):\n",
    "    \"\"\" Embed a list of n documents/words into an n-dimensional\n",
    "    matrix of embeddings\n",
    "\n",
    "    Arguments:\n",
    "        documents: A list of documents or words to be embedded\n",
    "        verbose: Controls the verbosity of the process\n",
    "\n",
    "    Returns:\n",
    "        Document/words embeddings with shape (n, m) with `n` documents/words\n",
    "        that each have an embeddings size of `m`\n",
    "    \"\"\"\n",
    "    truncated_docs = [truncate_document(doc) for doc in documents]\n",
    "    embeddings = embedding_model.encode(truncated_docs, show_progress_bar=verbose, convert_to_tensor=convert_to_tensor)\n",
    "    return embeddings\n",
    "\n",
    "def truncate_document(document, tokenizer=tokenizer):\n",
    "    tokens = tokenizer.encode(document)\n",
    "\n",
    "    if len(tokens) > 77:\n",
    "        # Skip the starting token, only include 75 tokens\n",
    "        truncated_tokens = tokens[1:76]\n",
    "        document = tokenizer.decode(truncated_tokens)\n",
    "\n",
    "        # Recursive call here, because the encode(decode()) can have different result\n",
    "        return truncate_document(document)\n",
    "\n",
    "    else:\n",
    "        return document"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "a5a051b2",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Token indices sequence length is longer than the specified maximum sequence length for this model (80 > 77). Running this sequence through the model will result in indexing errors\n"
     ]
    }
   ],
   "source": [
    "df['text_EN_trun'] = df['text_EN'].apply(truncate_document)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "3d195eeb",
   "metadata": {},
   "outputs": [],
   "source": [
    "def embed_images(images, verbose=True, batch_size=32, embedding_model=en_model, convert_to_tensor=True):\n",
    "    if batch_size:\n",
    "        nr_iterations = int(np.ceil(len(images) / batch_size))\n",
    "\n",
    "        # Embed images per batch\n",
    "        embeddings = []\n",
    "        for i in tqdm(range(nr_iterations), disable=not verbose):\n",
    "            start_index = i * batch_size\n",
    "            end_index = (i * batch_size) + batch_size\n",
    "\n",
    "            images_to_embed = [Image.open(img_name_dict[image]) if (isinstance(image, str) and not str(image)=='nan') \n",
    "                               else (Image.new('RGB', (800, 600), color = (255,255,255)) if str(image)=='nan'\n",
    "                               else image) for image in images[start_index:end_index]]\n",
    "            img_emb = embedding_model.encode(images_to_embed, show_progress_bar=False, convert_to_tensor=convert_to_tensor)\n",
    "            embeddings.extend(img_emb.tolist())\n",
    "\n",
    "            # Close images\n",
    "            if isinstance(images[0], str):\n",
    "                for image in images_to_embed:\n",
    "                    image.close()\n",
    "        if not convert_to_tensor:\n",
    "            embeddings = np.array(embeddings)\n",
    "        else:\n",
    "            embeddings = torch.tensor(embeddings).to(device='cuda')\n",
    "    else:\n",
    "        images_to_embed = [Image.open(filepath) for filepath in images]\n",
    "        embeddings = embedding_model.encode(images_to_embed, show_progress_bar=False, convert_to_tensor=convert_to_tensor)\n",
    "    return embeddings"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "54bd8cdf",
   "metadata": {},
   "outputs": [],
   "source": [
    "def return_report(labels, preds, ids=[]):\n",
    "    if len(ids)==0:\n",
    "        print(metrics.classification_report(labels.cpu(), preds.cpu(), digits=4))\n",
    "        return metrics.classification_report(labels.cpu(), preds.cpu(), digits=4, output_dict=True)\n",
    "    else:\n",
    "        print(metrics.classification_report(labels[torch.tensor(list(ids))].cpu(), preds[torch.tensor(list(ids))].cpu(), digits=4))\n",
    "        return metrics.classification_report(labels[torch.tensor(list(ids))].cpu(), preds[torch.tensor(list(ids))].cpu(), digits=4, output_dict=True)\n",
    "\n",
    "def evaluate_dataset_F1(df, preds_all, device = 'cuda', val_set = 'val'):\n",
    "    preds_all = preds_all.float()\n",
    "    label_all = torch.tensor(np.array(df['skepticism'])).to(device).float()\n",
    "    train = df[df['split'] == 'train'].index\n",
    "    val = df[df['split'] == 'val'].index\n",
    "    test = df[df['split'] == 'test'].index\n",
    "    image = df[df['available']].index\n",
    "    unseen = df[df['unseen_images']==1].index\n",
    "    \n",
    "    if val_set=='val':\n",
    "        temp = val.copy()\n",
    "    elif val_set=='test':\n",
    "        temp = test.copy()\n",
    "    \n",
    "    a = temp.intersection(image)\n",
    "    b = temp.difference(image)\n",
    "    c = temp.intersection(image).intersection(unseen)    \n",
    "    \n",
    "    print('Overall Dataset:')\n",
    "    all_report = return_report(label_all, preds_all)\n",
    "    \n",
    "    print('Training Set:')\n",
    "    train_report = return_report(label_all, preds_all, train)\n",
    "    \n",
    "    print('Validation Set:')\n",
    "    val_report = return_report(label_all, preds_all, temp)\n",
    "    \n",
    "    print('Validation Set with both Images and Texts:')\n",
    "    val_img_report = return_report(label_all, preds_all, a)\n",
    "    \n",
    "    print('Validation Set with only Texts:')\n",
    "    val_txt_report = return_report(label_all, preds_all, b)\n",
    "    \n",
    "    print('Validation Set with unseen Images:')\n",
    "    val_unseen_report = return_report(label_all, preds_all, c)\n",
    "    \n",
    "    all_acc = ((label_all==preds_all).sum()/len(preds_all)).item()*100\n",
    "    train_acc = ((label_all[torch.tensor(list(train))]==preds_all[torch.tensor(list(train))]).sum()/len(train)).item()*100\n",
    "    val_acc = ((label_all[torch.tensor(list(temp))]==preds_all[torch.tensor(list(temp))]).sum()/len(temp)).item()*100\n",
    "    val_img_acc = ((label_all[torch.tensor(list((a)))]==preds_all[torch.tensor(list(a))]).sum()/len(a)).item()*100\n",
    "    val_txt_acc = ((label_all[torch.tensor(list((b)))]==preds_all[torch.tensor(list((b)))]).sum()/len(b)).item()*100\n",
    "    val_unseen = ((label_all[torch.tensor(list((c)))]==preds_all[torch.tensor(list((c)))]).sum()/len(c)).item()*100\n",
    "    \n",
    "    print('''\n",
    "    Overall Accuracy: {:.2f}\n",
    "    Training Set Accuracy: {:.2f}\n",
    "    Validation Set Accuracy: {:.2f}\n",
    "    Validation Set with both Images and Texts: {:.2f}\n",
    "    Validation Set with only Texts: {:.2f}\n",
    "    Validation Set with unseen Images: {:.2f}\n",
    "    '''.format(all_acc, train_acc, val_acc, val_img_acc, val_txt_acc, val_unseen))\n",
    "    \n",
    "    return all_report, train_report, val_report, val_img_report, val_txt_report, val_unseen_report"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "10dd7c2d",
   "metadata": {},
   "outputs": [],
   "source": [
    "def evaluate_dataset(df, preds_all, device = 'cuda', val_set = 'val'):\n",
    "    preds_all = preds_all.float()\n",
    "    label_all = torch.tensor(np.array(df['skepticism'])).to(device).float()\n",
    "    #train = torch.tensor(np.array(df_all['split'] == 'train')).to(device).float()\n",
    "    train = df[df['split'] == 'train'].index\n",
    "    #val = torch.tensor(np.array(df_all['split'] == 'val')).to(device).float()\n",
    "    val = df[df['split'] == 'val'].index\n",
    "    #test = torch.tensor(np.array(df_all['split'] == 'test')).to(device).float()\n",
    "    test = df[df['split'] == 'test'].index\n",
    "    #image = torch.tensor(np.array(df_all['available'])).to(device).float()\n",
    "    image = df[df['available']].index\n",
    "    #unseen = torch.tensor(np.array(df_all['unseen_images'])).to(device).float()\n",
    "    unseen = df[df['unseen_images']==1].index\n",
    "    \n",
    "    if val_set=='val':\n",
    "        a = val.intersection(image)\n",
    "        b = val.difference(image)\n",
    "        c = val.intersection(image).intersection(unseen)\n",
    "        val_acc = ((label_all[torch.tensor(list(val))]==preds_all[torch.tensor(list(val))]).sum()/len(val)).item()*100\n",
    "            \n",
    "    elif val_set=='test':\n",
    "        a = test.intersection(image)\n",
    "        b = test.difference(image)\n",
    "        c = test.intersection(image).intersection(unseen)\n",
    "        val_acc = ((label_all[torch.tensor(list(test))]==preds_all[torch.tensor(list(test))]).sum()/len(test)).item()*100\n",
    "    \n",
    "    all_acc = ((label_all==preds_all).sum()/len(preds_all)).item()*100\n",
    "    train_acc = ((label_all[torch.tensor(list(train))]==preds_all[torch.tensor(list(train))]).sum()/len(train)).item()*100\n",
    "    \n",
    "    val_img_acc = ((label_all[torch.tensor(list((a)))]==preds_all[torch.tensor(list(a))]).sum()/len(a)).item()*100\n",
    "    val_txt_acc = ((label_all[torch.tensor(list((b)))]==preds_all[torch.tensor(list((b)))]).sum()/len(b)).item()*100\n",
    "    val_unseen = ((label_all[torch.tensor(list((c)))]==preds_all[torch.tensor(list((c)))]).sum()/len(c)).item()*100\n",
    "    \n",
    "    print('''\n",
    "    Overall Accuracy: {:.2f}\n",
    "    Training Set Accuracy: {:.2f}\n",
    "    Validation Set Accuracy: {:.2f}\n",
    "    Validation Set with both Images and Texts: {:.2f}\n",
    "    Validation Set with only Texts: {:.2f}\n",
    "    Validation Set with unseen Images: {:.2f}\n",
    "    '''.format(all_acc, train_acc, val_acc, val_img_acc, val_txt_acc, val_unseen))\n",
    "    \n",
    "    return all_acc, train_acc, val_acc, val_img_acc, val_txt_acc, val_unseen"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c374edb7",
   "metadata": {},
   "source": [
    "## Process Raw Data into Embeddings"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1ecee2fc",
   "metadata": {},
   "source": [
    "Two sentences to compute embeddings"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "64e022b7",
   "metadata": {},
   "outputs": [],
   "source": [
    "labels = ['Climate change is a very serious issue to pay attention to! We should care about climate change!', \n",
    "          'Climate change is not even a real problem to concern about! I do not believe in it!']\n",
    "lab_emb = en_model.encode(labels, convert_to_tensor=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "6b1f19aa",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "08cc1015e2164ae7a963759abc7426c0",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/1 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "empty_img_emb = embed_images(['nan'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "e4e48609",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "torch.Size([1, 512])"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "empty_img_emb.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "2b7d10eb",
   "metadata": {},
   "outputs": [],
   "source": [
    "txt_emb = embed_documents(list(df['text_EN']), convert_to_tensor=True, verbose=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "7799ab06",
   "metadata": {},
   "outputs": [],
   "source": [
    "img_emb = embed_images(list(df['target']), convert_to_tensor=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "00688ee3",
   "metadata": {},
   "outputs": [],
   "source": [
    "import pickle\n",
    "with open('data/CLIP32/embeddings.pkl', \"wb\") as fOut:\n",
    "    pickle.dump({'sentences': list(df['text_EN_trun']), 'txt_emb': txt_emb, 'img_emb': img_emb, \n",
    "                 'lab_emb': lab_emb, 'empty_img_emb': empty_img_emb}, fOut, protocol=pickle.HIGHEST_PROTOCOL)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "id": "96befd9a",
   "metadata": {},
   "outputs": [],
   "source": [
    "with open('data/CLIP32/embeddings.pkl', \"rb\") as fIn:\n",
    "    stored_data = pickle.load(fIn)\n",
    "    #stored_sentences = stored_data['sentences']\n",
    "    txt_emb = stored_data['txt_emb']\n",
    "    img_emb = stored_data['img_emb']\n",
    "    lab_emb = stored_data['lab_emb']\n",
    "    empty_img_emb = stored_data['empty_img_emb']\n",
    "\n",
    "labs = {}\n",
    "labs['txt_emb'] = lab_emb\n",
    "labs['img_emb'] = torch.vstack([empty_img_emb] * lab_emb.shape[0])\n",
    "mul_emb = (img_emb + txt_emb)/2"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4d53de3d",
   "metadata": {},
   "source": [
    "## Zero-shot Classification"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "da062fc9",
   "metadata": {},
   "source": [
    "Using Cosine Similarity to make classification"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "id": "990b44e0",
   "metadata": {},
   "outputs": [],
   "source": [
    "def zero_shot(df_all, model_name='CLIP32', merge = 'average', from_stored_embedding=False, pre_computed = True,\n",
    "             stored_data='', embedding_model = '', val_set='val'):\n",
    "    \n",
    "    if from_stored_embedding:\n",
    "        folder = 'model_storage/{}/embeddings.pkl'.format(model_name)\n",
    "        with open(folder, \"rb\") as fIn:\n",
    "            stored_data = pickle.load(fIn)\n",
    "            #stored_sentences = stored_data['sentences']\n",
    "            txt_emb = stored_data['txt_emb']\n",
    "            img_emb = stored_data['img_emb']\n",
    "            lab_emb = stored_data['lab_emb']\n",
    "    elif pre_computed:\n",
    "        stored_data = stored_data\n",
    "        #stored_sentences = stored_data['sentences']\n",
    "        txt_emb = stored_data['txt_emb']\n",
    "        img_emb = stored_data['img_emb']\n",
    "        lab_emb = stored_data['lab_emb']\n",
    "    else:\n",
    "        txt_emb = embed_documents(list(df_all['text_EN_trun']), embedding_model=embedding_model, convert_to_tensor=True, verbose=True)\n",
    "        img_emb = embed_images(list(df_all['target']), embedding_model=embedding_model, convert_to_tensor=True, verbose=True)\n",
    "        labels = ['Climate change is a very serious issue to pay attention to! We should care about climate change!', \n",
    "                  'Climate change is not even a real problem to concern about! I do not believe in it!']\n",
    "        lab_emb = embedding_model.encode(labels, convert_to_tensor=True)\n",
    "        with open('model_storage/CLIP/embeddings_re.pkl', \"wb\") as fOut:\n",
    "            pickle.dump({'sentences': list(df['text_EN_trun']), 'txt_emb': txt_emb, 'img_emb': img_emb, 'lab_emb': lab_emb}, fOut, protocol=pickle.HIGHEST_PROTOCOL)\n",
    "    \n",
    "    if merge == 'average':\n",
    "        mul_emb = (img_emb + txt_emb)/2\n",
    "    elif merge == 'max':\n",
    "        mul_emb = torch.max(img_emb, txt_emb)\n",
    "    else:\n",
    "        print(\"'Merger' input must be one of 'average' or 'max'\")\n",
    "    \n",
    "    TXT_LAB = util.cos_sim(lab_emb, txt_emb)\n",
    "    IMG_LAB = util.cos_sim(lab_emb, img_emb)\n",
    "    MUL_LAB = util.cos_sim(lab_emb, mul_emb)\n",
    "    \n",
    "    pred_TXT = torch.argmax(TXT_LAB[:2,:], dim=0).float()\n",
    "    pred_IMG = torch.argmax(IMG_LAB[:2,:], dim=0).float()\n",
    "    pred_MUL = torch.argmax(MUL_LAB[:2,:], dim=0).float()\n",
    "    \n",
    "    lab = torch.tensor(np.array(df_all['skepticism'])).to('cuda').float()\n",
    "    \n",
    "    set_seed_everywhere(42, cuda=True)\n",
    "    lab_rand = (torch.rand(len(lab))>0.5).float().to('cuda')\n",
    "    \n",
    "    print('Using textual embedding only:')\n",
    "    evaluate_dataset(df_all, pred_TXT, val_set=val_set)\n",
    "    \n",
    "    print('Using image embedding only:')\n",
    "    evaluate_dataset(df_all, pred_IMG, val_set=val_set)\n",
    "    \n",
    "    print('Using multi-modal embedding:')\n",
    "    evaluate_dataset(df_all, pred_MUL, val_set=val_set)\n",
    "    \n",
    "    print('Comparing to a random prediction:')\n",
    "    evaluate_dataset(df_all, lab_rand, val_set=val_set)\n",
    "    \n",
    "    print('Test set performance multi-modal:')\n",
    "    evaluate_dataset(df_all, pred_MUL, val_set='test')\n",
    "    \n",
    "    return"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "id": "cb85005f",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Using textual embedding only:\n",
      "\n",
      "    Overall Accuracy: 61.66\n",
      "    Training Set Accuracy: 61.63\n",
      "    Validation Set Accuracy: 62.58\n",
      "    Validation Set with both Images and Texts: 74.17\n",
      "    Validation Set with only Texts: 60.45\n",
      "    Validation Set with unseen Images: 70.19\n",
      "    \n",
      "Using image embedding only:\n",
      "\n",
      "    Overall Accuracy: 45.92\n",
      "    Training Set Accuracy: 45.95\n",
      "    Validation Set Accuracy: 46.59\n",
      "    Validation Set with both Images and Texts: 68.96\n",
      "    Validation Set with only Texts: 42.47\n",
      "    Validation Set with unseen Images: 66.25\n",
      "    \n",
      "Using multi-modal embedding:\n",
      "\n",
      "    Overall Accuracy: 57.28\n",
      "    Training Set Accuracy: 56.93\n",
      "    Validation Set Accuracy: 58.33\n",
      "    Validation Set with both Images and Texts: 75.30\n",
      "    Validation Set with only Texts: 55.21\n",
      "    Validation Set with unseen Images: 71.01\n",
      "    \n",
      "Comparing to a random prediction:\n",
      "\n",
      "    Overall Accuracy: 49.64\n",
      "    Training Set Accuracy: 49.67\n",
      "    Validation Set Accuracy: 49.99\n",
      "    Validation Set with both Images and Texts: 48.09\n",
      "    Validation Set with only Texts: 50.34\n",
      "    Validation Set with unseen Images: 47.20\n",
      "    \n",
      "Test set performance multi-modal:\n",
      "\n",
      "    Overall Accuracy: 57.28\n",
      "    Training Set Accuracy: 56.93\n",
      "    Validation Set Accuracy: 57.85\n",
      "    Validation Set with both Images and Texts: 73.59\n",
      "    Validation Set with only Texts: 55.21\n",
      "    Validation Set with unseen Images: 70.51\n",
      "    \n"
     ]
    }
   ],
   "source": [
    "zero_shot(df, stored_data = stored_data)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "92f68085",
   "metadata": {},
   "source": [
    "## Training an additional MLP"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "id": "496fe10f",
   "metadata": {},
   "outputs": [],
   "source": [
    "from argparse import Namespace\n",
    "args = Namespace(\n",
    "    # Data and Path information\n",
    "    model_state_file='MLP_classifier.pth',\n",
    "    save_dir='model_storage/CLIP32/',\n",
    "    result_dir='results/CLIP32/',\n",
    "    embedding_size=512,\n",
    "    early_stopping_criteria=20,\n",
    "    num_epochs=20,\n",
    "    seed=42,\n",
    "    \n",
    "    # Training hyper parameters\n",
    "    batch_size=32,\n",
    "    embedding_type='text',\n",
    "    fusion_type='average',\n",
    "    hidden_dim=128, \n",
    "    learning_rate=5e-5,\n",
    "    l2=1e-5,\n",
    "    dropout_p=0.1,\n",
    "    \n",
    "    # Runtime options\n",
    "    catch_keyboard_interrupt=True,\n",
    "    device= 'cuda',\n",
    "    cuda=True,\n",
    "    verbose=True,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "b8720053",
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch.nn as nn\n",
    "import torch.nn.functional as F\n",
    "import torch.optim as optim\n",
    "from torch.utils.data import Dataset, DataLoader\n",
    "\n",
    "class MLPClassifier(nn.Module):\n",
    "    \n",
    "    def __init__(self, embedding_size, embedding_type, fusion_type, hidden_dim, num_classes, dropout_p, \n",
    "                 pretrained_embeddings=None):\n",
    "        \"\"\"\n",
    "        Args:\n",
    "            embedding_size (int): size of the embedding vectors\n",
    "            embedding_type (str): type of embedding to be used, can be one of {'text', 'image', 'multi'}\n",
    "            fusion_type (str): type of fusing multimodal embeddings if embedding_type=='multi', can be one of \n",
    "                                {'average', 'concat', 'complex'}\n",
    "            hidden_dim (int): the size of the hidden dimension\n",
    "            num_classes (int): the number of classes in classification\n",
    "            dropout_p (float): a dropout parameter \n",
    "            pretrained_embeddings (numpy.array): previously trained embeddings\n",
    "            \n",
    "        \"\"\"\n",
    "        super(MLPClassifier, self).__init__()\n",
    "\n",
    "        self.embedding_type = embedding_type\n",
    "        self.fusion_type = fusion_type\n",
    "        self._dropout_p = dropout_p\n",
    "        self.dropout = nn.Dropout(dropout_p)\n",
    "        #self.prelu = nn.PReLU()\n",
    "        \n",
    "        self.fc1 = nn.Linear(embedding_size, hidden_dim)\n",
    "        self.fc2 = nn.Linear(hidden_dim, num_classes)\n",
    "        \n",
    "        self.fc_concat = nn.Linear(2*embedding_size, hidden_dim)\n",
    "        self.fc_complex = nn.Linear(5*embedding_size, 2*embedding_size)\n",
    "\n",
    "    def forward(self, x_in, apply_softmax=False):\n",
    "        \"\"\"The forward pass of the classifier\n",
    "        \n",
    "        Args:\n",
    "            x_in (dictionary): a dictionary of input data tensors with embeddings. \n",
    "                x_in.shape should be (batch, dataset._max_seq_length)\n",
    "            apply_softmax (bool): a flag for the softmax activation\n",
    "                should be false if used with the Cross Entropy losses\n",
    "        Returns:\n",
    "            the resulting tensor. tensor.shape should be (batch, num_classes)\n",
    "        \"\"\"\n",
    "        \n",
    "        # embed and permute so features are channels\n",
    "        x_embedded = x_in\n",
    "                \n",
    "        # logistic regression\n",
    "        if (not self.embedding_type=='multi') or self.fusion_type=='average':\n",
    "            intermediate_vector = F.relu(self.dropout(self.fc1(x_embedded)))\n",
    "            prediction_vector = self.fc2(intermediate_vector)\n",
    "        elif self.embedding_type=='multi' and self.fusion_type=='concat':\n",
    "            intermediate_vector = F.relu(self.dropout(self.fc_concat(x_embedded)))\n",
    "            prediction_vector = self.fc2(intermediate_vector)\n",
    "        elif self.embedding_type=='multi' and self.fusion_type=='complex':\n",
    "            first_vector = F.relu(self.dropout(self.fc_complex(x_embedded)))\n",
    "            intermediate_vector = F.relu(self.dropout(self.fc_concat(first_vector)))\n",
    "            prediction_vector = self.fc2(intermediate_vector)\n",
    "        \n",
    "        if apply_softmax:\n",
    "            prediction_vector = F.softmax(prediction_vector, dim=1)\n",
    "\n",
    "        return prediction_vector"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "4928e3f9",
   "metadata": {},
   "outputs": [],
   "source": [
    "class CustomDataset(Dataset):\n",
    "    def __init__(self, df_all, embeddings):\n",
    "        self.df = df_all\n",
    "        self.labels = torch.tensor(np.array(self.df['skepticism'])).to(args.device).float()\n",
    "        measure_len = lambda context: len(context.split(\" \"))\n",
    "        self._max_seq_length = max(map(measure_len, self.df.text_EN_trun)) + 0\n",
    "        self.embeddings = embeddings\n",
    "        \n",
    "        train = torch.tensor(df[df['split'] == 'train'].index)\n",
    "        self.train_df = self.df[self.df.split=='train']\n",
    "        self.train_size = len(self.train_df)\n",
    "        self.train_emb = self.embeddings[train]\n",
    "        \n",
    "        val = torch.tensor(df[df['split'] == 'val'].index)\n",
    "        self.val_df = self.df[self.df.split=='val']\n",
    "        self.validation_size = len(self.val_df)\n",
    "        self.val_emb = self.embeddings[val]\n",
    "                \n",
    "        test = torch.tensor(df[df['split'] == 'test'].index)\n",
    "        self.test_df = self.df[self.df.split=='test']\n",
    "        self.test_size = len(self.test_df)\n",
    "        self.test_emb = self.embeddings[test]\n",
    "\n",
    "        self._lookup_dict = {'train': (self.train_df, self.train_size, self.train_emb),\n",
    "                             'val': (self.val_df, self.validation_size, self.val_emb),\n",
    "                             'test': (self.test_df, self.test_size, self.test_emb)}\n",
    "\n",
    "        self.set_split('train')\n",
    "            \n",
    "    def set_split(self, split=\"train\"):\n",
    "        \"\"\" selects the splits in the dataset using a column in the dataframe \n",
    "        \n",
    "        Args:\n",
    "            split (str): one of \"train\", \"val\", or \"test\"\n",
    "        \"\"\"\n",
    "        self._target_split = split\n",
    "        self._target_df, self._target_size, self._target_emb = self._lookup_dict[split]\n",
    "        \n",
    "    def get_num_batches(self, batch_size):\n",
    "        \"\"\"Given a batch size, return the number of batches in the dataset\n",
    "        \n",
    "        Args:\n",
    "            batch_size (int)\n",
    "        Returns:\n",
    "            number of batches in the dataset\n",
    "        \"\"\"\n",
    "        if len(self) % batch_size==0:\n",
    "            return len(self) // batch_size\n",
    "        else:\n",
    "            return len(self) // batch_size + 1\n",
    "\n",
    "    def __len__(self):\n",
    "        return self._target_size\n",
    "\n",
    "    def __getitem__(self, idx):\n",
    "        emb = self._target_emb[idx].to(args.device)\n",
    "        row = self._target_df.iloc[idx]\n",
    "        label = torch.tensor(row[['mainstream', 'skepticism']].astype(int)).to(args.device)\n",
    "        \n",
    "        return {'emb': emb, \n",
    "                'label': label}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "e98c3e81",
   "metadata": {},
   "outputs": [],
   "source": [
    "def make_train_state(args):\n",
    "    return {'stop_early': False,\n",
    "            'early_stopping_step': 0,\n",
    "            'early_stopping_best_1_acc_val': 0,\n",
    "            'learning_rate': args.learning_rate,\n",
    "            'epoch_index': 0,\n",
    "            'golden_loss': [],\n",
    "            'train_loss': [],\n",
    "            'train_1_acc': [],\n",
    "            'train_acc': [],\n",
    "            'val_loss': [],\n",
    "            'val_1_acc': [],\n",
    "            'val_acc': [],\n",
    "            'val_img_txt': [],\n",
    "            'val_txt_only': [],\n",
    "            'val_unseen_img': [],\n",
    "            'test_loss': -1,\n",
    "            'test_1_acc': -1,\n",
    "            'test_acc': -1,\n",
    "            'test_img_txt': -1,\n",
    "            'test_txt_only': -1,\n",
    "            'test_unseen_img': -1,\n",
    "            'model_filename': args.model_state_file}\n",
    "\n",
    "def update_train_state(args, model, train_state):\n",
    "    \"\"\"Handle the training state updates.\n",
    "\n",
    "    Components:\n",
    "     - Early Stopping: Prevent overfitting.\n",
    "     - Model Checkpoint: Model is saved if the model is better\n",
    "\n",
    "    :param args: main arguments\n",
    "    :param model: model to train\n",
    "    :param train_state: a dictionary representing the training state values\n",
    "    :returns:\n",
    "        a new train_state\n",
    "    \"\"\"\n",
    "\n",
    "    # Save one model at least\n",
    "    if train_state['epoch_index'] == 0:\n",
    "        torch.save(model.state_dict(), args.save_dir + args.embedding_type + \"_\" + args.fusion_type + \"_\" + args.model_state_file)\n",
    "        train_state['stop_early'] = False\n",
    "\n",
    "    # Save model if performance improved\n",
    "    elif train_state['epoch_index'] >= 1:\n",
    "        acc_tm1, acc_t = train_state['val_1_acc'][-2:]\n",
    "\n",
    "        # If accuracy worsened\n",
    "        if acc_t <= train_state['early_stopping_best_1_acc_val']:\n",
    "            # Update step\n",
    "            train_state['early_stopping_step'] += 1\n",
    "        # Loss decreased\n",
    "        else:\n",
    "            # Save the best model from sklearn\n",
    "            if acc_t > train_state['early_stopping_best_1_acc_val']:\n",
    "                train_state['early_stopping_best_1_acc_val'] = acc_t\n",
    "                torch.save(model.state_dict(), args.save_dir + args.embedding_type + \"_\" + args.fusion_type + \"_\" + args.model_state_file)\n",
    "                \n",
    "            # Reset early stopping step\n",
    "            train_state['early_stopping_step'] = 0\n",
    "\n",
    "        # Stop early ?\n",
    "        train_state['stop_early'] = \\\n",
    "            train_state['early_stopping_step'] >= args.early_stopping_criteria\n",
    "\n",
    "    return train_state"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "63db575f",
   "metadata": {},
   "outputs": [],
   "source": [
    "def compute_cross_entropy(y_pred, y_target):\n",
    "    y_target = y_target.cpu().float()\n",
    "    y_pred = y_pred.cpu().float()\n",
    "    criterion = nn.BCEWithLogitsLoss()\n",
    "    return criterion(y_target, y_pred)\n",
    "\n",
    "def compute_1_accuracy(y_pred, y_target):\n",
    "    y_target_indices = y_target.max(dim=1)[1]\n",
    "    y_pred_indices = y_pred.max(dim=1)[1]\n",
    "    n_correct = torch.eq(y_pred_indices, y_target_indices).sum().item()\n",
    "    return n_correct / len(y_pred_indices) * 100\n",
    "\n",
    "def compute_k_accuracy(y_pred, y_target, k=3):\n",
    "    y_pred_indices = y_pred.topk(k, dim=1)[1]\n",
    "    y_target_indices = y_target.max(dim=1)[1]\n",
    "    n_correct = torch.tensor([y_pred_indices[i] in y_target_indices[i] for i in range(len(y_pred))]).sum().item()\n",
    "    return n_correct / len(y_pred_indices) * 100\n",
    "\n",
    "def compute_k_jaccard_index(y_pred, y_target, k=3):\n",
    "    y_target_indices = y_target.topk(k, dim=1)[1]\n",
    "    y_pred_indices = y_pred.max(dim=1)[1]\n",
    "    jaccard = torch.tensor([len(np.intersect1d(y_target_indices[i], y_pred_indices[i]))/\n",
    "                            len(np.union1d(y_target_indices[i], y_pred_indices[i]))\n",
    "                            for i in range(len(y_pred))]).sum().item()\n",
    "    return jaccard / len(y_pred_indices)\n",
    "\n",
    "def compute_jaccard_index(y_pred, y_target, k=3, multilabel=False):\n",
    "    \n",
    "    threshold = 1.0/(k+1)\n",
    "    threshold_2 = 0.5\n",
    "    \n",
    "    if multilabel:\n",
    "        y_pred_indices = y_pred.gt(threshold_2)\n",
    "    else:\n",
    "        y_pred_indices = y_pred.gt(threshold)\n",
    "    \n",
    "    y_target_indices = y_target.gt(threshold)\n",
    "        \n",
    "    jaccard = ((y_target_indices*y_pred_indices).sum(axis=1)/((y_target_indices+y_pred_indices).sum(axis=1)+1e-8)).sum().item()\n",
    "    return jaccard / len(y_pred_indices)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "6414e4bb",
   "metadata": {},
   "outputs": [],
   "source": [
    "def softmax_sensitive(T):\n",
    "    T = np.exp(T) - np.exp(0) + 1e-9\n",
    "    if len(T.shape)==1:\n",
    "        return T/T.sum()\n",
    "    return  T/(T.sum(axis=1).unsqueeze(1))\n",
    "\n",
    "def cross_entropy(pred, soft_targets):\n",
    "    logsoftmax = nn.LogSoftmax(dim=1)\n",
    "    return torch.mean(torch.sum(- soft_targets * logsoftmax(pred), 1))\n",
    "\n",
    "# convert a df to tensor to be used in pytorch\n",
    "def df_to_tensor(df):\n",
    "    device = args.device\n",
    "    return torch.from_numpy(df.values).float().to(device)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "f9b7c48a",
   "metadata": {},
   "outputs": [],
   "source": [
    "def prepare_embedding(embeddings, embedding_type, fusion_type):\n",
    "    '''\n",
    "    Args:\n",
    "        embeddings (dictionary): a dictionary of input data tensors with embeddings. \n",
    "        embedding_type (str): type of embedding to be used, can be one of {'text', 'image', 'multi'}\n",
    "        fusion_type (str): type of fusing multimodal embeddings if embedding_type=='multi', can be one of \n",
    "                        {'average', 'concat', 'complex'}'''\n",
    "    if embedding_type=='text':\n",
    "        x_embedded = embeddings['txt_emb']\n",
    "    elif embedding_type=='image':\n",
    "        x_embedded = embeddings['img_emb']\n",
    "    elif embedding_type=='multi':\n",
    "        if fusion_type=='average':\n",
    "            x_embedded = (embeddings['txt_emb']+embeddings['img_emb'])/2\n",
    "        elif fusion_type=='concat':\n",
    "            x_embedded = torch.hstack([embeddings['txt_emb'], embeddings['img_emb']])\n",
    "        elif fusion_type=='complex':\n",
    "            txt_emb = embeddings['txt_emb']\n",
    "            img_emb = embeddings['img_emb']\n",
    "            x_embedded = torch.hstack([txt_emb, img_emb, txt_emb+img_emb, txt_emb-img_emb, txt_emb*img_emb])\n",
    "        else:\n",
    "            print('fusion_type is not supported')\n",
    "    else:\n",
    "        print('embedding_type is not supported')\n",
    "        \n",
    "    return x_embedded"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "72be7dee",
   "metadata": {},
   "outputs": [],
   "source": [
    "def initialization(df, stored_data):\n",
    "    set_seed_everywhere(args.seed, args.cuda)\n",
    "    embeddings = prepare_embedding(stored_data, args.embedding_type, args.fusion_type)\n",
    "    dataset = CustomDataset(df, embeddings)\n",
    "    \n",
    "    classifier = MLPClassifier(embedding_size=args.embedding_size, \n",
    "                        hidden_dim=args.hidden_dim, \n",
    "                        embedding_type=args.embedding_type, \n",
    "                        fusion_type=args.fusion_type,\n",
    "                        num_classes=len(labels), \n",
    "                        dropout_p=args.dropout_p).to(args.device)\n",
    "    \n",
    "    return embeddings, dataset, classifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "789f15fb",
   "metadata": {},
   "outputs": [],
   "source": [
    "def write_train_state_evaluation(classifier, embeddings, train_state, df):\n",
    "    preds_all = classifier(embeddings).argmax(dim=1)\n",
    "    _, train_acc, val_acc, val_img_acc, val_txt_acc, val_unseen = evaluate_dataset(df, preds_all)\n",
    "    \n",
    "    train_state['train_acc'].append(train_acc)\n",
    "    train_state['val_acc'].append(val_acc)\n",
    "    train_state['val_img_txt'].append(val_img_acc)\n",
    "    train_state['val_txt_only'].append(val_txt_acc)\n",
    "    train_state['val_unseen_img'].append(val_unseen)\n",
    "    \n",
    "    return train_state"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "id": "fff070d9",
   "metadata": {},
   "outputs": [],
   "source": [
    "def training_loop(df, stored_data, labs=None, verbose=True):\n",
    "    \n",
    "    embeddings, dataset, classifier = initialization(df, stored_data)\n",
    "    if labs:\n",
    "        lab_embeddings = prepare_embedding(labs, args.embedding_type, args.fusion_type)\n",
    "    classifier = classifier.to(args.device)\n",
    "\n",
    "    loss_func = cross_entropy\n",
    "    optimizer = optim.Adam(classifier.parameters(), lr=args.learning_rate, weight_decay=args.l2)\n",
    "    scheduler = optim.lr_scheduler.ReduceLROnPlateau(optimizer=optimizer,\n",
    "                                               mode='min', factor=0.5,\n",
    "                                               patience=1)\n",
    "    train_state = make_train_state(args)\n",
    "    \n",
    "    print('Using {} embedding with {} fusion method'.format(args.embedding_type, args.fusion_type))\n",
    "    \n",
    "    set_seed_everywhere(args.seed, cuda=True)\n",
    "    train_state = write_train_state_evaluation(classifier, embeddings, train_state, df)\n",
    "    \n",
    "    set_seed_everywhere(args.seed, cuda=True)\n",
    "    try:\n",
    "        for epoch_index in trange(args.num_epochs, desc=\"Epoch\", disable=not args.verbose):\n",
    "            train_state['epoch_index'] = epoch_index\n",
    "\n",
    "            # Iterate over training dataset\n",
    "\n",
    "            # setup: batch generator, set loss and acc to 0, set train mode on\n",
    "            dataset.set_split('train')\n",
    "            train_dataloader = DataLoader(dataset, batch_size=args.batch_size, shuffle=True)\n",
    "            \n",
    "            running_loss = 0.0\n",
    "            running_1_acc = 0.0\n",
    "            classifier.train()\n",
    "\n",
    "            for batch_index, batch_dict in enumerate(train_dataloader):\n",
    "\n",
    "                # step 1. zero the gradients\n",
    "                optimizer.zero_grad()\n",
    "\n",
    "                # step 2. get the data compute fuzzy labels\n",
    "                X = batch_dict['emb']\n",
    "                y = batch_dict['label']\n",
    "\n",
    "                size = y.shape[0]\n",
    "\n",
    "                # step 3. compute the output\n",
    "                y_pred = classifier(X)\n",
    "\n",
    "                # step 4. compute the loss\n",
    "                loss = loss_func(y_pred, y)\n",
    "                loss_t = loss.item()\n",
    "\n",
    "                running_loss = (running_loss * batch_index * args.batch_size + loss_t * size)/(batch_index * args.batch_size + size)\n",
    "                \n",
    "                # step 5. use loss to produce gradients\n",
    "                loss.backward()\n",
    "\n",
    "                # step 6. use optimizer to take gradient step\n",
    "                optimizer.step()\n",
    "\n",
    "                # -----------------------------------------\n",
    "                # compute the accuracies\n",
    "                acc_1_t = compute_1_accuracy(y_pred, y)\n",
    "                running_1_acc = (running_1_acc * batch_index * args.batch_size + acc_1_t * size)/(batch_index * args.batch_size + size)\n",
    "            \n",
    "            if labs:\n",
    "                optimizer.zero_grad()\n",
    "                y_pred = classifier(lab_embeddings)\n",
    "                y = torch.tensor([[1,0], [0,1]]).to(args.device)\n",
    "                loss = loss_func(y_pred, y)\n",
    "                loss.backward()\n",
    "                optimizer.step()\n",
    "                train_state['golden_loss'].append(loss.item())\n",
    "            else:\n",
    "                train_state['golden_loss'].append('not implemented')\n",
    "            \n",
    "            train_state['train_loss'].append(running_loss)\n",
    "            train_state['train_1_acc'].append(running_1_acc)\n",
    "\n",
    "            # Iterate over val dataset\n",
    "\n",
    "            # setup: batch generator, set loss and acc to 0; set eval mode on\n",
    "\n",
    "            dataset.set_split('val')\n",
    "            val_dataloader = DataLoader(dataset, batch_size=args.batch_size, shuffle=True)\n",
    "\n",
    "            running_loss = 0.0\n",
    "            running_1_acc = 0.0\n",
    "            classifier.eval()\n",
    "\n",
    "            for batch_index, batch_dict in enumerate(val_dataloader):\n",
    "\n",
    "                # step 2. get the data compute fuzzy labels\n",
    "                X = batch_dict['emb']\n",
    "                y = batch_dict['label']\n",
    "                size = y.shape[0]\n",
    "\n",
    "                # step 3. compute the output\n",
    "                y_pred = classifier(X)\n",
    "\n",
    "                # step 4. compute the loss\n",
    "                loss = loss_func(y_pred, y)\n",
    "                loss_t = loss.item()\n",
    "                running_loss = (running_loss * batch_index * args.batch_size + loss_t * size)/(batch_index * args.batch_size + size)\n",
    "\n",
    "                # -----------------------------------------\n",
    "                # compute the accuracies\n",
    "                acc_1_t = compute_1_accuracy(y_pred, y)\n",
    "                running_1_acc = (running_1_acc * batch_index * args.batch_size + acc_1_t * size)/(batch_index * args.batch_size + size)\n",
    "\n",
    "            train_state['val_loss'].append(running_loss)\n",
    "            train_state['val_1_acc'].append(running_1_acc)\n",
    "\n",
    "            train_state = write_train_state_evaluation(classifier, embeddings, train_state, df)\n",
    "            \n",
    "            print('Current Running Accuracy Training Set: {:.2f}, Validation Set: {:.2f}, Golden Set: {:.2f}'.format(train_state['train_1_acc'][-1], train_state['val_1_acc'][-1], train_state['golden_loss'][-1]))\n",
    "\n",
    "            train_state = update_train_state(args=args, model=classifier, train_state=train_state)\n",
    "\n",
    "            scheduler.step(train_state['val_loss'][-1])\n",
    "\n",
    "            if train_state['stop_early']:\n",
    "                break\n",
    "\n",
    "            with open(args.save_dir + args.embedding_type + \"_\" + args.fusion_type + \"_\" + 'train_state.p', 'wb') as fp:\n",
    "                pickle.dump(train_state, fp, protocol=pickle.HIGHEST_PROTOCOL)\n",
    "    \n",
    "    except KeyboardInterrupt:\n",
    "        print(\"Exiting loop\")\n",
    "        pass\n",
    "    \n",
    "    return train_state"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "id": "1a3c2e58",
   "metadata": {},
   "outputs": [],
   "source": [
    "def experiment(df, stored_data, labs = None, hyperdict={}):\n",
    "    emb_types = ['text', 'image', 'multi']\n",
    "    fusion_types = ['average', 'concat', 'complex']\n",
    "    combi = {(a,b) if a=='multi' else (a, fusion_types[0]) for a in emb_types for b in fusion_types}\n",
    "    \n",
    "    for key in combi:\n",
    "        print(key)\n",
    "        args.embedding_type = key[0]\n",
    "        args.fusion_type = key[1]\n",
    "        if not key in hyperdict:\n",
    "            train_state = training_loop(df, stored_data, labs=labs)\n",
    "            hyperdict[key] = train_state\n",
    "            \n",
    "        with open(args.save_dir+'hyperdict.p', 'wb') as fp:\n",
    "            pickle.dump(hyperdict,fp, protocol=pickle.HIGHEST_PROTOCOL)    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "0465b2d3",
   "metadata": {},
   "outputs": [],
   "source": [
    "def demonstration(hyperdict):\n",
    "    emb_types = ['text', 'image', 'multi']\n",
    "    fusion_types = ['average', 'concat', 'complex']\n",
    "    metrics = ['train_acc', 'val_acc', 'val_img_txt', 'val_txt_only', 'val_unseen_img']\n",
    "    combi = {(a,b) if a=='multi' else (a, fusion_types[0]) for a in emb_types for b in fusion_types}\n",
    "    \n",
    "    for m in metrics:\n",
    "        print(m)\n",
    "        for key in combi:\n",
    "            train_state = hyperdict[key]\n",
    "            pd.Series(train_state[m]).plot(label=str(key))\n",
    "            plt.legend()\n",
    "        plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "76eee907",
   "metadata": {},
   "outputs": [],
   "source": [
    "def statistics(hyperdict, stored_data, df, labs=None, subfolder = ''):\n",
    "    emb_types = ['text', 'image', 'multi']\n",
    "    fusion_types = ['average', 'concat', 'complex']\n",
    "    combi = {(a,b) if a=='multi' else (a, fusion_types[0]) for a in emb_types for b in fusion_types}\n",
    "    train = df[df['split'] == 'train'].index\n",
    "    val = df[df['split'] == 'val'].index\n",
    "    results = {}\n",
    "    \n",
    "    for key in combi:\n",
    "        print(key)\n",
    "        print()\n",
    "        results[key] = {}\n",
    "        args.embedding_type = key[0]\n",
    "        args.fusion_type = key[1]\n",
    "        classifier = MLPClassifier(embedding_size=args.embedding_size, \n",
    "                        hidden_dim=args.hidden_dim, \n",
    "                        embedding_type=args.embedding_type, \n",
    "                        fusion_type=args.fusion_type,\n",
    "                        num_classes=len(labels), \n",
    "                        dropout_p=args.dropout_p).to(args.device)\n",
    "        classifier.load_state_dict(torch.load(args.save_dir + subfolder + args.embedding_type + \"_\" + args.fusion_type + \"_\" + args.model_state_file, map_location=torch.device('cpu')))\n",
    "        classifier.eval()\n",
    "        y = F.softmax(classifier(prepare_embedding(stored_data, args.embedding_type, args.fusion_type)), dim=1)\n",
    "        pd.DataFrame(y.cpu().detach().numpy()).to_csv(args.save_dir + subfolder + args.embedding_type + \"_\" + args.fusion_type + \"_\" +'prediction.csv')\n",
    "        \n",
    "        conf, pred = y.max(dim=1)\n",
    "        pd.Series(conf.cpu().detach().numpy()).hist()\n",
    "        \n",
    "        results[key]['performance']={}\n",
    "        \n",
    "        print('Validation Set:')\n",
    "        results[key]['performance']['val'] = evaluate_dataset(df, pred)\n",
    "        \n",
    "        print('Test Set:')\n",
    "        results[key]['performance']['test'] = evaluate_dataset(df, pred, val_set='test')\n",
    "        \n",
    "        if labs:\n",
    "            print('Prediction for Golden Sentences:')\n",
    "            print(F.softmax(classifier(prepare_embedding(labs, args.embedding_type, args.fusion_type)), dim=1).cpu().detach().numpy())\n",
    "            results[key]['golden'] = F.softmax(classifier(prepare_embedding(labs, args.embedding_type, args.fusion_type)), dim=1).cpu().detach().numpy()\n",
    "            print()\n",
    "        \n",
    "        print('Full Dataset:')\n",
    "        print(pd.Series(conf.cpu().detach().numpy()).describe())\n",
    "        print('Training Set:')\n",
    "        print(pd.Series(conf[train].cpu().detach().numpy()).describe())\n",
    "        print('Validation Set:')\n",
    "        print(pd.Series(conf[val].cpu().detach().numpy()).describe())\n",
    "        plt.show()\n",
    "        \n",
    "    return results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "id": "731aeca7",
   "metadata": {},
   "outputs": [],
   "source": [
    "def statistics_F1(hyperdict, stored_data, df, labs=None, subfolder = ''):\n",
    "    emb_types = ['text', 'image', 'multi']\n",
    "    fusion_types = ['average', 'concat', 'complex']\n",
    "    combi = {(a,b) if a=='multi' else (a, fusion_types[0]) for a in emb_types for b in fusion_types}\n",
    "    train = df[df['split'] == 'train'].index\n",
    "    val = df[df['split'] == 'val'].index\n",
    "    results = {}\n",
    "    \n",
    "    for key in combi:\n",
    "        print(key)\n",
    "        print()\n",
    "        results[key] = {}\n",
    "        args.embedding_type = key[0]\n",
    "        args.fusion_type = key[1]\n",
    "        classifier = MLPClassifier(embedding_size=args.embedding_size, \n",
    "                        hidden_dim=args.hidden_dim, \n",
    "                        embedding_type=args.embedding_type, \n",
    "                        fusion_type=args.fusion_type,\n",
    "                        num_classes=len(labels), \n",
    "                        dropout_p=args.dropout_p).to(args.device)\n",
    "        classifier.load_state_dict(torch.load(args.save_dir + subfolder + args.embedding_type + \"_\" + args.fusion_type + \"_\" + args.model_state_file, map_location=torch.device('cpu')))\n",
    "        classifier.eval()\n",
    "        y = F.softmax(classifier(prepare_embedding(stored_data, args.embedding_type, args.fusion_type)), dim=1)\n",
    "        #pd.DataFrame(y.cpu().detach().numpy()).to_csv(args.result_dir + subfolder + args.embedding_type + \"_\" + args.fusion_type + \"_\" +'prediction.csv')\n",
    "        \n",
    "        conf, pred = y.max(dim=1)\n",
    "        pd.Series(conf.cpu().detach().numpy()).hist()\n",
    "        \n",
    "        results[key]['performance']={}\n",
    "        \n",
    "        print('Validation Set:')\n",
    "        results[key]['performance']['val'] = evaluate_dataset_F1(df, pred)\n",
    "        \n",
    "        print('Test Set:')\n",
    "        results[key]['performance']['test'] = evaluate_dataset_F1(df, pred, val_set='test')\n",
    "        \n",
    "        if labs:\n",
    "            print('Prediction for Golden Sentences:')\n",
    "            print(F.softmax(classifier(prepare_embedding(labs, args.embedding_type, args.fusion_type)), dim=1).cpu().detach().numpy())\n",
    "            results[key]['golden'] = F.softmax(classifier(prepare_embedding(labs, args.embedding_type, args.fusion_type)), dim=1).cpu().detach().numpy()\n",
    "            print()\n",
    "        \n",
    "        print('Full Dataset:')\n",
    "        print(pd.Series(conf.cpu().detach().numpy()).describe())\n",
    "        print('Training Set:')\n",
    "        print(pd.Series(conf[train].cpu().detach().numpy()).describe())\n",
    "        print('Validation Set:')\n",
    "        print(pd.Series(conf[val].cpu().detach().numpy()).describe())\n",
    "        plt.show()\n",
    "        \n",
    "    return results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "3f9528cd",
   "metadata": {},
   "outputs": [],
   "source": [
    "hyperdict = {}\n",
    "experiment(df, stored_data, hyperdict)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "396fbf8c",
   "metadata": {},
   "outputs": [],
   "source": [
    "hyperdict = {}\n",
    "experiment(df, stored_data, labs, hyperdict)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "efffe68d",
   "metadata": {},
   "source": [
    "## Performance on Trained Models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "id": "f6295688",
   "metadata": {},
   "outputs": [],
   "source": [
    "perform_results = {}\n",
    "for o in ['O1', 'O2']:\n",
    "    perform_results[o] = {}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "id": "52cde508",
   "metadata": {},
   "outputs": [],
   "source": [
    "with open(args.save_dir+'001/hyperdict.p', 'rb') as fp:\n",
    "    hyperdict = pickle.load(fp)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "id": "52c83fcf",
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train_acc\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": [
      "val_acc\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAh8AAAGdCAYAAACyzRGfAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjcuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/bCgiHAAAACXBIWXMAAA9hAAAPYQGoP6dpAACV1UlEQVR4nOzdd3hUZd7G8e/0ZNJ7gZCEEpqAdEGlWRCkKHZxFeuquyq6WHhdFVYEG1h3dXVdULGvDRHEAthAqSqCQgiEEtJ7ZiZTz/vHSSaZNJKQZIL5fa5rrplT5pxnhnLuedrRKIqiIIQQQgjRQbT+LoAQQgghuhYJH0IIIYToUBI+hBBCCNGhJHwIIYQQokNJ+BBCCCFEh5LwIYQQQogOJeFDCCGEEB1KwocQQgghOpTe3wWoy+PxcOzYMUJCQtBoNP4ujhBCCCGaQVEUysvLSUxMRKttum6j04WPY8eOkZSU5O9iCCGEEKIVjhw5Qvfu3Zvcp9OFj5CQEEAtfGhoqJ9LI4QQQojmKCsrIykpyXsdb0qnCx/VTS2hoaESPoQQQoiTTHO6TLSow2lKSgoajabe4y9/+QsAc+bMqbfttNNOa13phRBCCPGH1KKaj61bt+J2u73Lv/76K+eccw6XXHKJd915553H8uXLvctGo7ENiimEEEKIP4oWhY+YmBif5UcffZRevXoxfvx47zqTyUR8fHzblE4IIYQQfzitnufD4XCwcuVKrrvuOp/2nY0bNxIbG0taWho33ngjeXl5bVJQIYQQQvwxtLrD6UcffURJSQlz5szxrpsyZQqXXHIJycnJHDx4kAceeIBJkyaxfft2TCZTg8ex2+3Y7XbvcllZWWuLJIQQQoiTgEZRFKU1b5w8eTJGo5FPPvmk0X2ys7NJTk7m7bffZtasWQ3us2DBAhYuXFhvfWlpqYx2EUIIIU4SZWVlhIWFNev63apml0OHDvHll19yww03NLlfQkICycnJpKenN7rP/PnzKS0t9T6OHDnSmiIJIYQQ4iTRqmaX5cuXExsby/nnn9/kfoWFhRw5coSEhIRG9zGZTI02yQghhBDij6fFNR8ej4fly5dzzTXXoNfXZJeKigrmzZvH5s2byczMZOPGjUyfPp3o6GguvPDCNi20EEIIIU5eLa75+PLLLzl8+DDXXXedz3qdTseuXbt47bXXKCkpISEhgYkTJ/LOO+80a6pVIYQQQnQNre5w2l5a0mFFCCGEEJ1Du3c4FUIIIYRorU53YzkhhBCdj6IoOA8domLTJtyFRQQMOgXz0KHowsL8XTRxEpLwIYQQokGu4mKsP/yAZdMmLN9vwnnsWL19TH16Ezh0GIHDhmIeNgxDUlKz7moqujYJH0IIIQDwOBzYduzE8v33WDZtonLPHqjdLdBgwDx0KIaEBGw//4wjMxN7+n7s6fspefddAHQx0ZhrhZGA/v3RGAx++kSis5LwIYQQXZSiKNj3pas1G5s2Yd22DcVm89nH1KcPQWPHEnT6WMwjRqA1m73bXEVF2HbuxLp9B7YdO7Dt3o07v4Dyzz+n/PPPAdAEBBA4eLA3jASeeiq6dhxM4KmsxJWfjysvT31UvXaXloJGC1oNGq226rUWjVajvtZo6mxr7HWt91S/NhjQGo1oTCY0RhMakxGtyeRd1pqqtplMaIzGmn31J3YJVhQFnE48DieKw45iVx8eu8O77LHbUaqXHQ7vMopC5J+uaqNvveVktIsQQrQxRVFQrFbcFRY8lgo8FerDXVGBx2JVl6vWo9Ggi4pCHx2DPjoafXQU+uhotGFh7dJ84czL84YNy+bNuPMLfLbrYqIJHjuWoLFjMY8ZgyE2ttnH9lRWUvnrr1h37FTDyM6d6kW/No0GU+/eBA4fpoaRYcMwdOt23M/qEyrqhAund10+npPp/mA6HRqTqSaMmKqCilENKRq9HsXpxOOoChDV4cJR85pWXsI1JhP9fv6pTT9OS67fEj6EEKIJHquViu++w11c4g0N7ooKPBWWqmX12W3xXdfai4KXwYA+KqoqkESji65+7RtSdNExaIPMjV68PVYr1m3bsHyvBg57ndtdaAICMI8c6a3dMPXp02ahR/F4cBw4gHXHDmw7dmLduQPnocP19tPHxBA4bBiBQ09Fo9Xhyq8TLFoYKjQBAehjY9HHxqCPicEQG4suPNxbJjwKKB6f1ygKikcBj6dqW+3XDbzH40FRPOD2oLhcahhwVNU6eEOCHcXh9AkNOJ1t8t02+LkNhpoaFpMRrdFUE2pqBRy1ZiaAxEeXtOn5JXwIIcQJcubmUbxyJcXvvoun7q/35tLp0AYFoQ0OQhcUjDa4+hGENkhdpyge3AWFuAoKcBWqzy09nyYgoF5I0YWEYPv5F2w7d6LUvuBpNAQMHKiGjbFjCRw2FK3R2LrP1wquggKfMFK5ew+4XM16r8ZkqgoVvsFCHxuLPibGu00bHNxpO70qbjdKVc1Fo80jVcu4XGoNSIPNN3WWDQa1aciPJHwIIUQrVf7+O0XLV1C6Zo33V6qhe3dMffuiCw5CGxRcFSjUEKGrDhTe9TXrNAEBrboIehwO3FVBxJVfgKsgX13OL6gVUvJx5xfgsVqPezxDYiJBp1c1pZx2GvqIiBaXqb14bDZsu3Zh27ET265daAyGRoOFNiSk04YKIeFDCCFaRPF4sHz7LYUrVmDd/IN3feCI4UTNmUPwxIlodDo/lrBxHqtVDSP5BbgKC3BXBRZ3SQnG3r0IHjsWQ3KyXLRFu2vJ9VtGuwghuixPZSWlq1ZR9OprODIy1JU6HaGTzyVyzhwCBw/2bwGbQWs2YzSbMSYl+bsoQjSbhA8hRJfjKiyk+M23KH7rLdxFRQBog4IIv/RSIq+ajaFbNz+XUIg/NgkfQpyE3KWllH+1HmdWFmEXzJRfvc1kz8igaMUKSj9eheJwAKBPTCDy6qsJv/hidMHBfi6hEF2DhA8hThLukhLKv1pP2brPsGz+wdsZsvC//yV27h1EXHVVp+2X4E+KomD94QcKly/H8s233vUBgwYRde0cQs4994QnexJCtIz8ixOiE3MVF1Oxfj1ln63Dsnmzz5BEU58+aM1mbD//TO6SRylds4bERYsw9enjxxJ3HorDQemaNRSteBX777+rKzUaQs4+S+3PMWyYdMIUwk8kfAjRybiKiyn/8kvKP1uH5ccffQNHWhoh500m9LzzMPXsieLxUPLue+Q98QSVP//CgVkXEX3zn4m+8UY0HTh3Q2fiLimh+J13KV65Eld+PgCawEDCZ80i8uo/YUxO9nMJhRAy1FaITsBVXEz5F1/UBA6327vN1LcvoedNJmTyeZh6pjb4fmdODjkLFlKxcaP6nj59SHhk0UkxWqM1FLcbj82mTlVutaDYbLjLKyhft46SDz/03p9EHxtLxFVXEXHpJd4ZLoUQ7UPm+RDiJOAqKqL8iy8pX/cZlh+3+AaO/v0JnTyZkMnnYkptOHDUpSgKZWvWkLvoEdzFxaDVEnn11cTccTvawMD2+hgtorhcVP72O57yMjxWa83DUvVss+GxWrzrlXrbqtZXVjZ5HlO/fkRdO4fQKVM6bw2QpQA2PQcF+yC0G4T3gPCkqudkMEepNzsT4iQh4UOITspVWEj5F19Stu4zrFu2+gaOAf0JnXweoZPPxZiS0vpzFBeTu3gJZZ98AoAhKYmEh/9B0GmnnWjxW82Zm0vJe/+j5L33cOXmtt2BtVq0ZrP3YezVi8g/XYV59OjO25+jshQ2PQ8//AscFY3vZzBDWFKtQNKjajlZXRcUC36eTlu0I0WB8hwIigHdydFDQsKH8BvF5aLkgw9wHMwk6vrr0EdH+7tIfqcGji8o+2wd1i1b1JtVVQkYMICQ86oCRxv3RSjfuJGcBQtx5eQAEH7JxcTefXe73s68NsXjwbJ5MyVvv035+g3eoKUNDcUQH+8NDBpzYK0AEeQTJrQ+28xoqvcJMqMNDFTvadFZQ0ZdDitsfRm+ewpsxeq6hFNhyOVQkQclh9VH6REozz7+8XQmCOteE0zCq4JJWFVYCYkHbRcf/eRxQ0UulB4FjQ66DTs5apMKM+DDP8PRraAPhITB6t+VxFMhcShEp3XKP1sJH8IvLD/8SO7ixdj37QPUi0zcPXcTdtFFJ88Foo3YDxykYv1XlH+1HttPP/nc4TRg4EC10+jkyRh79GjXcrgrKshbupSSt94G1LuHxi94iJCzzmq3c7qKiyn94EOK333H5w6mgSOGE3H5FYSce06H3sjM71wO2PkafP0EVKhBkOg0mPR36D+j4Yuhy65eMGsHkpLDUFL1XH5MvRNrU3RG6DkRhl8DfSafNL+em01RoLIESrPU76rsqPpcerRmXfkx8NS6aV3vs2Ha02pQ64wUBbYvh3X3g7OJe/YYzBA/SA0i1aGkEwQSCR+iQzmOHiXvsccp/+ILALRhYRji4rwhxDxyJPELFzbaWfKPQPF4sP30szdwOA4e9NkeMGiQ2mn03HP9MiGYdetWsv/+AI5DhwAImXIe8fff32Y1U4qiYNv5E8Vvv0X5Z+u8E3hpg4MJmzmTiMsv63pDgD1u2PU/2LgYijPVdWE9YOJ8GHzZiV0o3E4oy/INJN6Qcki9+Co1TXoEx8OpV8KwP0FkzxP6WB3GWal+xrKsWoHiSFXQqFrXVLNVNY0OQhPV2iW3HYzBcM4/YPi1navZqjwXVv0V0j9Xl1PHwcx/qt9D9k9w7Cc4thNyfmn4cxvMED+4pnYk4VSI7tOhgUTCh+gQHouFgpdepmj5cvVio9MRcfnlRP/1L+hCQih6fSX5zz6LYrOhMRiIuuVmom+4ofN2AGwhT2Ullk2bKV//FRUbNuIuLKzZaDAQNGoUwWdNImTSJAzx8f4raBVPZSUF//wnhf9dDm43urAw4v5vPqEzZrS6ZspdYaHsk1UUv/0O9r17vesDBgwg/IrLCZs6FW1QUFt9hJODosDeNbB+EeTtUdcFxcK4u9VaCL2p/cvgdkFhOvz0pvqwFtRsSx0Pw66G/tM7pizNUZGnXnQzNkBRhho0LHnNe685Sm1+Cu2uPod1q3pOUjvyVjc/5e9TL+5HflTfl3ImzHi2c4SxPavgkzvAVqQ2p529AEbf3HA48rjVZpljO2tCSfbP4LTU39cQVL/JJqp3uwUSCR+iXSkeD2WffELek0u98yiYTzuNuP+bT0Bams++jqNZ5CxciOVbdWZJY+9eJPzjH5iHDevwcrcFV3ExFRs2Ur7+Kyzfb/IO6QTQhoQQPG4cIWdNIujMM9GFhPixpI2z7d5N9t8fwP7bbwAEnXkmCQseatH9TCp//53it9+mbNUn3lu6a0wmQqdOJeKKywkYNKjLNbUBcGAjfPUPyNquLgeEwelzYfSfweinEOZyqGFox6vqxZ2q//IDI2HIFWoQie3XsWXyeCB7J+z7HNLXqRfShugDq4JEdahIqgoaVa9DE8FobsF53bDlZfhqodqsoQ+Esx5U/3z80WRRWQpr74Of31SX4wfBrJchtn/LjuNxQ+F+9Xs89pMaSrJ/brjppjqQJA6Fif8Hprb7f0rCh2g3tl9+IfeRxdh+/hlQR1LE3XsPwWed1ejFxjsEdPESb+1A+GWXEfu3uzqs8+OJcBw6RPlX6ylf/xW2HTt9OozqExIImTSJkLMmYR4x4qSp1VGcTgr/u5yCf/4TxeFAazYT87e7iLjiCjSNVEV77HbK1q6l5O131H4sVYypqURcfhlhF1yALiysgz5BJ3N0mxo6Dn6tLhvMcNqtMPY2CAz3a9F8FB+CnSvVR/mxmvVJo2HYNTDwgvYLSZVlcGBDVeD4vH7NRsKpkDZZfa4OGoER7dNBtOggrLoNMqum2+8+Sm3iiElr+n1tKfM7+PAWKD0MGq0aUifMB30b/R/icUNBelXtSFUoyfmlJpAYzDD/aJuGLgkfos058/LIX/YUpR99BIDGbCb65puJvOZqtKbmVd26S0rIffJJSv/3PgC6mGji7/87IZPP7VS/khWPh8pdu7yBw7E/w2e7qX9/b+Aw9e/fqcreUvYDB8j++wPYduwAIHDYMBIWPYypZ01VtCMzk+J33qX0gw9wl5aqK/V6Qs4+m4jLL8c8epT/vgOPB0oyIXeP2sThcasXkOi+avWyIaB9z5+7R21e2fupuqwzwojr4My/QXBs+567ER6Pgq3MgbXMQWCIgaDwBkYEuV2Q8RVsfxX2fVbTP8QUCoMuVoNI4qknVhBFUX+N71un1m4c2gweZ812YzD0mqh2hu1zjto80pEUBbavgM8fAEe52twx4V4Ye0f7ds512WH9w+pwaxSISIEL/w09OmAovMetzitz7Ce1iWfMX9r08BI+RJvx2O0UvfoahS++6K1eD7vgAmLuvBNDXOv+c7X8uIWchx7CkZkJQPDEicQ/+ACGhIS2KnaLeex2rD/8QPlX66nYsMHbnASAXo955AhCJp1FyKSJf7jbrSseD8Vvv03+k0vxWK1oDAai//IXjKmplLzzNpZNm7376hMTiLj0UsIvugh9TEzHFtRaBLm71ZCR+2tV4Pit4bZuUH9NRqRATD91JEBMv6pgknbiVc1FB2Djo/DLu4CinmvIlerFK7z9RjC5nR4spXYqSuxYiu1UFNuxlNipKKn0vraUOlA8Nf+t641awmLNhMcGEh5rVl/HqcsBwQY05Tlqtf+O12o6xgIkDFGbZAZdojYfNYfLrv6iT/9cDR3Fvh2vieqtho20c6HH2Lb7lX8iSo/CJ3Nhv9phnoQhai1I/KC2P1fOr/DBTZC3W10edg1MfqRNmz78ScKHOGGKolDx1VfkPvY4ziNHAAgYMpj4++9vkym7PXY7hf/+NwUv/wecTrXaf+5cImZf2WF3ZnUVFFDx9deUb9iAZdNmFGtN+6g2KIigcWcSMuksgsed6Z/mBLdTnQWzIlftkGfJq3ld/XDZIPl06DtFrTo+gV9szmPHyH5ogbd/jpdGQ9C4M4m4/HKCx41r/z8flx3y91aFjN01gaOxuS90JojpC3EDazoWFuxV29MbE9pdfU/1I7rq2RzZdNnKjsE3T6gX6uohnAMugIn3n3CVvaPSVRUk6gYLOxXFlVhK7NjKncc/EGpLRUCIkcoKp08Qqctk1hMWE+gNI2EcIjx3NeFH38OoVH1/BjMMvFANIkmj6zeDlB2rChufq31eaodBrQFSzlCbU/qcC1G9WvitdBBFgV/egbX3qsN3tXq19urMeW0TkDxu2Py8WkvmdqgTh814DvpOwe3yYK2qqaqusap+OGwuAkIMmEOMmMOMmEOrHyYCgg1otZ2r1lXChzghlfv2kbtkCdbNPwDq/TFi5/2N0GnTGu0P0Fr2/fvJfvAhb7V/wKBBJDz8DwL6tX0HOEVRsO/bR8WGDZRv2EDlL7t85t/Qx8URPGkiIZPOwjx6VPvMReHxgLWwgSCRW/NsyVefrYXHP15tgRHqf/Bp50Hvs5r/a7UWRVEo++QTchcvAZ2O8IsuIvzSSzB2797iYzXjZOrQydoBI3e32k5de5hobeHJasiIHaA+xw2EyF71Q5eiqN9h/l71UbC35nVToyiCYurXlMT0U5tTvlumdlZ0VU3t3vtsmPRAi5sn3G4PBUcqyMkoJXt/CUU5Viwldhw21/HfDOj0WoIiTASHmwgKNxEcUfMcHB5AULgJc6gBrU6L2+2hvKCSkjwrJblWSvNs3tcVxfYmzxMY4CJce4RwJYMwXTbh+mOEx5gIG3Ue+qQhkLEeZd/nuLL34sKISzGpj4BEnN1Ox5UwGlfUKbgUIy6nB5fDjcvhwVn1rC5XvXbWLGu0GnR6LXqDFl3VQ6/XojPo1GW9Rt2mV5e9++l9X3u36Wut02vQ6rXodBo0Wk395qjyXPj0Lvh9tbocOwBmPg/dhrfoz9jj9mCrcKohIuso1q9fwZZfgNUTjjVoINbQwVgtCtYyB3Zr8/7c66oOmOZQI0GhRgJDa4eTmpBiDjViCtJ3SNOohA/RKu6SEvKffY7it98GjweN0UjkddcSfeON7Tpc0ntn1iefxFNRATodUdddS/Stt57wPUk8DgfWH7eogWPjBlzHfH89BwwcSPDEiQRPnEDAgAFt/w+09Ch89bBazVqRrwaLxi6sDdHo1AticGzVI059DqpaVjyQ/oX6y7OypOZ9WgMkj4W+U6HveWrzQwtUD51us1oOl0MNF9k/QfYvNWHDXtbw/gFhEHeKb8iI7Q+mEDxu9WLldnlwO6teVy0bTDpCYwLR6ZoIydYitd27bjApPdLEB9DgHSXSY4waOlJOb9ZHt9tc5BxQg0ZORim5mWW4HA1PEGYM0BEUEVATKBoIF211IXE63JTl2+oHkzwbtjJHk+81aSpwKQbcdJKhuq2hAZ1Oi1avqf/sqkBbcRStYkeHC214ArroVLQGvRpgdLWCjFaDo9KNtcxeU4NR4fT+dWkOrVaDOcxIYO0ajhAjxkA9ldUhptyBtdSBtcze8uPrNJhDax2/KrSMnJ7a9L+VFpLwIVpEcbkofucdCp59ztuhMOScc4i95+4OnRDLmZtH7uLFlK9bB6gjaeIXPETw6c37T76aq7CQio1fU7FxI5bvv/f2VQF1OGjQmDFq4JgwodX9Vo5LUdT5FT67r+ELrDmqfpCoXva+jlOHQzantsntUucv2LcW9n6mzvFQW0x/teq77xToPrJ9hxW6HJD/W82kSNk/Qe5uFJeTbGd/jtiH4FQCcCsGXATgNkXiNkbi1ofh0ofg1gTi9uhxuRTcTrcaMFzVAUNpshkB1P9ow2LNRCaYiYgPIjIhiIgEtZ+D3tDE57ZXqN9b/l7fYFJ0QA158YPgrIfUGo8mRnaVF1aSnVFKdkYpORklFB6z1LtQmMx6EnqFEd8rjJgeIYREqjUWxoDOMQupw+aqCiJVoSTXSklOOaU55dgdDX+H1TUNeqMOvVF9Nhirl2utq9pHZ9RiqLVeb9SiN+hQFKVeqPS+rv13ofa2BoKoq+rvTvV7PK6OvdRp8BCgLcWsLcFsBnOvQZhjIhuumTDr0bSgCcWnZqV2c01pVUgps3tf2y0N16zo9Fr+/Nz4Nv3BJeFDNJtl82Z1SvT0/QCY0tKI+7/5fr0JWfn69eT842HvPUnCZs4g9t570Uc23B6vNqekU7FhAxUbNmD75Rff5pSYGIInTCB44kSCxpzW/nd4Lc9VJwzat1Zd7j5SbT8OSVADRVA06AztW4bCDNi7Vh3JcGiTb22LOcq3eeZEOru5nWqnz9oTHuX+qrZrVylwJrOvchzpleOpcEe1/lwN0Oo0PtXrlRZno7UKGg2ERAeqYSTeXPWsBpMmL/ouu1pjFZJYLwh63B4KjlaoYWO/GjYspfVrDUJjAkmsChsJvcKJiDe36GLTWSiKQqXFia3cid4bHtTmj87W/6AuRVHwuBXcLk/9Z5eCx6OGW4/Lg9td6/nwdtzbXsdjq8Ct6PEkT8Dd6zw8GoP3GAaTrqZGoXgb5m/nE1B5EK1OB5Puh7G3g1aH2+Om3FFOqaOUMnsZpY5SSu3qo8xRVu+50lVJjDmGhKAE4oPi1Yc5noTgBGLNsRi0x/9/xO30YC13YPPWnKgPt8vD6BltO8GahI8/CI/NxrH/+z8sX39TUwWu16Op+1qvA51e7Y/RwHb0OjS6+q9d+flYNm0CQBcWRszcOwi/5BI0ev//+nJXWMh/5hmKV64ERUEXHk7svfcSdsFMNBqN2pyyZasaODZuxJmV5fN+04D+hEyYSPDEiQQMHNDmfVUa9ev78Onf1BuHaQ3qJD5jb2+XoXuKR8HpcB//17KtGPZ/pYaR/V/4dsSs7hDYd4oaRiKauLldddCoPdVz7m51yuo6yvQ9SddeyL6SUykqC/auNwboSBkSTXC4qVa7vK5WO70GfVVbft02+7rt+Tp9/Que4lEoL66kONtKcY6FomyL93VTbevBESYi6oSSyIQgAoJ9/3N32FzkHCz1ho3czDJcdt9mNK1WQ3SPEBJ6h6m1Gz3DCAo7iZsnTgLVF/XqC3dDF3Wby4ZWo0Wn0aHX6tFpdOi0unrLeo2+/nqXHd3uD9FnfodWUdAHxaAbdRO6hFPRaXXYXDbKLHmU7XqX0uztlGq1lJojKIvrRyketRz2Msqd5W32mTVoiAmMIT64KpAEJZAQnEC8Od67LjIgskOHwUv4+APw2O0cvfUvWL7/vn1PpNMRceWVxPzlVnTh4e17rpbyeLBt30z2wkXY92cCYB6Ygi62G5YtP+Gx1PSq1xiNmMecRkh1c0pHT2duKYQ1f4PdH6rL8YPhwhfVvgonyFHpojTPRnGuheIctX2+OMdKaa4Vl9NDeJyZxLRwuvUJJ7FPBMERTVzo3E44/INaI7J3rTqVdW2xA9QQ0ncKGAJrZks8tlMdJthA0MAUBolDqIwcwX7LcPZlRpCdWbOfVq8h5ZRo+oyMI2VQFHpjx88kqShq577ibPU7LM62UJSjBhNrE/0bAkMMRMQHERIZQEFWBUVZFdT9H9Nk1hPfs7pWI4zYlFAMHfgZFUXB4rSQZ8ujwFpAni2PfGs+xZXFRAVGkRqWSkpoConBiei1/v9h0RSH20GJvcQnNDRUK1AdMKprD8odbXdR7whBhiDCjGGEmcIINYYSagolzBTmsy7MFIZRZyTPmkeOJYdsSzY5lhzvw+Fpul8OgEln8taWVNecVNeiJAQlkBKWglYjfT4ACR+gdvY7etvtVHz9NZrAQLo//RSGHj3A7UZxu1FcLvW1yw1uV9W62q+rtrs96jqXG8Vd857q1wDBEyZg6tUBw988HrCXqp39rEXqSI7qh616uc42WxEoHhQPFO4NpuDXEBR3TYrXRYYTPOksQiZOJGjMGLTmFkyz3JZ+X6M2s1jy1A6i4+a1eIieoihUFNspybFSnGulJMeiPjdjVEJdYTGBNWEkLYKQyCYm2ipIr2meObz5+HdKNYXVTM2ceCrOqMFkHg1h39Y8Du8uxOOu+u9EA936hJM2Kp6eQ2MICGrnZqYTUGlx1gkkaigpL6pscP/Q6AASeoV7w0ZkQlC7NaFYnBbyrGqYyLflk2/Nrxcy8m352Fy24x7LoDXQI6SHGkbCUryhJCUshVBjx/xf61E85FvzOVpxlKyKLI6W1zwfLT9Knq2Z93NphFlvVi/iVRfyUFOo90Ju1pvx4MHtceNW3Lg9blyKy7vs8rgaXO99rt7mtuMuzsRdkYtLo8Gl0xPochLm8RCqCyAsdSJhMf19QkT1c5gpjBBjSLOaS5qiKApFlUU+oSTbkk22JZtcSy7ZlmwKbAUoTfRK1aBh+1XbMbRhE7CEj5OY4nRy9M47qfjyKzQmE0n//jdBp432d7Gaz+1U50DIWF8TJGxVoaIlozxqM4aAORKHI4yinyrRVmYT0q2SgEgnmv7T1It94tC2/RzNUVkKn82Hn95Ql2P6wQUvQLfG71vjdLirOu/VDxmN9VUA9Rd4eJyZiDgz4XFq80B4nBmTWU/OgVKy0ks4tq+EgiPl9X6Zh0QF0C1NrRXplhZOSFRAw1Wx1iLY/2VV88yX6rqEIepQ0oRT1e84IhWPAkd/L2bfllwO/JSPs1azQ3RSMH1GxpE2Mo7giHaeXbSdOSpdai1TtoXyokrC44JI6N22TSiZpZnsKdxDvi1fDRlVAaN6uTmholqwIZgYcwyxgbFEm6OJMEWQb8vnYOlBDpUdwt5QrVWVqIComlASWvOcGJyIroWdk8sd5T7B4kj5Ee/ysYpjx/3FrtVoG60VCDWFerf5bDeqr9vyQnpcmd+rN6orOqAuD7kCpjzWqiHu7cHpdpJrzfWpMakdVNyKm1UXrGrTc0r4OEkpLhdZ8+6m/LPP0BiNdP/Xvwg+o2UjPfxGUeC3T+DLBfWr8mszBqsTOZmj1JEc5qhaj8iabdWPwIj6d97M2QXfPAl7PsY7jKD3OepdQ3t0UFDLWA8f/1W9tTca9R4eE+/3TuddPeoh52ApeQfLKcquoDjXSkVR4xcArVZDaEygN1hExKujNcLjzM2uObDbXGTvV4NIVnoJ+YfL640OCY4wqTUjaREk9gknLCawfhip/m+har2iKORllrNvSw7p2/N8hmKGRAWQNjKOtFHxRCZ2sTvYtkKpvZTPDn7GqoxV/FLwy3H3DzYEEx0YTaw5lhhzDDGB6iPWHOtdHx0YjdnQeM2fR/GQbckmszSTg6UHySyrei7NbLK2wag10iO0h7eWpPo51BRar+YiqyKLoxVHKbU3MbkboNPoiA+Kp3tId7oHd6d7SHe6BXeje3B3uoV0I9wU3qZNAe3KYYXty9WZW9Mm+7s0fifh4ySkuN0cu/c+ylavBoOBpOefI3j8eH8Xq3kO/6DeH+HoFnXZHA2n3aLeqrpWsFACInBhxFnpxmFz4ah04ah67ax+XfXstNVZrrU9ONxEdPdgosKtROd9QNThVzFpqtp8U8epISTlzPa5IZW9Ar54ELa9oi5HpMKFL2KPGUFeZhm5B0vJPVhGbmZZo7NRmoL0RNSqvah+Pu78FK3gqHSRk1FdM1JMXmY5njphJCjMSGJVEOmWFk54nNkbRkpyrezdkkP6llxK82t+hQcEGeg9PJa0UXHE9wo7qe9v0xGcHiffZ33PqoxVbDyyEWfVPU50Gh1DYoYQHxTvEyZiAmO8QaOpUNEWKhwVHCo7xIHSAzWhpCyTQ6WHmtWvoCGRAZE+gaL2c3xQfKfveyJaR8LHSUbxeMi+/++Ufvgh6PV0f+ZpQs46y9/FOr6CdLWmo3o2QIOZ8sF38Zt7OrlH7FWBwYXDVhMijjdHQ2uFBlqIVn4lSn+AaH0m0T3CCTn7z2j6nNV2IeTQJvjoFjxFhyl09SC32/XkBk4k95DadFK3eVWr0xDdPZi41DCiuwcTHq8GjcBg/93Pwml3k3OglGPpJWTtKyY3s6ze/AeBoUYSe4dRXlhJ3qGajnx6g5bUU2NIGxVH0oDINg9KfzSKovB70e+syljFmoNrKKos8m7rG9GXGb1mMLXnVKIDo/1Yysa5PW6yLdm+NSVlmWSWZlLuKCcxONFbe9EtuFtNDUZId4IMUgPWFUn4OIkoikLOQwsoefdd0OnotnQpoed18uq7ijz4+jHYthwUN24MZCb+jT2WSRzeV39CpXo0YDTpMAbqMQToMQaor40mHYbAquUAvfoI1GGotaw3aikrqKQwq4KCI+UUHK1otEOmUWMhOqiAqN7diB44kOgeIUQmBjU90VRDHze/hNxPVpC7O4NcRxp5rt64lPrt/qHRAcSlhBKXGkZcaijRScEtPldHcznc5Bws49i+Yo6ll5BzoAy3q6bviUarIal/JGmj4kgdEn3Ck2ApikKOJQery4rdbcfhdnifq1/7rPf4rqu9f0PPYaYwTok+hcHRgxkUM4hYc8ffWbbAVsCnBz7l44yPSS+umewtMiCS83uez8xeM+kb2bfDyyVEe5PwcZJQFIXcRY9Q/MYboNGQ+PjjhE2f5u9iNc5hgc3/hO+fAUcFJa4E9gTcxO/Fw7BZai5Y3fqG03t4HIEhBp8QYQzQYwjQYTDq2nR0QKXFScHRCgqPVgWSw8UUZdvwKPV/mWu0EBEfRFS3YKK7BxOdFEx09xDMoWpthNPuJv9wGTkHy9Tmk/0FWMrr/xMxBuiITQklLrUqbKSEeo9xMnM53eRllpGdUYoxQE+vYbFt8rmyKrJYnbGa1QdWk1mWeeIFbab4oHgGRQ9icPRgBscMpn9UfwL1bT/JnN1tZ8PhDazKWMWmY5twV3WuNmgNTEyayMzeMxmTOOaERzkI0ZlJ+DgJKIpC3mOPU7RiBWg0JCxeTPiFF/i7WA1zu9QRHRsW4yov5EDlaexxzyKrIsW7S2Cokf5jEuh/egLhsX4a8lqL2+Wh+GAWhd+sJv/3AxTaEylwplKpNPx3yhxqJCDYQHGOtV7TkAY3kcZjxPdLJO7UgcSlhPlthkqXx8Wewj1sydnC1pytHC0/ytDYoUzsMZGxiWPb5cLaWqX2Uj4/9DmrM1azI2+Hd71eoyfYGIxRZ8SkM2HSmbyvjVpjzes6z3XX+WzXqs+51lx+yf+FXwp+IaMkA0+docM6jY60iDQGxwxmUPQgBsUMIiW0dXMdKIrCz/k/83HGx6w7uM5nAqnBMYOZ2Wsmk1MmE2bqHKMfhGhvEj46OUVRyF/2FIUvvwxA/D8WEnHppX4uVQMUBfatgy8fovCYlT22c9hbORG7p6o9VwM9BkQx8IxEkgdHdd4+ALZi+PHfKJv/hcWqo8CVQoFuKIWhEymoiKIkv9KnqSgoREucdhdxnm3EG/YRM2QQhumPHv926+3A7XGzt3gvW7K3sCVnCzvydmCpfcvyWkw6E2MSxjCxx0TGdx9PVGDbTmXeHE63k2+zvmX1gdU+HSs1aBiVMIrpPadzVo+zCDYGN32gNmB1WtlduJuf839mV/4udhXsIt+WX2+/EGOIGkSiB3lDSURARKPHPVZxjFUZq/gk4xMOlx/2ro8Pimd6z+nM6DWDlLCU9vhIQnRqEj46ufznnqfgn/8EIO6BvxM5e7afS9SArO04PlvE/r0a9tjOIddZ00YdHGGi/+mJ9B+b0PQEVp1NZRls/Q9sfr7mdvWh3XCOupPCuJlUWhSic94meMvD4HGqQ4GnLYOBF3ZYET2Kh/TidLbmbGVLzha25W6rN3tjqDGUEXEjGJUwiqSQJDYf28z6w+s5Zjnm3UeDhiExQ5jYYyITkyaSGpbabmWurgFYfWA1n2V+5jPUsnd4b6b3ms7U1KnEB3XwrLN1KIpCrjXXJ4zsLtzd4PwXSSFJ3jAyOHowSSFJbDy6kVUZq9ias9W7X6A+kHOSz2FGrxmMjB958gwRFaIdSPjoxApe/Df5Tz8NQOx99xI1Z45fy1OXUniAvI9eYM+vWtIrx+FU1Gp8rRZShsQw4IxEkvpHdvqbSDXJYYHtr6p9VyrUm9cRFAshceocIgD9psG0p9Q7zLYjRVE4WHqQLTlqzca2nG0U24t99gk2BDM8bjgj40cyKn4UaRFp9SZ+UhSFfcX7WH9kPRuPbGRP4R6f7SmhKUzsMZFJSZMYHDO4TS6SR8qOsPqA2o+jdg1ATGAMU1OnMr3XdNIi0jr1MFynx0l6cTq78nfxS8Ev/JL/y3H7pGjQMCp+FDN6z+DsHme3+1BYIU4WEj46qcJX/kveE08AEPO3u4i+8cbjv+nARjj4DQTFQEi8emfUkHgIjvdOaNUW7IV57HvnPXbvCaTQleJdHxZlYMC4HvQbk/CH6FDpw1kJP62E756G0iPqOlMYTH0cBl/WLvOEKIrC4fLDap+N7K1szd1Kga3AZ59AfSDDYod5w0b/qP4tnhchx5LDhiMb2HB4A1tzt+Ly1NxULSogiglJE5iYNJHRCaMJ0Df/71H1BFmrD6zmp/yffMp8Vo+zmN5zOqMTRrd4VszOpNReyu6C3fxcUFNDUmIvITk0mRm9ZjC953QSghP8XUwhOh0JH51Q0Wuvk7t4MQDRt/2VmL/85fhv+uVd+OAmGh27GhBeFUbifEOJN6TENRlSFEUh+/c89qz6jv2ZQbgVNVzoNC56nhLEwLMHkJgW3ql/ubYJlwN+eUe9Q+vY2yCsW5sePseSw+Zjm71NKbnWXJ/tJp2JU2NPZVT8KEbFj2Jg9MA2HRVR7ijnu6zv2HB4A99mfUuFs8K7LVAfyNjEsUxMmsi47uMa7OvgcDv45ug3fJLxCd9kfeMNMlqNltMSTmNaz2mc1eOsP2wNgKIolNhLCDd1gX8LQpwACR+dTPHbb5OzYCEAUTf/mZg77jj+f2K/r4F3rlLvh9L7bHVa8vIctZmgPAdcDd/0qi63osNiSKXC1IcKbQ/KSaTCHU2FM5SiEiNl5TW1GZEBuQwYG0vf88d16huBnQwqHBV8cegLPs74mO252322GbQGBscMZlT8KEbGj2RIzBCMuo6pVXK6nWzN3cqGwxvYeHQjOZYc7zatRquOnElS+4kU2Ar45MAnrMtc59PvpG9EX6b3ms6U1Cl+mUdDCNE5SfjoREref5/s+/8OQOR11xF797zjB48DX8Mbl6i3Lx98uXqzMm2tNnpFgcoSlLIcKvOyKc8toiK/lPLCSirK3FSU66ioDKDcEYrVHY5C41Xgeo2NPiE7GXBWX+LOvhCNTqY9bi23x82POT+yKmMVXx36ikq3GhA1aLxhY1TCKIbEDOkUQ2IVReG3ot+8zTN7i/c2um+sOZbze57PtJ7TSItI68BSCiFOFhI+OonSVas4du99oChE/OlPxP3f/OMHj6Pb4NUZ4LRAv2k4pr1CzmELFcV2yosqqSiqrHldbMftPM4t0AGtDoJDIMTsIDigghB9McGaHEK0ucQPSsN45o1gaJuLYfUF7bfC39rkeHVpNBqSQ5MZFD2ow2oLjudAyQE+zviY1QdWk2etuUlXSmgKM3vPZFrPaX4f6dEcWRVZbDyykQ2HN7AtdxsmnYmzk89meq/pjIwbeVL34xBCtD8JH51A2Zo1ZM27Gzwewi+/jPiHHjp+8MjdDcunQmUJ9JxA/pn/ZfULv2Eta/rmTuYwI8ERAYREmqqeAwiu9Tow2NDuE2KVVJbw6cFP+TD9wyZ/QbcVk87EkJghjIgbwYj4EQyOGYxJ13a3Oj+eksoS1hxcwycZn/Br4a/e9aHGUKakTmFmr5mcEn3KSdtHwOq0otfqO03AE0J0fhI+/Kzsiy/ImnsnuN2EXTSLhIcfRqM9ztDGwgz473lgyYPuozg6+lXWvJKOs9JNUJiRqO4htcKFieDIAIIjAggON6Ez+GduAY/i4YfsH/gw/UO+OvyVd0Ipo9bIiPgR7XLhqp7hs/ZNukDtRzEoehAj4kd4+1G0ddNG9QRaqzJW8fXRr70dL/UaPWd0O4MZvWcwvvt4uWALIbokCR9+VL5+A0fvuAOcTsJmziBh8WI0uuNUV5dmqcGj9DDEnUL6kNf58s1DeFwK3dLCmXLLYEyBnacvxrGKY3y0/yM+2v8R2ZZs7/p+kf24sPeFnN/z/HadUlpRFA6WHWRbzja25W5je8528mx5PvvotXpOiTqFEfEjGBE3gqGxQ1s1GkNRFPYU7eGTjE9Yc2CNzxwc/SP7M6PXDKakTvHLbKJCCNGZSPjwk4pvv+XorX9BcToJnTqFxMcfR6M/TmiwFMDyKVCwDyJ78UvaG3z7cTYo0GtYDGdfO6BT3BnV7raz/vB6Pkz/kB+yf0CpGv4bYgzh/NTzmdVnFv2j+vulbNVzZ1SHkW2523xGcYB6T4+BUQMZHj+cEXEjGBY7rMkpvvOt+aw+sJpVGavYX7Lfuz4qIIppPacxo/cM6XgphBC1tFv4SElJ4dChQ/XW33rrrfzzn/9EURQWLlzISy+9RHFxMaNHj+af//wnAwcObJfCdyaWzZs5cvMtKHY7IeecQ7dlS9EYjjNctbIUVkyDnF9QQrrzQ7eV7Nio/rIeNL4bZ1yW5veZRPcW7eWD9A9YfWA1ZY4y7/rRCaO5sPeFnNXjrBZNUtURFEUhqyKLrTlb1ZqR3O1kVWT57KPVaOkX2Y+RcSMZET+CYXHDMGqNbDiygY8zPmbzsc3em5IZtUYm9ZjE9F7TGZs4tsUTfgkhRFfQbuEjPz8ft9vtXf71118555xz2LBhAxMmTOCxxx7jkUceYcWKFaSlpbFo0SK++eYb9u7dS0hISJsXvrOo3LuXzMsuR6msJHjiRLo/8zQa43Ha/R1WWDkLDm/GExjLhoiV/L5DvWHY6Bk9GT4l2W+dFcscZaw5sIYP93/oM013nDmOC3pfwAW9L6B7SHe/lK21siuy2Za7zRtIjpQf8dmuQUOAPgCby+Zdd2rMqczoPYPJKZMJNZ4cfxeFEMJfOqzZZe7cuaxevZr09HQAEhMTmTt3Lvfeey8AdruduLg4HnvsMf785z+3eeE7i7wnn6TwP69gHjGCpP++gvZ4wcPlgLevgP1f4jTEsC5gOYfSnWg0MGF2PwackdgxBa/Fo3jYmrOVD/d/yJeHvvTebEuv1TMpaRKz+szitITT/jDDLXMtud4mmm0527z380gMSmR6r+lM7zWd5NBk/xZSCCFOIi25fre6/tjhcLBy5UruuusuNBoNBw4cICcnh3PPPde7j8lkYvz48WzatKnR8GG327Hba+4qWVZW1uB+nZmjqikq5Nxzjx88PG744AbY/yWV2lhWe14mN92JzqBl8g0DSR0S0wElrpFjyeHj/R/z0f6POFpx1Lu+d3hvZvWZxbSe05q8vfjJKi4ojvN7ns/5Pc8H1D4eRZVF9InoI3cmFUKIdtbq8PHRRx9RUlLCnKq7subkqB384uLifPaLi4trsJ9ItSVLlrBw4cLWFqNTcGRmAmBMOc4vZY8HPrkd9nxMuZLAJ5X/orjIg8ms5/xbB5PQO9x3d8XD0m1LWX94PRqNBq1Giwb1WavRquto5PXx9tVosTgt7Mzb6e3bEGwIZkrqFGb1mcXAqIEn7RwVrRFjjiHG3LHBTwghuqpWh49XXnmFKVOmkJjo20RQ94KlKEqTF7H58+dz1113eZfLyspISkpqbbE6nOLx4Dis9h8wpqQ0saMCn98PO1dS6ErhE9tSLBYIjjAx/bZTiUwMqveW53c+z2t7XmunktcYETeCC/tcyDnJ53SKab+FEEL8sbUqfBw6dIgvv/ySDz74wLsuPl6dPjonJ4eEhJrbTefl5dWrDanNZDJhMnXczJRtzZWTg2K3g16PIbGJvhpfPwY//Itjjv58WrEIh0NLREIQ028bQkhk/dEiaw+u5eVdLwNw78h7OSX6FBQUPIoHj+JBURQ81HqteI673UOt11W1HcPihknfBiGEEB2qVeFj+fLlxMbGcv7553vXpaamEh8fzxdffMHQoUMBtV/I119/zWOPPdY2pe2EvE0u3bs3PqfH5n/BxiUcqBzF5+X34nZrSegVxtRbBzd499jdBbt54PsHALh24LVcNeCq9iq+EEII0eFaHD48Hg/Lly/nmmuuQV/rYqvRaJg7dy6LFy+mT58+9OnTh8WLF2M2m7nyyivbtNCdSXVn00abXHa8Duvm86v1XL4pvxlF0ZAyOJrJNwxEb6w/ciTPmsft62/H7rYzrvs47hh2RzuWXgghhOh4LQ4fX375JYcPH+a6666rt+2ee+7BZrNx6623eicZ+/zzz5s9x8fJyJFZFT6SG2i62P0Ryqrb2VpxKVsrrgBgwOkJjL+yL1pd/REVla5K5m6YS54tj15hvXjszMf+MENbhRBCiGotDh/nnnsujU0NotFoWLBgAQsWLDjRcp00Gh3pkv4lnv/dyDelN7HbNhmAEVNTGDU9tcEOuIqisGDzAnYV7CLMFMZzk55rcvpvIYQQ4mQl80SfoAabXQ5txvX2tXxedCcH7aeBBsZdlsagCY3PCvrKr6/w6YFP0Wl0LB2/lKTQk2fEjxBCCNESEj5OgOJy4TiqTszlbXY59hP2lXP4NO8esp0D0eo1nHPtQHoPj230OBsOb+DZHc8CMH/UfEYnjG73sgshhBD+IuHjBDizssDlQmMyoY+Ph/x9VKy4jk+y76PIlYwxQMfUWwbTrW/jM4TuK97Hfd/eh4LCZX0v47J+l3XgJxBCCCE6noSPE+BtcunRA03pEYr+cxOfHL2XCk8M5lAD028/lejujXe2La4s5vb1t2N1WRkVP4p7R93bUUUXQggh/EbCxwnwjnTpnkDOv+9g9eE7sSshhMeYmH7HMEKjG58t1Ol2ctfGu8iqyCIpJIml45di0Naf80MIIYT4o5HwcQKqR7pYDHo2Zv4ZFyZikwKYdvsIAkMav8Gcoigs3rKYbbnbCDIE8dyk5wgPCO+YQgshhBB+JuHjBFQ3uxzV9cGFicToUqbNG4/B1PTcHG/9/hb/2/c/NGh4fNzj9Arv1RHFFUIIIToFuXf4CagOHyVKJAD90qzHDR6bj23m8a2PA3Dn8DsZ131c+xZSCCGE6GQkfLSSx+HAeewYClDsjAIgulvTd4Q9VHaIv339N9yKm+k9pzNn4Jz2L6gQQgjRyUj4aCXnkSPg8eAMS8DuDkSDm4ju4Y3uX+4o57b1t1HuKGdwzGAeGvtQgzOdCiGEEH90Ej5aqbrJxZY6BIAIfRb6sJgG93V73Nz9zd0cLD1InDmOZyY+g0ln6rCyCiGEEJ2JhI9WchzMBMAa3RuAKP1BMEc3uO+y7cv4Put7AnQBPDvpWaIDG95PCCGE6AokfLRSdc1HuUmdNj1anwnmqHr7fbT/I17b8xoAi85YxICoAR1WRiGEEKIzkvDRStXho8yjzmAaHZQPet+5PXbm7eQfm/8BwM1DbmZyyuSOLaQQQgjRCUn4aCVHZiZurYFymzoraVSY1Wd7dkU2czfMxelxck7yOdwy5BZ/FFMIIYTodCR8tILHZsOVm4slKBEFDYHaEoLCajqQWp1Wblt/G0WVRfSL7Mei0xeh1chXLYQQQoCEj1ZxHD4MgCVG7Wwarc+EILUTqUfxcP9397O3eC+RAZE8O/FZzAazv4oqhBBCdDoSPlrBO9IlNg2AKH0mBKnDbF/4+QW+PPwlBq2BZyY+Q0Jwgp9KKYQQQnROEj5aobqzaUVQIgDRhkwIiuazzM948ecXAXhwzIOcGnuqn0oohBBCdF4SPlrBcegQClCmhAFqzccencID3z0AwNUDruaC3hf4r4BCCCFEJybhoxUcmZlUBkTi9OjQaty4jdncnvUZle5KTu92OncNv8vfRRRCCCE6LQkfreA4dIiKoO4ARJhyeT8skFxnGSmhKTwx7gl02qbvbCuEEEJ0ZRI+WshdXo67sJCK4G6AOtLlF5M6udjs/rMJMYb4s3hCCCFEpyfho4UcmWpnU0tkKgBRmr3sqgofg6IH+a1cQgghxMlCwkcLeUe6hCQB4DEdpkynw6A1kBaR5s+iCSGEECcFCR8t5DiUiUtnwqoNBSAr+BgA/SP7Y9AZ/Fk0IYQQ4qQg4aOFHJmHsFTN7xEUDL+ZHQCcEn2KP4slhBBCnDQkfLSQ49AhyoPVkS5RUU52mdR7ukj4EEIIIZpHwkcLKIqCIzPTO9IlMrSM34zS2VQIIYRoCQkfLeAuKcFTVkZFkBo+HKZDOLQaQjR6eoT28HPphBBCiJODhI8WcGRmoqDBEqI2u2Qb9gJwiikarUa+SiGEEKI55IrZAo5Dh7AFRuPWGtEZtPymTQfglKAkP5dMCCGEOHlI+GgBtb9HVWfTxCB+dZcCMCiijz+LJYQQQpxUJHy0gHpPF7W/R1hCABm4ADglSjqbCiGEEM0l4aMFHIcOeUe62MPLUDQQ73IRE9nbzyUTQgghTh4SPppJHWZbM8dHjukAAIPsDgiK9mfRhBBCiJOKhI9mcuXn43Ao2AMiAdjr2QbAKXYHBEb4s2hCCCHESUXCRzM5a/X3CI408XPFzwAMIgC0On8WTQghhDipSPhoJnutmU1DE4zkVBaiURQGGCP9XDIhhBDi5CLho5mctTqbOsLLAejldBIUFOPPYgkhhBAnHQkfzaQOs1U7m+YGHgaq+ntIZ1MhhBCiRSR8NFNl5iEsQQkA7NP+AlSNdDFL+BBCCCFaQsJHMygeD6X5lXh0RvQGDTsqfwTgFLsdpNlFCCGEaBEJH83gysmh3BgLQHC8kXJnGUY09HE4ISjKz6UTQgghTi4SPpqh9rTqrsgKAPoregwgNR9CCCFEC0n4aIbaN5QrMGcBMMih3tdFwocQQgjRMhI+mkGdVl2t+UjX7QLgFKtF3SgdToUQQogWkfDRDOWHsnGYwgHY6VQ7mw6qKFE3ylBbIYQQokUkfDRDUU4lAAFBLqzackINwSS5XKDRQUC4fwsnhBBCnGQkfByH4nJRbDWqryPVEDIoNBUNqLUeWvkKhRBCiJaQK+dxOI8do8KsTi5WEJ4HwCmB6rL09xBCCCFaTsLHcdQe6ZKh3wPAIEOEulH6ewghhBAtJuHjOGwHDmExxwOwhx0ADNSY1I0SPoQQQogWa3H4yMrK4qqrriIqKgqz2cypp57K9u3bvdvnzJmDRqPxeZx22mltWuiOVHQgH0WrR6dxUmYqJDEokWh71TBbmeNDCCGEaDF9S3YuLi7m9NNPZ+LEiaxdu5bY2FgyMjIIDw/32e+8885j+fLl3mWj0dgmhfWHwmNW0IM+oAw0cEr0KVBaoG6UPh9CCCFEi7UofDz22GMkJSX5BIuUlJR6+5lMJuLj40+4cJ1BcbkWIqA8OB+oCh/H1qkbpdlFCCGEaLEWNbusWrWKESNGcMkllxAbG8vQoUN5+eWX6+23ceNGYmNjSUtL48YbbyQvL6/RY9rtdsrKynwenYXH4aCUMAAygvYDVeHDWlXzIeFDCCGEaLEWhY8DBw7wwgsv0KdPH9atW8fNN9/M7bffzmuvvebdZ8qUKbzxxhusX7+epUuXsnXrViZNmoTdbm/wmEuWLCEsLMz7SEpKOrFP1IacR454byj3W/DvaDVaBkYNBItaCyJ9PoQQQoiW0yiKojR3Z6PRyIgRI9i0aZN33e23387WrVvZvHlzg+/Jzs4mOTmZt99+m1mzZtXbbrfbfYJJWVkZSUlJlJaWEhoa2pLP0ubyPl3Pe58Aiof/jL6HlKhkPpz5ISzuDo5y+Ot2iO7t1zIKIYQQnUFZWRlhYWHNun63qOYjISGBAQMG+Kzr378/hw8fbvI9ycnJpKenN7jdZDIRGhrq8+gs8vbmAmBUSnDpnAyKHgROmxo8QJpdhBBCiFZoUfg4/fTT2bt3r8+6ffv2kZyc3Oh7CgsLOXLkCAkJCa0roR8VZqlDat16NYScEn0KWKr6e2gNEBDmr6IJIYQQJ60WhY8777yTH374gcWLF7N//37efPNNXnrpJf7yl78AUFFRwbx589i8eTOZmZls3LiR6dOnEx0dzYUXXtguH6A9FZWqz/mGTAC15qN2Z1ONxj8FE0IIIU5iLQofI0eO5MMPP+Stt97ilFNO4eGHH+bpp59m9uzZAOh0Onbt2sXMmTNJS0vjmmuuIS0tjc2bNxMSEtIuH6A9lbqDATgccgiTzkTviN41NR/S5CKEEEK0Sovm+QCYNm0a06ZNa3BbYGAg69atO+FCdQaOMgtWYxQAB2KO0T+yPwatoSZ8yARjQgghRKvIvV0akffTARSNDr3LQl5YqdrfA2SYrRBCCHGCJHw0Iu/3HAAMzmzQVPX3AJlgTAghhDhBEj4aUXBEHU5r5yhQK3xInw8hhBDihEj4aERxsTr3WonpKGGmMLqHdFc3SJ8PIYQQ4oRI+GiAoiiUOs0A5IZkcUr0KWiqh9VKnw8hhBDihEj4aEBFsR2nxoRGcZMZlVvT5ALS50MIIYQ4QRI+GpC/X63dMFtzORbt8g0f0udDCCGEOCESPhqQtycbAKMtC5tJo97JFsBhAadVfS19PoQQQohWkfDRgIJDJQC4OEq34G5EBaqTjXlrPXQmMJ18M7YKIYQQnYGEjwYUFbkBKDcdq5lcDHybXOS+LkIIIUSrSPiow+lwU243ApAXnCWdTYUQQog2JuGjjqIsC6DB4CgjK7K8Ts2HDLMVQgghTpSEjzoKjqozmwZXZJEbpaV/ZP+ajTLBmBBCCHHCJHzUkZ9RCEBIRRZBKb0xG8w1G701HxI+hBBCiNaS8FFHQWYJAIr7KP0Sh/hutKrBRMKHEEII0XoSPmpRFIWiAicAFkOdkS4gfT6EEEKINiDho5bywkqcLg0aj4v84DrTqoP0+RBCCCHagISPWgqOVgAQZM0hP1pLr/Bevjt45/mQmg8hhBCitSR81FIdPoIrjmJMTkav1ddsVJRa83xE+aF0QgghxB+DhI9aag+zje5Tp8nFUQGuSvW11HwIIYQQrSbho5aCQ6UABFmySO0/xndjdZOLPhCMQR1cMiGEEOKPQ8JHFYfNRXmxOtKlUnuUUxJP9d1B+nsIIYQQbULCR5XCLLW/h9FeQkm4g27B3Xx3kP4eQgghRJuQ8FGlurNpSEUWrm6xaOretVbm+BBCCCHahISPKtU1H8EVRzGn9qq/g4QPIYQQok3oj79L1+AdZmvJIqDv2Po7WKqmVjdLs4sQQghxIiR8AIpH8an5SB50Zv2dpOZDCCGEaBPS7AKU5ttwOTxo3Q5MlflEp/avv5O3w6lMrS6EEEKcCAkf1JpW3ZKNJSYQjb6BCiGp+RBCCCHahIQPanU2tRzF0z2+4Z2kz4cQQgjRJiR8UPueLlmE9Eyrv4OiSM2HEEII0UYkfAB5R9Rp1YMrsojtO6T+DvYy8Kizn0qfDyGEEOLEdPnwUWlxYq2aVj3YkkVwrwZqPqqnVjcGgyGwA0snhBBC/PF0+aG21f09TJVFGFw2jMnJ9XeqDh/S30MIIYSfKIqCw+2h0umh0unG5nBjc6qPytqvnZ5662wOt/oep/qsQcOLfxrut8/S5cNHzbTqR/EY9ejjG+hwKv09hBCi2RRFodLpweJwYbW71WeHC4vdXfPsdGO1u7A4aj07XCgKaDWg1WjQaDTe11otvssaDRrv6wb211TvX7VOq0Gn1aD3edbWLOsaWV/1rPMua+vt7/Io2J0e7C43dpcHh8uD3VW17Kz12uXx2c/uctfs67O+5r3VYaE6aHiUtvkzMur82/Ah4eNoOaD299B2T0SjbeAPROb4EOIPwe5yY7G7qah0UW53UlHposJe61G1XF7pu2x1uDDotAQadQQa1EdArdeBRh0BBh0BBm2j2wMN6j6BRh0Bei36VvznrygKigIeRcFT9VyzrK5Tqp7dHgWXx4PTpeD0eHC61dcOd9XrqofDpXhfu9x1tys4XHWW3Z6qi2KdMOGoCRJWpxuljS6SomE6rabW36laf++8f8dq/l4G1tqn9nZFUerfx6yDdPnwcexQEaD29wga0qfhnbw1HxI+hGgOt0ehoMJOdmklOaWV5JTayCmzk1Nqo6DCAdD0r1ZtnV+tzfyV63B7sFSFhvJa4aE6SDjcHj9/MzUMOg0BBh1GnbZZYaJ628nGbNRhNuoJMqkXwSCTHrNRR5BRj9lU82w2qOu1Wk3VZ67znXhqL1d9Nxx/H+9xPApuRakKZQpujxq2apbVsOaz7PZd71YU3O7a+yu43B70Oi0mvRajXn026XWYDLVe67WYqv6s66+v9dp7jJr31w6vtUOtwc81FyeqS4cPj9tDWY4d0BBccZSAlGEN7+jt8yHhQwi7y01emRossktt5JZVkl1a6X3OKa0kr9yOu63qh9uB2agj2KRXHwH6Rl+HVD0HGvQ43WoVuN3bhl6/Srx2m3pNO7vHZ111gHC6FZxuV7t+ToNOg0Gn9T6MOg0GvbbWOo3Ps7F6vV5dV72sr/XaoFMvkEGmqlBh1GE2VT1Xh4yqcBFoUMOEEHV16fBRkmsDtwY8dgJtBRhTUhresTp8SJ8P0c6qL3B1O4qpFz2Pz3KlU20fbqhGQAPedu4m28F9ahRqtms0UGZzqrUWZZU+z4UWR7M+i1YDcaEBxIUGkBBW8xwTYkKn1ai/SD11f+03/Ku1Ob+EFdSq6JAAA8EmHcEmgzdMhAToCaoKFUFGXauaPNqCoijYXR6fcOJ0K+iarOmp/edWq7+Dtu6fe/39heisunT4KMhS+3sE2rLQoDQ80gWkw2kX5PYojXYWc7iP14msgc5mzoZ7nVf/gq6sChOuTlxbUJtRryUhLID40ADiw6oeVeEiPiyQ+NAAooONfrvId1Yajaaqb4iOcH8XRgg/6tLhI/ewOrlYeGkWAIbGwoe1amr1IBlq29Fcbg9FFgd55XbyK+wUVjiwOd04a3WCc7irOsxVr/MotbbX6UBXq8Odq+p9Dm/Hu5pe6v4OAVoNPh0Z672uWjbqtPXbvevWIHiOU4PQSP+CYJO+VsAI9Km9CDcb5Je1EKLVunT4OJKZB0BIRRZasxl9TCM1G1Lz0aY8HoUSm5P8crv6qKikoNxBfoW6XFD1nF9up8jq8HsnO71W4+0wZqrToaymA1n9bdWvjVWdxgKqRjkcb8SE2plMIxd3IcQfVpcOHyXHKgEdIRVHMaQkN/yfvcdTU/MhHU6bpCgKRRYHh4qsZBXbfEJEfkVNqCiscLSoZkGrgahgE9HBJqKDjQSb9Oh1vh3i1E5ydZZ1Gox6LXptzWufbd6OdTUd7gJqBQejXg0X0nQghBBtq8uGD1uFA8WiAyDIcgxj8sSGd6wsAU9Vj3QZaovHo5BbXsmhQiuHCi1kFlo5XGgls9DC4UIr5fbm996PMBuICVFDRUyIiZiqZ+9y1evIICM66TEvhBB/GF02fFTPbKq489G77Y2PdKmu9TCFgt7UMYXzM6fbw7ESW1WwUANGddg4XGTF7mp6roSEsACSIszeAKGGCGNVwFBHO0QGGTHqpUZBCCG6oi4bPrIy1eGzAZVqZ9Pjj3T5Y9V6VDrdHCmyVgULC4eqay+KrBwttjU5R4NOq6F7RCDJUUGkRJnpEWkmJSqI5CgzSZFmAgy6DvwkQgghTjZdNnwcPJANQGTZMaCp8HHyTjDmcns4WmzjYKGFg/kWMgstHCxQH1kltiY7cpr0WpKjzPSIVANGcpSZ5KqAkRgeeNLPrieEEMJ/umz4KM6yAiaiC44ANDHBWOce6eLxKOSUVZJZYOFAVbDIrHo+XGRtsmNniElPcnRVqKiqvegRpT7HhphkZkIhhBDtokuGD7fLg6fYgAb1ni7asDD0EREN79wJ5vhQFIVCi8Nba1E7YGQWWqh0Nt4Hw6TXkhodREpUEKkxQaRWP0cHERVklOGcQgghOlyXDB9F2RY0Hi1ubARUFmJMG9z4zh1Y86EoCrlldvblltd6VJCRV9HkKBK9VkOPSLMaMqLVYFH9iA8NkBoMIYQQnUqXDB8HDqidTN0cQwMYUxrp7wE14aON+3wUVFSFjJxy9uVVqM+55ZRVNhwyNBpIDAukZ0xVLUatgNE9IlDmohBCCHHS6JLhIyMjC9AQ7FSDRaOdTeGEbypXYnWwL7eiXm1GUSM359JpNaREmUmLC6FPXAh940LoHRtMcpSMIhFCCPHH0CXDR2FWBVpCiKlQp1c3Jqc0vrM3fDRd81Fhd7Evt5z03HL25lSQnlfO3pxy8srtDe6v0UCPSDN9YkPoGx9MWlwIaXEh9IwJwqSXkCGEEOKPq8XhIysri3vvvZe1a9dis9lIS0vjlVdeYfjw4YDab2HhwoW89NJLFBcXM3r0aP75z38ycODANi98ayiKgjvfgBaIyT8ENDHSBcDadPjIKa3kqS/28b8dRxudG6NbeCBpcTUBI62qNiPQKCFDCCFE19Oi8FFcXMzpp5/OxIkTWbt2LbGxsWRkZBAeHu7d5/HHH2fZsmWsWLGCtLQ0Fi1axDnnnMPevXsJCQlp6/K3WHmJDYMjAA8ewnIygCb6fNS+r0udZpfySif//voA//nugHe0SVyoqVbACPY2nQSbumQFkxBCCNGgFl0VH3vsMZKSkli+fLl3XUqtWgNFUXj66ae5//77mTVrFgCvvvoqcXFxvPnmm/z5z39um1KfgD37DgBQaShA53Gii4pCFxzc8M62YlCqhrGa1aG2TreHt7Yc5pkv0yms6rcxIjmC+VP7Mzy5keG6QgghhPBq0RCJVatWMWLECC655BJiY2MZOnQoL7/8snf7wYMHycnJ4dxzz/WuM5lMjB8/nk2bNrVdqU9A+v7DABj1xepzU00u1SNdAsJRtHrW7srm3Ke+4cGPd1NocdAzOoh//2k47908RoKHEEII0Uwtqvk4cOAAL7zwAnfddRf/93//x5YtW7j99tsxmUxcffXV5OTkABAXF+fzvri4OA4dOtTgMe12O3Z7TafMsrKyln6GFik4Wo6RAKIoB44z0qWqv0elMYIrX9jEjsMlAEQHG7nj7DQuH5kk04wLIYQQLdSi8OHxeBgxYgSLFy8GYOjQoezevZsXXniBq6++2rtf3VkzFUVpdCbNJUuWsHDhwpaWu9Wc+TqMQFxlVc1HE+EjJ/sI8cAvJUZ2OEoINOi4cVxPbhrXU/pxCCGEEK3Uop/tCQkJDBgwwGdd//79OXxYbcqIj48H8NaAVMvLy6tXG1Jt/vz5lJaWeh9HjhxpSZFaxFJpI7AiDIC4Ju7pkl9u5+8f7eJfn/4IQLESyhWjkvj67gncdU6aBA8hhBDiBLToKnr66aezd+9en3X79u0juar2IDU1lfj4eL744guGDh0KgMPh4Ouvv+axxx5r8JgmkwmTydSasrfYT3v3oEWHXW/FeDgdBd+RLlaHi/98e5B/f52BxeFmrr4UgNMG9WXyrCamYBdCCCFEs7UofNx5552MHTuWxYsXc+mll7JlyxZeeuklXnrpJUBtbpk7dy6LFy+mT58+9OnTh8WLF2M2m7nyyivb5QO0xN70Q0A4ntBylFK1b4mxRw9cbg/vbT/Ksi/2kV81KdiQ7mFcHmWGvRAWneDHUgshhBB/LC0KHyNHjuTDDz9k/vz5/OMf/yA1NZWnn36a2bNne/e55557sNls3Hrrrd5Jxj7//PNOMcdH3pFSQggnPMgJgD4+ng2ZZTy69nfS8yoASIoM5J7J/Zg2OAHNe2qo6oibygkhhBBdRYs7L0ybNo1p06Y1ul2j0bBgwQIWLFhwIuVqF448tYtLokG9eVu6IZxbX90GQLjZwO2T+jD7tB4105tbqiYYq5rjQwghhBAnrsv0nCypLCGoLBKA7H1HSAV+04Vj1Gu57vRUbpnQi7BAg++bquf5kJoPIYQQos10mfBRmGsnwB2EBw+OzH0ARPXtzYZ5E+gWHtjwm7zho+mbygkhhBCi+bpM+DCFhvBypJ0gByxyqM0pF80cQ0hjwcPtUqdXB6n5EEIIIdpQlwkf3SPM3DW9P6lRZmLX5+PhOLOb2ooABdBAYGQHlVIIIYT44+tSc4NfPSaFMREaPFYraLUYkpIa39miTq1OYAToukxGE0IIIdpdlwofAI7MTAAMiYlojcbGd5TOpkIIIUS76Hrho+oGd002uYD3pnLS2VQIIYRoW103fDRwTxcfFgkfQgghRHvoeuGjqtnluDUf1eHDLOFDCCGEaEtdMHxU13wcL3xInw8hhBCiPXSp8KF4PDgOHwaa0ewifT6EEEKIdtGlwocrJwfFbge9HkNiYtM7S58PIYQQol10qfDh7WzavTsa/XHm7pA+H0IIIUS76Jrh43hNLiB9PoQQQoh20rXCx8FMoBkjXdxOqCxRX0uzixBCCNGmulb4ONTMkS5W9cZzaLTq9OpCCCGEaDNdNHykNL1jdZNLYCRode1bKCGEEKKL6TLhQ3G5cBw5ArRggjHp7yGEEEK0uS4TPtxlZZjS+qCLjEQfH9/0zjLMVgghhGg3XeZe8frISHp+8AGKoqDRaJreWSYYE0IIIdpNl6n5qHbc4AEyzFYIIYRoR10ufDSLTDAmhBBCtBsJHw2RPh9CCCFEu5Hw0RDp8yGEEEK0GwkfDZE+H0IIIUS7kfDREEvVDKfS50MIIYRocxI+6nLZwV6qvpZmFyGEEKLNSfioy3tfFx0EhPu1KEIIIcQfkYSPurz9PaJBK1+PEEII0dbk6lpXdfiQ/h5CCCFEu5DwUVd1Z1Pp7yGEEEK0CwkfdckwWyGEEKJdSfioSyYYE0IIIdqVhI+6anc4FUIIIUSbk/BRl0wwJoQQQrQrCR91SZ8PIYQQol1J+KhL+nwIIYQQ7UrCR12W6vAhNR9CCCFEe5DwUZvTBo4K9bU5yr9lEUIIIf6gJHzUVl3roTVAQJh/yyKEEEL8QUn4qK12fw+Nxr9lEUIIIf6gJHzUZpHOpkIIIUR7k/BRm9xUTgghhGh3Ej5qk5EuQgghRLuT8FGbTK0uhBBCtDu9vwvQqVirplaX8CGEOEFutxun0+nvYgjRpgwGAzqd7oSPI+GjNplaXQhxghRFIScnh5KSEn8XRYh2ER4eTnx8PJoTGBUq4aO26j4f0uFUCNFK1cEjNjYWs9l8Qv9BC9GZKIqC1WolLy8PgISEhFYfS8JHbdLhVAhxAtxutzd4REXJLMnijycwMBCAvLw8YmNjW90EIx1Oa/NOMib/aQghWq66j4fZbPZzSYRoP9V/v0+kT5OEj2oOCzit6mup+RBCnABpahF/ZG3x91vCR7XqJhedCYzB/i2LEEII8Qcm4aNa7f4e8qtFCNHFFBYWEhsbS2ZmZoPbV6xYQXh4eIeWSXSs559/nhkzZnTIuSR8VJP+HkKILmzJkiVMnz6dlJQUADIzM32q1y+77DL27dvnp9K1XEpKChs3bmz2/hs3bvR+9q6i7me+8cYb2bp1K9999127n7tF4WPBggVoNBqfR3x8vHf7nDlz6m0/7bTT2rzQ7ULm+BBCdFE2m41XXnmFG264odF9AgMDiY2N7cBSdU3+nJjOZDJx5ZVX8txzz7X7uVpc8zFw4ECys7O9j127dvlsP++883y2r1mzps0K265kjg8hRBe1du1a9Ho9Y8aMaXSfus0uCxYs4NRTT+W///0vPXr0IDg4mFtuuQW3283jjz9OfHw8sbGxPPLIIz7HWbZsGYMGDSIoKIikpCRuvfVWKioqfPZ5+eWXSUpKwmw2c+GFF7Js2bJ6TT6ffPIJw4cPJyAggJ49e7Jw4UJcLtcJfxfVMjIymDlzJnFxcQQHBzNy5Ei+/PJL7/b58+c3+ON68ODBPPTQQ97l5cuX079/fwICAujXrx//+te/vNuqa5feffddJkyYQEBAACtXrqSwsJArrriC7t27YzabGTRoEG+99ZbPecrLy5k9ezZBQUEkJCTw1FNPMWHCBObOnevdx+FwcM8999CtWzeCgoIYPXr0cWuDZsyYwUcffYTNZmvhN9YyLZ7nQ6/X+9R21GUymZrc3mnJfV2EEG1MURRsTrdfzh1o0DV7VMI333zDiBEjWnyOjIwM1q5dy2effUZGRgYXX3wxBw8eJC0tja+//ppNmzZx3XXXcdZZZ3kv1FqtlmeffZaUlBQOHjzIrbfeyj333OO9KH///ffcfPPNPPbYY8yYMYMvv/ySBx54wOe869at46qrruLZZ5/lzDPPJCMjg5tuugnA58J/IioqKpg6dSqLFi0iICCAV199lenTp7N371569OjB7NmzefTRR8nIyKBXr14A7N69m127dvG///0PUEPUQw89xPPPP8/QoUPZuXMnN954I0FBQVxzzTXec917770sXbqU5cuXYzKZqKysZPjw4dx7772Ehoby6aef8qc//YmePXsyevRoAO666y6+//57Vq1aRVxcHA8++CA7duzg1FNP9R732muvJTMzk7fffpvExEQ+/PBDzjvvPHbt2kWfPn0a/NwjRozA6XSyZcsWxo8f3ybfZUM0iqIozd15wYIFPPHEE4SFhWEymRg9ejSLFy+mZ8+egNrs8tFHH2E0GgkPD2f8+PE88sgjTVbV2e127Ha7d7msrIykpCRKS0sJDQ09gY/WQh/8GX55G85eAGfc2XHnFUL8YVRWVnLw4EFSU1MJCAjA6nAx4MF1finLnn9Mxmxs3u/LCy64gKioKF555ZVG91mxYgVz5871ThtffT3IyckhJCQEUGu+9+7dS0ZGBlqtWrHer18/5syZw3333dfgcd977z1uueUWCgrU2ufLL7+ciooKVq9e7d3nqquuYvXq1d5zjxs3jilTpjB//nzvPitXruSee+7h2LFjzfrMrTFw4EBuueUW/vrXvwIwZMgQLr74Ym84+r//+z++/PJLtmzZAkCPHj147LHHuOKKK7zHWLRoEWvWrGHTpk1kZmaSmprK008/zR133NHkuc8//3z69+/Pk08+SXl5OVFRUbz55ptcfPHFAJSWlpKYmMiNN97I008/TUZGBn369OHo0aMkJiZ6j3P22WczatQoFi9e3Oi5IiMjeeqpp3wCUm11/55XKysrIywsrFnX7xbVfIwePZrXXnuNtLQ0cnNzWbRoEWPHjmX37t1ERUUxZcoULrnkEpKTkzl48CAPPPAAkyZNYvv27ZhMpgaPuWTJEhYuXNiSYrQPq8xuKoTommw2m89FpLlSUlK8wQMgLi4OnU7nDR7V66qn4wbYsGEDixcvZs+ePZSVleFyuaisrMRisRAUFMTevXu58MILfc4zatQonzCyfft2tm7d6tOk43a7qaysxGq1tskkbxaLhYULF7J69WqOHTuGy+XCZrNx+PBh7z6zZ8/mv//9Lw888ACKovDWW295mz3y8/M5cuQI119/PTfeeKP3PS6Xi7CwMJ9z1a11crvdPProo7zzzjtkZWV5f6QHBQUBcODAAZxOJ6NGjfK+JywsjL59+3qXd+zYgaIopKWl+Rzbbrcfd/bdwMBArFZrM76l1mtR+JgyZYr39aBBgxgzZgy9evXi1Vdf5a677uKyyy7zbj/llFMYMWIEycnJfPrpp8yaNavBY86fP5+77rrLu1xd89HhpMOpEKKNBRp07PnHZL+du7mio6MpLi5u8TkMBoPPskajaXCdx+MB4NChQ0ydOpWbb76Zhx9+mMjISL777juuv/56b0dLRVHqNRfVraD3eDwsXLiwwetKa0JUQ+6++27WrVvHk08+Se/evQkMDOTiiy/G4XB497nyyiu577772LFjBzabjSNHjnD55Zd7ywhq00t1U0m1ulOSV4eKakuXLuWpp57i6aef9vaPmTt3rvfc1d9HU9+Tx+NBp9Oxffv2eucLDm56LquioiJiYtr3WnhC93YJCgpi0KBBpKenN7g9ISGB5OTkRreD2keksVqRDmUpVJ+lw6kQoo1oNJpmN33409ChQ1m5cmW7n2fbtm24XC6WLl3qrR159913ffbp16+ft9mi9vtqGzZsGHv37qV3797tVtZvv/2WOXPmeGthKioq6s2B0r17d8aNG8cbb7yBzWbj7LPPJi4uDlBrfLp168aBAweYPXt2i889c+ZMrrrqKkANEunp6fTv3x+AXr16YTAY2LJli/fHellZGenp6d5+GkOHDsXtdpOXl8eZZ57Z7HNnZGRQWVnJ0KFDW1TmljqhfxV2u53ffvut0Q9WWFjIkSNHTujOdx1CUaTDqRCiy5o8eTLz58+nuLiYiIiIdjtPr169cLlcPPfcc0yfPp3vv/+eF1980Wef2267jXHjxrFs2TKmT5/O+vXrWbt2rc+v/AcffJBp06aRlJTEJZdcglar5ZdffmHXrl0sWrSoTcrau3dvPvjgA6ZPn45Go+GBBx7w1mbUNnv2bBYsWIDD4eCpp57y2bZgwQJuv/12QkNDmTJlCna7nW3btlFcXOxT49/Qud9//302bdpEREQEy5YtIycnxxs+QkJCuOaaa7j77ruJjIwkNjaWhx56CK1W6/2e0tLSmD17NldffTVLly5l6NChFBQUsH79egYNGsTUqVMbPPe3335Lz549vZ1o20uLhtrOmzePr7/+moMHD/Ljjz9y8cUXU1ZWxjXXXENFRQXz5s1j8+bNZGZmsnHjRqZPn050dHS99rtOx1EB7qpOrxI+hBBdzKBBgxgxYkS9Woi2duqpp7Js2TIee+wxTjnlFN544w2WLFnis8/pp5/Oiy++yLJlyxgyZAifffYZd955p09zyuTJk1m9ejVffPEFI0eO5LTTTmPZsmUkJyc3eu4JEyYwZ86cZpf1qaeeIiIigrFjxzJ9+nQmT57MsGHD6u13ySWXUFhYiNVq5YILLvDZdsMNN/Cf//yHFStWMGjQIMaPH8+KFStITU1t8twPPPAAw4YNY/LkyUyYMIH4+Ph6x162bBljxoxh2rRpnH322Zx++uneIb3Vli9fztVXX83f/vY3+vbty4wZM/jxxx+b7Nrw1ltv+fRRaTdKC1x22WVKQkKCYjAYlMTERGXWrFnK7t27FUVRFKvVqpx77rlKTEyMYjAYlB49eijXXHONcvjw4ZacQiktLVUApbS0tEXvOyGFGYryUKiiLIrvuHMKIf5wbDabsmfPHsVms/m7KC326aefKv3791fcbre/i1LPDTfcoJxxxhkndIzk5GRl+fLlbVOgTqiiokIJCwtT/vOf/7T6GLt27VJiY2OVkpKSJvdr7O95S67fLWp2efvttxvdFhgYyLp1/hlSdsKkv4cQooubOnUq6enpZGVl+afTfy1PPvkk55xzDkFBQaxdu5ZXX33VZ3Kulvr9998JCQnh6quvbsNS+tfOnTv5/fffGTVqFKWlpfzjH/8AYObMma0+5rFjx3jttdfqjcZpD52/J1RHkP4eQghx3LkmOsqWLVt4/PHHKS8vp2fPnjz77LNNTv1+PP369as3G/cfwZNPPsnevXsxGo0MHz6cb7/9lujo1l/Hzj333DYsXdMkfECtOT4kfAghhL+1d9+TP4KhQ4eyfft2fxej1eSutiBzfAghhBAdSMIH1Orz0fSsb0IIIYQ4cRI+QGo+hBBCiA4k4QOkw6kQQgjRgSR8gNxUTgghhOhAEj4ALFXhQ/p8CCGEEO1Owoei1IQPqfkQQnRRhYWFxMbG1rt5WnvLzMxEo9Hw008/NbnfhAkTvLerF75GjhzJBx984O9itIiEj8pS8Ki3cpY+H0KIrmrJkiVMnz6dlJQUoCYUdLSNGzei0WgoKSnxWf/BBx/w8MMPe5cnTJjAihUrmn1cf32eE9HYdzFnzhwWLFjgXX7ggQe47777GrzxXWcl4cNaNczWGAyGQP+WRQgh/MBms/HKK6+c0Cyi7S0yMpKQkBB/F6NTOv/88yktLT2pbnEi4UNGugghuri1a9ei1+sZM2ZMo/usWLGC8PBwVq9eTd++fTGbzVx88cVYLBZeffVVUlJSiIiI4LbbbsPtdnvfp9Fo+Oijj3yOFR4e3mCtRWZmJhMnTgQgIiICjUbjvRNtezS7fP/994wfPx6z2UxERASTJ0+muLgYALvdzu23305sbCwBAQGcccYZbN261fve6lqJr776ihEjRmA2mxk7dix79+71OceqVasYMWIEAQEBREdHM2vWLO+2lStXMmLECEJCQoiPj+fKK68kLy/vuN9FXTqdjqlTp/LWW2+15dfTriR8eDubSvgQQrQxRQGHxT8PRWl2Mb/55htGjBhx3P2sVivPPvssb7/9Np999hkbN25k1qxZrFmzhjVr1vD666/z0ksv8b///a9VX1dSUhLvv/8+AHv37iU7O5tnnnmmVcc6np9++omzzjqLgQMHsnnzZr777jumT5/uDU733HMP77//Pq+++io7duygd+/eTJ48maKiIp/j3H///SxdupRt27ah1+u57rrrvNs+/fRTZs2axfnnn8/OnTu9QaWaw+Hg4Ycf5ueff+ajjz7i4MGD3oDR0u9i1KhRfPvtt2319bQ7ubeLTDAmhGgvTissTvTPuf/vGBiDmrVrZmYmiYm+5UxJSUGpE2CcTicvvPACvXr1AuDiiy/m9ddfJzc3l+DgYAYMGMDEiRPZsGEDl112WYuLrNPpiIyMBCA2Npbw8PBG9924cWOLjl338zz++OOMGDHC5265AwcOBMBisfDCCy+wYsUKpkyZAsDLL7/MF198wSuvvMLdd9/tfc8jjzzC+PHjAbjvvvs4//zzqaysJCAggEceeYTLL7+chQsXevcfMmSI93XtoFJ9A71Ro0ZRUVFBcHBwo99FQ7VG3bp14/Dhw3g8HrTazl+v0PlL2N68c3zIMFshRNdks9kICAg47n5ms9kbPADi4uJISUkhODjYZ11100FnVl3z0ZCMjAycTienn366d53BYGDUqFH89ttvPvsOHjzY+zohIQHA+/mbOgfAzp07mTlzJsnJyYSEhDBhwgQADh8+3OLPExgYiMfjwW63t/i9/iA1HzLMVgjRXgxmtQbCX+dupujoaG9fhyYPaTD4LGs0mgbX1R51odFoGqxB8bfAwMYHGFSXt+7oGEVR6q2r/fmrt1V//qbOYbFYOPfcczn33HNZuXIlMTExHD58mMmTJ+NwOFr2YYCioiLMZnOT5+xMpOZD+nwIIdqLRqM2ffjj0YJhpUOHDmXPnj3t8hXExMSQnZ3tXU5PT8dqtTa6v9FoBPDptNoeBg8ezFdffdXgtt69e2M0Gvnuu++865xOJ9u2baN///5tco7ff/+dgoICHn30Uc4880z69etXr8aoJd/Fr7/+yrBhw5pdNn+T8CF9PoQQXdzkyZPZvXt3s2o/WmrSpEk8//zz7Nixg23btnHzzTfXqy2pLTk5GY1Gw+rVq8nPz6eioqLNywQwf/58tm7dyq233sovv/zC77//zgsvvEBBQQFBQUHccsst3H333Xz22Wfs2bOHG2+8EavVyvXXX9/sczz00EO89dZbPPTQQ/z222/s2rWLxx9/HIAePXpgNBp57rnnOHDgAKtWrfKZxwRa9l18++23nHvuua37MvxAwkf1PB/S50MI0UUNGjSIESNG8O6777b5sZcuXUpSUhLjxo3jyiuvZN68eZjNjTcJdevWjYULF3LfffcRFxfHX//612adZ86cOd4+E82RlpbG559/zs8//8yoUaMYM2YMH3/8MXq92hvh0Ucf5aKLLuJPf/oTw4YNY//+/axbt46IiIhmn2PChAm89957rFq1ilNPPZVJkybx448/AmqN0IoVK3jvvfcYMGAAjz76KE8++aTP+5v7XWRlZbFp0yauvfbaZpfN3zRK3cY4PysrKyMsLIzS0lJCQ0Pb/4RPpkFFLvz5G0gYcvz9hRCiEZWVlRw8eJDU1NRmdeDsTNasWcO8efP49ddfT4rREnVNmDCBCRMm+Mz82VXcfffdlJaW8tJLL3XI+Rr7e96S63fX7nDq8UifDyGEAKZOnUp6ejpZWVkkJSX5uzgtUl5eTkZGBqtXr/Z3UfwiNjaWefPm+bsYLdK1w0dlCShVHXlkhlMhRBd3xx13+LsIrRISEsKRI0f8XQy/qT3vyMni5Ktba0vVtR6mMNCb/FsWIYQQoovo2uFDJhgTQgghOlzXDh8yzFYIIYTocF08fEhnUyGEEKKjSfgA6WwqhBBCdKCuHT6sEj6EEEKIjta1w4f0+RBCCCE6XBcPH9LnQwghAAoLC4mNjSUzM7NDz5uZmYlGo+Gnn35qcr8JEyYwd+7cDinTH9Xq1asZOnSoz12H/UXCB0izixCiy1uyZAnTp08nJSUFqAkFHW3jxo1oNBpKSkp81n/wwQc+N16bMGECK1asaPZx/fV5/KnuZ542bRoajYY333zTj6VSde3wIX0+hBACm83GK6+8wg033ODvojQqMjKSkJAQfxfjhDmdTr+e/9prr+W5557zaxmgK4cPj7vWHW2lz4cQoutau3Yter2eMWPGNLrPihUrCA8PZ/Xq1fTt2xez2czFF1+MxWLh1VdfJSUlhYiICG677Tbcbrf3fRqNho8++sjnWOHh4Q3WWmRmZjJx4kQAIiIi0Gg0zJkzB2j7ZpfCwkKuuOIKunfvjtlsZtCgQbz11lve7f/+97/p1q1bvSaKGTNmcM0113iXP/nkE4YPH05AQAA9e/Zk4cKFuFwu73aNRsOLL77IzJkzCQoKYtGiRbjdbq6//npSU1MJDAykb9++PPPMMz7ncblc3H777YSHhxMVFcW9997LNddcwwUXXODdR1EUHn/8cXr27ElgYCBDhgzhf//7X5Ofe8aMGWzZsoUDBw605mtrM1333i62YlCq/lKZZYZTIUTbUxQFm8vml3MH6gOb3czwzTffMGLEiOPuZ7VaefbZZ3n77bcpLy9n1qxZzJo1i/DwcNasWcOBAwe46KKLOOOMM7jssstaXOakpCTef/99LrroIvbu3UtoaCiBgYEtPk5zVFZWMnz4cO69915CQ0P59NNP+dOf/kTPnj0ZPXo0l1xyCbfffjsbNmzgrLPOAqC4uJh169bxySefALBu3Tquuuoqnn32Wc4880wyMjK46aabAHjooYe853rooYdYsmQJTz31FDqdDo/HQ/fu3Xn33XeJjo5m06ZN3HTTTSQkJHDppZcC8Nhjj/HGG2+wfPly+vfvzzPPPMNHH33kDWcAf//73/nggw944YUX6NOnD9988w1XXXUVMTExjB8/vsHPnZycTGxsLN9++y09e/Zsl++2Obpu+Kju7xEQDjqDX4sihPhjsrlsjH5ztF/O/eOVP2I2mJu1b2ZmJomJiT7rUlJSUBTFZ53T6eSFF16gV69eAFx88cW8/vrr5ObmEhwczIABA5g4cSIbNmxoVfjQ6XRERkYC6p1aw8PDG91348aNLTp23c/TrVs3nzvB3nbbbXz22We89957jB49msjISM477zzefPNNb/h47733iIyM9C4/8sgj3Hfffd6akJ49e/Lwww9zzz33+ISPK6+8kuuuu86nPAsXLvS+Tk1NZdOmTbz77rve8PHcc88xf/58LrzwQgCef/551qxZ432PxWJh2bJlrF+/3ltj1bNnT7777jv+/e9/M378+Ab/DKs/e0d3LK6rC4eP6mG20t9DCNG12Ww2AgICjruf2Wz2Bg+AuLg4UlJSCA4O9lmXl5fXLuVsS263m0cffZR33nmHrKws7HY7drudoKAg7z6zZ8/mpptu4l//+hcmk4k33niDyy+/HJ1OB8D27dvZunUrjzzyiM9xKysrsVqtmM1q+GuoVunFF1/kP//5D4cOHcJms+FwODj11FMBKC0tJTc3l1GjRnn31+l0DB8+3NsMtGfPHiorKznnnHN8jutwOBg6dGiTnz0wMBCr1dqCb6vtdd3w4e1sKv09hBDtI1AfyI9X/ui3czdXdHQ0xcXFx93PYPCtJdZoNA2uq91PQqPRNFiD4m9Lly7lqaee4umnn2bQoEEEBQUxd+5cHA6Hd5/p06fj8Xj49NNPGTlyJN9++y3Lli3zbvd4PCxcuJBZs2bVO37tMFc70AC8++673HnnnSxdupQxY8YQEhLCE088wY8/+v5dqdtsVvt7rP6OP/30U7p16+azn8nU9F3ai4qKiInx77Wv64YPGWYrhGhnGo2m2U0f/jR06FBWrlzZLseOiYkhOzvbu5yent7kr26j0Qjg02m1PXz77bfMnDmTq666ClAv5unp6fTv39+7T2BgILNmzeKNN95g//79pKWlMXz4cO/2YcOGsXfvXnr37t3ic48dO5Zbb73Vuy4jI8P7OiwsjLi4OLZs2cKZZ54JqN/Hzp07vbUjAwYMwGQycfjw4Ub7dzSksrKSjIyM49aOtDcJHzLBmBCii5s8eTLz58+nuLiYiIiINj32pEmTeP755znttNPweDzce++99WpLaktOTkaj0bB69WqmTp1KYGCgT7NOW+nduzfvv/8+mzZtIiIigmXLlpGTk+MTPkBtepk+fTq7d+/2BpVqDz74INOmTSMpKYlLLrkErVbLL7/8wq5du1i0aFGT537ttddYt24dqampvP7662zdupXU1FTvPrfddhtLliyhd+/e9OvXj+eee47i4mJvbUhISAjz5s3jzjvvxOPxcMYZZ1BWVsamTZsIDg72GZFT2w8//IDJZGpyZFNH6LpDbWVqdSGEAGDQoEGMGDGCd999t82PvXTpUpKSkhg3bhxXXnkl8+bN8/aFaEi3bt1YuHAh9913H3Fxcfz1r39t1nnmzJnDhAkTml2uBx54gGHDhjF58mQmTJhAfHy8zzDWapMmTSIyMpK9e/dy5ZVX+mybPHkyq1ev5osvvmDkyJGcdtppLFu2jOTk5CbPffPNNzNr1iwuu+wyRo8eTWFhoU8tCMC9997LFVdcwdVXX82YMWMIDg5m8uTJPs05Dz/8MA8++CBLliyhf//+TJ48mU8++cQnxNT11ltvMXv27Cb/DDqCRmmoK6wflZWVERYWRmlpKaGhoe13onevhj0fw5THYfSf2+88Qoguo7KykoMHD5KamtqsDpydyZo1a5g3bx6//vorWu3J97t0woQJTJgwgQULFvi7KO3C4/HQv39/Lr30Up+ZXlsiPz+ffv36sW3btiYDyvE09ve8JddvaXaRPh9CCMHUqVNJT08nKyuLpKQkfxenRcrLy8nIyGD16tX+LkqbOXToEJ9//jnjx4/Hbrfz/PPPc/DgwXq1Ly1x8OBB/vWvf51Q8GgrEj6kz4cQQgBwxx13+LsIrRISEsKRI0f8XYw2pdVqWbFiBfPmzUNRFE455RS+/PLLen1SWmLUqFE+w3f9qQuHD+nzIYQQonNKSkri+++/93cx2s3J17DXFtwudXp1kGYXIYQQooN1zfBhKwIUQAOBkf4ujRBCCNGldM3wUd3fIzACdF235UkIIYTwhy4aPqS/hxBCCOEvXTx8SH8PIYQQoqN1zfBhLVSfJXwIIYQQHa5rho/qmg+Z40MIIQAoLCwkNjaWzMzMDj1vZmYmGo2Gn376qcn9JkyYwNy5czukTP6WkpLC008/3W7HnzdvHrfffnu7Hb85umj4qJ7dVPp8CCEEwJIlS5g+fTopKSlATSjoaBs3bkSj0VBSUuKz/oMPPvCZVnzChAmsWLGi2cf11+fpDFasWOFz35t77rmH5cuXc/DgQb+VqYuGD+nzIYQQ1Ww2G6+88go33HCDv4vSqMjISEJCQvxdjD+E2NhYzj33XF588UW/laFF4WPBggVoNBqfR3x8vHe7oigsWLCAxMREAgMDmTBhArt3727zQp8w6fMhhBBea9euRa/XN3mb9RUrVhAeHs7q1avp27cvZrOZiy++GIvFwquvvkpKSgoRERHcdtttuN1u7/s0Gg0fffSRz7HCw8MbrLXIzMxk4sSJAERERKDRaJgzZw7QPs0u33//PePHj8dsNhMREcHkyZMpLlYnoLTb7dx+++3ExsYSEBDAGWecwdatW73vra6hWbduHUOHDiUwMJBJkyaRl5fH2rVr6d+/P6GhoVxxxRVYrVbv+yZMmMBf//pX/vrXvxIeHk5UVBR///vfaeoer6Wlpdx0003ExsYSGhrKpEmT+PnnnwH1ZnHx8fEsXrzYu/+PP/6I0Wjk888/b/SYM2bM4K233mr1d3eiWjzJxcCBA/nyyy+9yzqdzvv68ccfZ9myZaxYsYK0tDQWLVrEOeecw969eztXYpWhtkKIDqAoCorN5pdzawIDm93M8M033zBixIjj7me1Wnn22Wd5++23KS8vZ9asWcyaNYvw8HDWrFnDgQMHuOiiizjjjDO47LLLWlzmpKQk3n//fS666CL27t1LaGgogYGBLT5Oc/z000+cddZZXHfddTz77LPo9Xo2bNjgDU733HMP77//Pq+++irJyck8/vjjTJ48mf379xMZWTM55YIFC3j++ecxm81ceumlXHrppZhMJt58800qKiq48MILee6557j33nu973n11Ve5/vrr+fHHH9m2bRs33XQTycnJ3HjjjfXKqSgK559/PpGRkaxZs4awsDD+/e9/c9ZZZ7Fv3z5iYmL473//ywUXXMC5555Lv379uOqqq7j11ls599xzG/38o0aN4siRIxw6dIjk5OQ2/Gabp8XhQ6/X+9R2VFMUhaeffpr777+fWbNmAeoXHBcXx5tvvsmf/9yJblsvN5UTQnQAxWZj77Dhfjl33x3b0ZjNzdo3MzOTxMREn3UpKSn1fo07nU5eeOEFevXqBcDFF1/M66+/Tm5uLsHBwQwYMICJEyeyYcOGVoUPnU7nvbDHxsYSHh7e6L4bN25s0bHrfp7HH3+cESNG8K9//cu7buDAgQBYLBZeeOEFVqxYwZQpUwB4+eWX+eKLL3jllVe4++67ve9ZtGgRp59+OgDXX3898+fPJyMjg549ewLqd7Rhwwaf8JGUlMRTTz2FRqOhb9++7Nq1i6eeeqrB8LFhwwZ27dpFXl4eJpMJgCeffJKPPvqI//3vf9x0001MnTqVG2+8kdmzZzNy5EgCAgJ49NFHvceYM2eOtwapWrdu3QD1z94f4aPFfT7S09NJTEwkNTWVyy+/nAMHDgDqrXpzcnJ8kpbJZGL8+PFs2rSp0ePZ7XbKysp8Hu3K7YTKEvW11HwIIQQ2m42AgIDj7mc2m73BAyAuLo6UlBSCg4N91uXl5bVLOdtSdc1HQzIyMnA6nd5QAWAwGBg1ahS//fabz76DBw/2vo6Li8NsNnuDR/W6ut/Haaed5lMrNWbMGNLT032aq6pt376diooKoqKiCA4O9j4OHjxIRkaGd78nn3wSl8vFu+++yxtvvHHcP8/qGqXaTUIdqUU1H6NHj+a1114jLS2N3NxcFi1axNixY9m9ezc5OTmA+kXXFhcXx6FDhxo95pIlS1i4cGErit5K1f09NFp1enUhhGgnmsBA+u7Y7rdzN1d0dLS3r0NTDAaD7zk0mgbXeTwen+WGalD8ranmnOry1m22UhSl3rran78530dLeTweEhISGqzpqV0zdODAAY4dO4bH4+HQoUM+oaghRUVFAMTE+OdHeIvCR3X1E8CgQYMYM2YMvXr14tVXX+W0004DmveHVdv8+fO56667vMtlZWUkJSW1pFgt453jIwq0XXOwjxCiY2g0mmY3ffjT0KFDWblyZbscOyYmhuzsbO9yenp6k7+2jUYjQIO1AG1p8ODBfPXVVw3++O3duzdGo5HvvvuOK6+8ElAD07Zt29qk0+sPP/xQb7lPnz4+fSirDRs2jJycHPR6vXcYdF0Oh4PZs2dz2WWX0a9fP66//np27dpVrzKgtl9//RWDweBtaupoJ3T1DQoKYtCgQaSnp3v7gVTXgFTLy8tr8gswmUyEhob6PNqV9PcQQggfkydPZvfu3c2q/WipSZMm8fzzz7Njxw62bdvGzTffXK92oLbk5GQ0Gg2rV68mPz+fioqKNi8TqD98t27dyq233sovv/zC77//zgsvvEBBQQFBQUHccsst3H333Xz22Wfs2bOHG2+8EavVyvXXX3/C5z5y5Ah33XUXe/fu5a233uK5557jjjvuaHDfs88+mzFjxnDBBRewbt06MjMz2bRpE3//+9/Ztm0bAPfffz+lpaU8++yz3HPPPfTv3/+45fz2228588wz261D7/GcUPiw2+389ttvJCQkkJqaSnx8PF988YV3u8Ph4Ouvv2bs2LEnXNA2451gTMKHEEKAWpM9YsQI3n333TY/9tKlS0lKSmLcuHFceeWVzJs3D3MTtUHdunVj4cKF3HfffcTFxfHXv/61WeeZM2eOz0Rax5OWlsbnn3/Ozz//zKhRoxgzZgwff/wxer3aIPDoo49y0UUX8ac//Ylhw4axf/9+1q1bR0TEiTfXX3311dhsNkaNGsVf/vIXbrvtNm666aYG99VoNKxZs4Zx48Zx3XXXkZaWxuWXX05mZiZxcXFs3LiRp59+mtdff53Q0FC0Wi2vv/463333HS+88EKjZXjrrbca7ODaYZQW+Nvf/qZs3LhROXDggPLDDz8o06ZNU0JCQpTMzExFURTl0UcfVcLCwpQPPvhA2bVrl3LFFVcoCQkJSllZWbPPUVpaqgBKaWlpS4rWfJv+qSgPhSrKu9e0z/GFEF2WzWZT9uzZo9hsNn8XpcU+/fRTpX///orb7fZ3UVpl/PjxykMPPeTvYhzX+PHjlTvuuMOvZVi9erXSv39/xel0tur9jf09b8n1u0V9Po4ePcoVV1xBQUEBMTExnHbaafzwww/eYTr33HMPNpuNW2+9leLiYkaPHs3nn3/eueb4sMrU6kIIUdfUqVNJT08nKyurffvdtYPy8nIyMjJYvXq1v4tyUrBYLCxfvtxby+MPLTrz22+/3eR2jUbDggULWLBgwYmUqX3JTeWEEKJBjfU76OxCQkI4cuSIv4tx0rj00kv9XYSWTzJ20rPI1OpCCCH8o6WTo/1Rdb2xpjK1uhBCCOFXXS98WGW0ixBCCOFPXS98WKTDqRBCCOFPXSt8uOxgr7p3jDnKv2URQgghuqiuFT6qaz20eggI92tRhBBCiK6qa4WP6v4ecl8XIYQQwm+61hVYRroIIUSDCgsLiY2NJTMz099FEa20evVqhg4dekJ30e0oXSx8VM3xIf09hBDCx5IlS5g+fbr3zqmZmZk+dyRfsWKFzy3c20pjx50wYQIrVqxo9nHqlrcrqPuZp02bhkaj4c033/RjqZqni4UPqfkQQoi6bDYbr7zyCjfccIO/i3LSczqdfj3/tddey3PPPefXMjRH1wofMseHEELUs3btWvR6PWPGjGlw+8aNG7n22mspLS1Fo9F4b6UB6t3L77nnHrp160ZQUBCjR4/2zuJZWVnJwIEDfe7YevDgQcLCwnj55ZebPO6JKiws5IorrqB79+6YzWYGDRrEW2+95d3+73//m27dutVropgxYwbXXHONd/mTTz5h+PDhBAQE0LNnTxYuXIjL5fJu12g0vPjii8ycOZOgoCAWLVqE2+3m+uuvJzU1lcDAQPr27cszzzzjcx6Xy8Xtt99OeHg4UVFR3HvvvVxzzTVccMEF3n0UReHxxx+nZ8+eBAYGMmTIEP73v/81+blnzJjBli1bOHDgQGu+to7TqlvataN2vavtR7eqd7T9+vG2P7YQosure7dPj8ejOCpdfnl4PJ5ml/uOO+5QzjvvPJ91Bw8eVKovEXa7XXn66aeV0NBQJTs7W8nOzlbKy8sVRVGUK6+8Uhk7dqzyzTffKPv371eeeOIJxWQyKfv27VMURVF27typGI1G5cMPP1RcLpdy+umnKzNnzjzuccePH68sX7682Z+hdnkVRVGOHj2qPPHEE8rOnTuVjIwM5dlnn1V0Op3yww8/KIqiKIWFhYrRaFS+/PJL73uKiooUo9GorFu3TlEURfnss8+U0NBQZcWKFUpGRoby+eefKykpKcqCBQu87wGU2NhY5ZVXXlEyMjKUzMxMxeFwKA8++KCyZcsW5cCBA8rKlSsVs9msvPPOO973LVq0SImMjFQ++OAD5bffflNuvvlmJTQ01PvdKIqi/N///Z/Sr18/5bPPPlMyMjKU5cuXKyaTSdm4cWODn7labGyssmLFimZ/dy3VFne11SiKovgz/NRVVlZGWFgYpaWlhIaGtu3B37wM9n0G056GEde27bGFEF1eZWUlBw8eJDU1lYCAAJx2Ny/d8bVfynLTM+MxmHTN2veCCy4gKiqKV155pdF9VqxYwdy5cykpKfGuy8jIoE+fPhw9epTExETv+rPPPptRo0axePFiAJ544gkef/xxrrjiCt577z127dpFdHR0o8dtL+effz79+/fnySefBGDmzJlER0d7P/dLL73EQw89xNGjR9HpdIwbN44pU6Ywf/587zFWrlzJPffcw7FjxwC15mPu3Lk89dRTTZ77L3/5C7m5ud6ai/j4eObNm8e8efMAcLvd9OzZk6FDh/LRRx9hsViIjo5m/fr1PjVSN9xwA1artcl+HcOGDWPmzJk89NBDrfiWjq/u3/NqLbl+d60by8nspkIIUY/NZvO5iDTXjh07UBSFtLQ0n/V2u52oqJqO/X/729/4+OOPee6551i7dq03eLQnt9vNo48+yjvvvENWVhZ2ux273U5QUJB3n9mzZ3PTTTfxr3/9C5PJxBtvvMHll1+OTqeGtu3bt7N161YeeeQRn+NWVlZitVoxm80AjBgxot75X3zxRf7zn/9w6NAhbDYbDoeDU089FYDS0lJyc3MZNWqUd3+dTsfw4cO9zUB79uyhsrKSc845x+e4DoeDoUOHNvnZAwMDsVqtLfi2Ol4XCx/VHU6lz4cQov3pjVpuema8387dXNHR0RQXF7f4HB6PB51Ox/bt270X7GrBwcHe13l5eezduxedTkd6ejrnnXdei8/VUkuXLuWpp57i6aefZtCgQQQFBTF37lwcDod3n+nTp+PxePj0008ZOXIk3377LcuWLfP5fAsXLmTWrFn1jl87rNUONADvvvsud955J0uXLmXMmDGEhITwxBNP8OOPP/rsV3d0Tu2GiOoQ8umnn9KtWzef/UwmU5OfvaioiJiYzv0ju2uFD2vVUFup+RBCdACNRtPspg9/Gjp0KCtXrmxyH6PRiNvtrvc+t9tNXl4eZ555ZqPvve666zjllFO48cYbuf766znrrLMYMGBAo8dtC99++y0zZ87kqquuAtSLeXp6Ov379/fuExgYyKxZs3jjjTfYv38/aWlpDB8+3Lt92LBh7N27l969e7f43GPHjuXWW2/1rsvIyPC+DgsLIy4uji1btni/N7fbzc6dO721IwMGDMBkMnH48GHGj29+gK2srCQjI+O4tSP+1nXCh9MGjgr1tdR8CCGE1+TJk5k/fz7FxcVEREQ0uE9KSgoVFRV89dVXDBkyBLPZTFpaGrNnz+bqq69m6dKlDB06lIKCAtavX8+gQYOYOnUq//znP9m8eTO//PILSUlJrF27ltmzZ/Pjjz9iNBobPG51c8aJ6N27N++//z6bNm0iIiKCZcuWkZOT4xM+QG16mT59Ort37/YGlWoPPvgg06ZNIykpiUsuuQStVssvv/zCrl27WLRoUZPnfu2111i3bh2pqam8/vrrbN26ldTUVO8+t912G0uWLKF3797069eP5557juLiYm9tSEhICPPmzePOO+/E4/n/9u49psn7+wP4u4WWm4DOgW3HdWSDOQhRlAGiGM1AnJPNRWAuCLuqkU2mRsgugWR/jN1ctjHGXBBcds0GGDP8qhABnYjDWScqQwIdkgFDyQQm4WbP74/vr80KvQD28rTf80qaSPv5fDzH00+eQ32ePmrEx8djaGgIjY2NmDdvns4VOf/W1NQEFxcXg1cuCYYlzoS9Gxa72mXsn/9e5fKfPKJZnAXOGGMzZegqAHsQExNDJSUlRsds376dFi5cSAAoPz+fiEh7ZUdQUBBJJBKSyWT05JNP0qVLl6i1tZXc3Nzom2++0a4xODhIQUFBtG/fPqPrTpWZmUkJCQkzzmdgYIBSUlJo3rx55OvrS2+88QZt3bpV52oSIqLJyUmSy+UEgDo6Oqatc+zYMYqLiyM3Nzfy8vKi6OhoOnDggPZ1AFRVVaUzZ3R0lLKyssjb25vmz59PO3bsoLy8PIqMjNSOmZiYoOzsbPLy8qIFCxZQbm4ubd68mdLT07Vj1Go1ffTRRxQaGkoSiYR8fHwoKSmJGhoaDOb90ksv0bZt22b87zQXfLULY4wJiKGrAOzB0aNHsXfvXly+fBliAd77avXq1Vi9erXZvgdEaNRqNR566CGkpqbirbfemtMaN27cQFhYGM6fP6/zKYu58dUujDHGzGL9+vVob2/Hn3/+CX9/f1uHo2N4eBgdHR346aefbB2K2XR1deHEiRNISEjA2NgYioqKoFKpsGXLljmvqVKpUFxcbNHGw1y4+WCMMQYA2LVrl61D0MvT0xPd3d22DsOsxGIxysvLsXfvXhARwsPDUVtbO+2clNmIjo7WuXxXyLj5YIwxxqzM398fZ86csXUYNiO8/9hjjDHGmEPj5oMxxhhjVsXNB2OMmdnUO6Uy5kjM8f7mcz4YY8xMpFIpxGIxenp64OPjA6lUOu0rtBmzV0SE8fFx3LhxA2KxGFKpdM5rcfPBGGNmIhaLERwcjN7eXu1dTxlzNO7u7ggICLir74Ph5oMxxsxIKpUiICAAk5OTFrlnCWO25OTkBGdn57v+RI+bD8YYMzORSASJRAKJRGLrUBgTJD7hlDHGGGNWxc0HY4wxxqyKmw/GGGOMWZXgzvnQ3GR3aGjIxpEwxhhjbKY0x23NcdwYwTUfw8PDACC4uyoyxhhjzLTh4WF4e3sbHSOimbQoVqRWq9HT0wNPT0+zfznP0NAQ/P390d3dDS8vL7OuLQSOnh/g+DlyfvbP0XN09PwAx8/RUvkREYaHh6FQKEx+B4jgPvkQi8Xw8/Oz6N/h5eXlkG8oDUfPD3D8HDk/++foOTp6foDj52iJ/Ex94qHBJ5wyxhhjzKq4+WCMMcaYVf1PNR8uLi7Iz8+Hi4uLrUOxCEfPD3D8HDk/++foOTp6foDj5yiE/AR3wiljjDHGHNv/1CcfjDHGGLM9bj4YY4wxZlXcfDDGGGPMqrj5YIwxxphVOVzzUVxcjODgYLi6uiIqKgqnT582Or6hoQFRUVFwdXXF/fffj5KSEitFOjtvv/02li9fDk9PT/j6+uKJJ55AW1ub0Tn19fUQiUTTHr///ruVop6dgoKCabHKZDKjc+ylfgAQFBSktx47d+7UO94e6nfq1Ck8/vjjUCgUEIlEOHz4sM7rRISCggIoFAq4ublh9erVuHLlisl1KyoqsHjxYri4uGDx4sWoqqqyUAbGGctvYmICubm5iIiIgIeHBxQKBbZu3Yqenh6ja5aXl+ut6+joqIWzmc5U/bKysqbFGRMTY3JdodQPMJ2jvlqIRCK89957BtcUSg1nclwQ6h50qObj+++/R05ODl5//XUolUqsXLkSycnJuH79ut7xKpUK69evx8qVK6FUKvHaa6/hlVdeQUVFhZUjN62hoQE7d+5EU1MTampqMDk5icTERNy+fdvk3La2NvT29mofDzzwgBUinpuHH35YJ9aWlhaDY+2pfgDQ3Nysk1tNTQ0AYPPmzUbnCbl+t2/fRmRkJIqKivS+/u6772L//v0oKipCc3MzZDIZHn30Ue09nPQ5e/Ys0tLSkJGRgd9++w0ZGRlITU3FuXPnLJWGQcbyGxkZwYULF/Dmm2/iwoULqKysxLVr17Bx40aT63p5eenUtLe3F66urpZIwShT9QOAdevW6cR59OhRo2sKqX6A6Ryn1uHgwYMQiUR46qmnjK4rhBrO5Lgg2D1IDiQ6Opq2b9+u81xYWBjl5eXpHb9v3z4KCwvTeW7btm0UExNjsRjNpb+/nwBQQ0ODwTF1dXUEgP7++2/rBXYX8vPzKTIycsbj7bl+RES7du2ikJAQUqvVel+3t/oBoKqqKu3ParWaZDIZFRYWap8bHR0lb29vKikpMbhOamoqrVu3Tue5pKQkSk9PN3vMszE1P31++eUXAkBdXV0Gx5SVlZG3t7d5gzMDffllZmZSSkrKrNYRav2IZlbDlJQUWrNmjdExQq3h1OOCkPegw3zyMT4+jl9//RWJiYk6zycmJqKxsVHvnLNnz04bn5SUhPPnz2NiYsJisZrD4OAgAOCee+4xOXbJkiWQy+VYu3Yt6urqLB3aXWlvb4dCoUBwcDDS09PR2dlpcKw91298fBxfffUVnnvuOZM3ULSn+v2bSqVCX1+fTo1cXFyQkJBgcE8ChutqbI5QDA4OQiQSYf78+UbH/fPPPwgMDISfnx82bNgApVJpnQDnoL6+Hr6+vnjwwQfx4osvor+/3+h4e67fX3/9herqajz//PMmxwqxhlOPC0Legw7TfNy8eRN37tzBokWLdJ5ftGgR+vr69M7p6+vTO35ychI3b960WKx3i4iwe/duxMfHIzw83OA4uVyOAwcOoKKiApWVlQgNDcXatWtx6tQpK0Y7c4888gi+/PJLHD9+HF988QX6+voQFxeHgYEBvePttX4AcPjwYdy6dQtZWVkGx9hb/abS7LvZ7EnNvNnOEYLR0VHk5eVhy5YtRm/WFRYWhvLychw5cgTffvstXF1dsWLFCrS3t1sx2plJTk7G119/jZMnT+KDDz5Ac3Mz1qxZg7GxMYNz7LV+AHDo0CF4enpi06ZNRscJsYb6jgtC3oOCu6vt3Zr6WyQRGf3NUt94fc8LSXZ2Ni5duoSff/7Z6LjQ0FCEhoZqf46NjUV3dzfef/99rFq1ytJhzlpycrL2zxEREYiNjUVISAgOHTqE3bt3651jj/UDgNLSUiQnJ0OhUBgcY2/1M2S2e3Kuc2xpYmIC6enpUKvVKC4uNjo2JiZG56TNFStWYOnSpfjkk0/w8ccfWzrUWUlLS9P+OTw8HMuWLUNgYCCqq6uNHqDtrX4aBw8exDPPPGPy3A0h1tDYcUGIe9BhPvm499574eTkNK0z6+/vn9bBachkMr3jnZ2dsXDhQovFejdefvllHDlyBHV1dfDz85v1/JiYGEH+hqWPh4cHIiIiDMZrj/UDgK6uLtTW1uKFF16Y9Vx7qp/mSqXZ7EnNvNnOsaWJiQmkpqZCpVKhpqZm1rcoF4vFWL58uV3UVS6XIzAw0Gis9lY/jdOnT6OtrW1O+9LWNTR0XBDyHnSY5kMqlSIqKkp7BYFGTU0N4uLi9M6JjY2dNv7EiRNYtmwZJBKJxWKdCyJCdnY2KisrcfLkSQQHB89pHaVSCblcbuboLGNsbAytra0G47Wn+v1bWVkZfH198dhjj816rj3VLzg4GDKZTKdG4+PjaGhoMLgnAcN1NTbHVjSNR3t7O2pra+fU9BIRLl68aBd1HRgYQHd3t9FY7al+/1ZaWoqoqChERkbOeq6tamjquCDoPWi2U1cF4LvvviOJREKlpaV09epVysnJIQ8PD/rjjz+IiCgvL48yMjK04zs7O8nd3Z1effVVunr1KpWWlpJEIqEff/zRVikYtGPHDvL29qb6+nrq7e3VPkZGRrRjpub34YcfUlVVFV27do0uX75MeXl5BIAqKipskYJJe/bsofr6eurs7KSmpibasGEDeXp6OkT9NO7cuUMBAQGUm5s77TV7rN/w8DAplUpSKpUEgPbv309KpVJ7tUdhYSF5e3tTZWUltbS00NNPP01yuZyGhoa0a2RkZOhckXbmzBlycnKiwsJCam1tpcLCQnJ2dqampiZB5TcxMUEbN24kPz8/unjxos6+HBsbM5hfQUEBHTt2jDo6OkipVNKzzz5Lzs7OdO7cOUHlNzw8THv27KHGxkZSqVRUV1dHsbGxdN9999lN/YhMv0eJiAYHB8nd3Z0+++wzvWsItYYzOS4IdQ86VPNBRPTpp59SYGAgSaVSWrp0qc6lqJmZmZSQkKAzvr6+npYsWUJSqZSCgoIMvvlsDYDeR1lZmXbM1PzeeecdCgkJIVdXV1qwYAHFx8dTdXW19YOfobS0NJLL5SSRSEihUNCmTZvoypUr2tftuX4ax48fJwDU1tY27TV7rJ/mcuCpj8zMTCL676V++fn5JJPJyMXFhVatWkUtLS06ayQkJGjHa/zwww8UGhpKEomEwsLCbNZwGctPpVIZ3Jd1dXXaNabml5OTQwEBASSVSsnHx4cSExOpsbHR+smR8fxGRkYoMTGRfHx8SCKRUEBAAGVmZtL169d11hBy/YhMv0eJiD7//HNyc3OjW7du6V1DqDWcyXFBqHtQ9P8JMMYYY4xZhcOc88EYY4wx+8DNB2OMMcasipsPxhhjjFkVNx+MMcYYsypuPhhjjDFmVdx8MMYYY8yquPlgjDHGmFVx88EYY4wxq+LmgzHGGGNWxc0HY4wxxqyKmw/GGGOMWRU3H4wxxhizqv8DP9ACaRfBvGUAAAAASUVORK5CYII=",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "val_img_txt\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": [
      "val_txt_only\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": [
      "val_unseen_img\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAh8AAAGdCAYAAACyzRGfAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjcuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/bCgiHAAAACXBIWXMAAA9hAAAPYQGoP6dpAACgDElEQVR4nOzdd3hUVfrA8e9kJplk0nsjpJAEQu+hqARUqqAiNrBgVyyru1jQdRdXxbKCrPpTV9cVFBvWVaSJgnQIvYcQEgiQ3pNJpt7fH5NMMpBABlLAvJ/nyUMy9869Z4Zk7nvf855zVIqiKAghhBBCtBGX9m6AEEIIIToWCT6EEEII0aYk+BBCCCFEm5LgQwghhBBtSoIPIYQQQrQpCT6EEEII0aYk+BBCCCFEm5LgQwghhBBtStPeDTid1Wrl1KlTeHt7o1Kp2rs5QgghhGgGRVGoqKggIiICF5ez5zYuuuDj1KlTREVFtXczhBBCCHEesrOz6dSp01n3ueiCD29vb8DWeB8fn3ZujRBCCCGao7y8nKioKPt1/GwuuuCjrqvFx8dHgg8hhBDiEtOckgkpOBVCCCFEm5LgQwghhBBtSoIPIYQQQrQpCT6EEEII0aYk+BBCCCFEm3Iq+DCbzfz1r38lNjYWDw8P4uLi+Mc//oHVarXvoygKs2fPJiIiAg8PD1JSUti/f3+LN1wIIYQQlyango/XXnuN999/n3feeYeDBw/y+uuv889//pO3337bvs/rr7/OvHnzeOedd0hNTSUsLIyrr76aioqKFm+8EEIIIS49TgUfmzZt4tprr2XChAnExMQwZcoURo8ezbZt2wBb1mP+/Pk899xzTJ48mZ49e7Jw4UL0ej2ff/55q7wAIYQQQlxanAo+LrvsMn799VcOHz4MwO7du1m/fj3jx48HIDMzk9zcXEaPHm1/jlarZcSIEWzcuLHRYxoMBsrLyx2+hBBCCPHH5dQMp08//TRlZWV069YNtVqNxWLh5Zdf5tZbbwUgNzcXgNDQUIfnhYaGcuzYsUaP+corr/DCCy+cT9uFEEIIcQlyKvPx1VdfsWjRIj7//HN27NjBwoULeeONN1i4cKHDfqdPraooSpPTrc6aNYuysjL7V3Z2tpMvQQghhBCXEqcyH08++STPPPMMt9xyCwC9evXi2LFjvPLKK9x5552EhYUBtgxIeHi4/Xn5+flnZEPqaLVatFrt+bZfCCGEEJcYpzIfer0eFxfHp6jVavtQ29jYWMLCwvjll1/s241GI7///jvDhg1rgeYKIeoUnaxk58rjGKvN7d0UIYRwilOZj4kTJ/Lyyy/TuXNnevTowc6dO5k3bx533303YOtuefzxx5kzZw4JCQkkJCQwZ84cdDodU6dObZUXIERHdHDjKX7//DAWs5WTh0uYMKM3KpdzryQphBAXA6eCj7fffpvnn3+eGTNmkJ+fT0REBA888AB/+9vf7Ps89dRTVFdXM2PGDEpKSkhOTmblypV4e3u3eOOF6GjMJgvrvkrnwPpT9seO7Sti+/JjDBwf034NE0IIJ6gURVHauxENlZeX4+vrS1lZGT4+Pu3dHCEuGuVF1az4YB/5xypABckTY9H5aln96SFUKpj4p75EdQto72YKITooZ67fsraLEJeA7APFfD1nG/nHKtB6aph4q46BPj/Qvfg1krobURT45aP9VJYY2rupQghxTk51uwgh2pZiVdi+PIstP2WCAsFeBYz1fQ2f3zLs+1yhfE6B62sUVsSw4u11XPfM5ajdZASZEOLiJd0uQlyMSrIwHFrHqp8Vsgo7AdDd4xcu9/kQjcoEWh+IuQz8oiHtZ8oKjSwuegOj4kkfn5Vcdnk19LwBooeDi7qdX0y9guMV7Fh5DJPB0irHD+7szcBxMag1F39SV1EUdq3Kpqygmu7DwwmJ7rifdyajhaw9hZw4WExAhBc9Lo9A43bx/N6K5nHm+i3BhxAXg8p8yFwLR9dA5loKC2BZ6dOUW8JRY+QK/4/pnlQDsSNsX+F9QF2buFQUOJHK0V/WsmxrfwDG+P2TePeN4BUGPSfbApHIAdDEZH9toeB4BT+8ubPVhwZHdvVn3AM90epcW/U8F8JisvLrJwdJT82zPxYe70vvkVHE9Q3CRX3xB08XymK2kn2gmMOpeWTuKcTcICD19HVj4IRYkoaFXxKBpLCR4EOclaIoVK1dS9n/fsQ1KgqfCeNxT0xs72Z1LNWlcGxDbcDxOxQctG9Kqx7BmvKHMCtavD1rGDvFi5BByaA5d1fKxm/T2flLNq5qEzeGPY+/Ja1+o3+MLQjpOQVCu7f8azqLopOV/DBvJzVVJsLjfUkaFn7uJznJoDez9adMTAYL/uGeXPNIb3wCPc79REUBQzlUFUJVQf1XZYPvq0vA3Qc8g8EzBDyDar+v/fIKBne/ZgV3NZUmlr6/h5wjZbi4qIjqEUD2/mKsVttHsZe/ll4pneh+WQTunhdvAHU+rFaFU4dLSN+WT8bOfAxV9YGod6A7Mb2CyNxTQGWxrXbJJ8idQdfEkjg4DJfzGUquL4as9ZD5OxSk2f6vW4iiwInyaA4W9ACNBwGhrgRGBxPQNQ6fmC6oNB0vcyPBh2hS1datFMz/F9U7djg8rk1IwGfCeHzGj8etc+d2at0fmFEP2Zvrg42cXaBYHXaxhPRlfcU97Mu0dbN07hHA1Xf1wN2r+Rcgq8XK/+bv4lR6KQHhOqbcUILroW8gbSmY9PU7BidBrxtswUhAXEu8wiaV5un5bu4OqsuNhMb6MOmxvrh5tE65WUF2BT+/s5uqMiM6bzUTbvEixLfEFlhU5tcGEw2DjNrvLS1QqOuiqQ1GTgtMGnyVGfxZ8lUNpYUm3NzVjH2gF1FJAVSVGti39iT7152kusIEgMbVhcQhYfQe2YnACK8Lb187URSFvMxy0rflcWR7Pvoyo32bzseN+IEhJAwMJTTWB5VKhcVkZf/6k2xbdozqctu+/mE6Bk+Mo0u/4LPPZ2OohOObIXON7W8tZw/Qspc4k+JGWnUKe/QTKDE3/lmpUdUQ4FFEYICRgDAdgbGhBHRLRBcegcrlj5vJkeBDnKF6zx4K5v+LqtrVhVWuavx6eWHSq6g6Uo5irr8Quvfujc/4cfiMG49raEh7NdnOajRSvXMX+i2b0e/ciSY4GM8hQ/EckoxrRER7N69p+mI48D/Y/z0c3wQWo+P2wPjabpQrqPRLZvmik+Rl2lZ1HjQhhoETYs/rbq+qzMDil1PRlxtJTA7lqundUZn0cHg57P0Wjvzi2JaI/tBrCvS4Hnwu8P00Gxwu7OU5xXz/gy+VejeCvEu4tvs3uJtyWvQOtJ4CRj2VZWaW5D5KkTkWjaqG0b5ziXXfdu6nu3k7Bg5eDQIHd7/a7EjBmYFLVQHUlJ3z8LnGrvxcMosaxRcvl3yuCZ1PoL8BtN6A7f/ZbFWTXtKL3QXJFFWH2Z/byTuDPsFbiPZJd67nzMOv6WyNZxB4hYBrM7JDTlIUhaKTVbaAY1se5YU19m1anYYu/YJJGBRKRKJ/k7/jJoOFvWtOsGPFMQx6W4YkKMqL5ElxRPcMtK0XZjbAiW22zEbmWjiRCtbTuvWCu9n+zjoNBPX5Z5IqKlTs3aPlwH5XDAZbAOHqqtCtmwEvTQlFuQaKS7UUVwdhpfHzuLtUEOhVSkCghcBIbwLiIglI6obW/48xRF6CD2FXk3aYgrfeovLXX20PuIBfXBVBPSpw9bAFHBajiooT7pQf86AqXwtK7YeBCnQ9E/CZMAHvSTehCfBvkzYrFgs1+/dTtXkL+s2b0G/fgWJo/M7UNboznslD8Bw6BF1yMpqAdv4jNlTAoaWw7xvI+M3xg9An0h5sEHsF+EYCcCKthJX/2Ud1hQmtTsNVd3UnplfQBTXjVHoJP7y5C8WqMGJqV3peEVm/sboUDv4E+761fWjbMzAqW4Fqrxug+3WgCwCrFWpKz7zo2rMIp12EDeX201RaAvmu+GUqLKH4a45zfcDzeLiU0xaMVg9WlD3JcUM/VFi5LHYtveNzbBdbh4twg+8v5CJsNjR4H+r+rc+0ZBzz5pe0q7AorgS7HmWC30t4qkuaPJyiQI6pO7urriHTMBgFWwrfV32KXrqlJHn8hptL9fm3tyE3r7NkaxoGYyHg4X/WAubSfD1HtuVxODWfkpwq++MarZrY3kEkDgolqnuAU3Uchmozu1YdZ/eqbHuhcnhwFcmhy4ks/RbMp70Pfp3ra6NirwDvxtcVaw5FUcg5Usae37I5uqvAHjP7BLnTe2QU3YaFoz0ti2c1mig7cpii9EyKjhdTnG+mqMyTMmMgTc1u4aUpJtC7goBgFYFRfgTEd8a/azc0Ot15t709SPAhMGZlUTB/HuUrfrFlHVUKvtHVBPWswM3Lahsp0f1a24dm/oHar0OYKw2UZ9sCkerCBjUGKgXPWB2+yQl4pVyBOrovhCTZLlAXSFEUDOnp6DdvoWrzZvSpqVgrKhz2UQcH4Zk8BN3AgZhyc9Bv2kz1vn1gcRw1oe3aFc8hyeiGDEE3aBBqrzZIV5tqIH2lLeA4vALM9Xd5hPWy1Vh0uwYCuzjUBCiKws6Vx9n8QwaKYrurG3t/L3yDPVCsViylpViKijAXFWMptv1rLi7CUliEubgYS1ERismE52WXNVq3s2PlMTZ9l4GLRsUNTw5ofDRFZT7s/8HW9uwt9Y+7aEAXCPqiM+8kz8XFlSptF3449RdKDUH46iq4fsQuPIP8bBcxj4DWG4Gjca8NMIKxuPqw9qsj9tlg+4yKYtiUeIc7bUVRsJaXO7zH9ve6qBBL3XteUmoLxJykAFk+A0nzGwEqFcH6DPoULkFjrQGrxfalnP24eo0vx/0Gc8KnH2a1LUBSWwx0Kt9JdNlWdKYmghgXFzy7dyYgJR43L1N9UFSZf/7dTSoX2+9Fg8Ck0iWKIwVdSM8OIb/Qrf70ahXRPQNJGBRKTO8gXM9n9Iqi2Go1Mn+nOm0LO/YFsrd8FBZsn01RbrtIDl5KaFJtwBE3wlbbdIEsJivp2/LY/Vs2hdmV9scju/rTZ1QnonsFOZ2VNFdVUnLoEEVHsik6UUpxARRV+lBlbvymToUFP20RAaFaApK6EhgTRGCEFz7BHudX/9IGJPjowEzHMih47e+Urd5u7+r0jqomuGcF2u59bBfCHtc1nl63WqAkC/IPQv4BTId3Ur75IGUHKjGU1KcRVWoFr4gafDpX45Xoj0tEEoR0r/1KsqU53c4esRtPnKBq0yb0mzZTtWULlqIih+0uPj7oBg+yd6+4deliS7M2YKmsRJ+aag9aDGlpDttRq/Ho2RPdEFtmxKNfP1xaagVli8lWu7HvGzi4BIwNgqXAeNv73PMGCK4PCKx6vT1o0OcWsWG9kewC24d1lOYEPavWQlGBbZ/iYqcvdva6nQkTcIuKQlEUlr2/l8zdhXgHuHPTs4POXj9Sehz2fWd7Tbl7Hbe5+579zrj2go9nENVmHT+8uYviU1V4B7hz/cz+eAe4O/VaLoTVaLQHbeaiQvZsq2JXuu384Zo8+tb8DsX5tYFFMZhMrdMOlQvp8TdyMvIKACJP/k5i+teozrMGweLiRk5YMiciU9B71nbJKFYCi/YTdXI1/iVpNHVJ8hxxBQG334Hn8GH1f0eKYsvUOWS1Ts9sNchqVRfbj1dt9SajZhjp1ZdxytSdujt6FRY6ue0hwX0dce5b0Go5aw2Mw++PR4BtBFfJMVsXSl1XSmWew2upUndmm/IAB3KSsNZmaWP7BJE8KY7AyAu72agqq629WVtfe6N2daHr4FB6j4q64OM3pqaogOKDaRQfPUnRqSqKitQUVwVgsHo2ur/a1YWAcE8CIjwJjPAiINKTwAhPPP20Z3xGtjUJPjoasxHztu8pfP99SrfmoFhtv4BeETUEXxGE+6ibbcMtz7ew0GzAsH0N5T/9QPna7Rjz6y+0Lhor3p1q8ImuxjPUgMoFQAUBsfXBSEh3zOpwqg7nUZW6Df2mzZhOnnQ4hcrdHd2AAbXdJ0Nw756ESu3cnZK5qAj91q1UbdpM1ZbNmI4ddzyHmxse/fvjOSQZzyFDcO/ZE5XGicJHq9VWu7HvWzjwA0plERajC+YaFyzqEMxBg7F4dcNsdsdcVFR/11x7kVP0toLPSl04+3reh14XispqIjH9GyJy1jd64VD7+qIODEQTGGj7NyAAdWAAmsAg1IEBKDU1lC9fQdXatSgNLqLuvXvjO2E8biNG8/1HWZQXVNO5RyDXPNzMBeiKj9rqGOpqBZox0gbAoDfxv/m7KDhegaevG9fP7I9vcOumjs0FBZQvW0758uUY0tPPyJoB5AX352DSHVhdXPEpz6L33vdxMzX4Pfbysr2vAYGO/wYGoQkMQO3v79Tvo8mksOa3ak6csGXmBidr6d7DtUUuDoqicOqkhf37jZw8UZ/58/NzoXsPV7rEu6LRqLCUlVH69TdU/v67fR+3Ll0IuP02fCdNwuW0lL6iKJhqLOgrjFRXmKguN9Z+b/tZX1ZDdZme6nIDZcVmrNb61xLuX0hCwEG66LaiMx63dTk1zACeQ02phvJjOipzdbi4mNH6mtD61f4bqEaTmFzbXZliH2ZeXlhN6pJM0rbk2rpDVJAwMJTB18TiF+rc71z+sXJ2/5bNkW35WC31o456joikx2WRThV9twTFakV/8gRFOzZRtGsbxSVaisydKTFHYabxv0WtTlMfkER4EhjpSUCEV5uOmJLgoyOwWiBrHZYtX1D03SqKD2hQLLa7D12EiuBbr0Q36aEWH1KpKAqGgwcp+/lnyn/+GXNu/V2J2sMF72gTvhElaH1N6Au0VOW5UZWnxVh+2h+AWoVH1xg8hwzDM2UM7n374OLmRksynTpF1eYtVG22ZVjMBQUO2108PdENGoRuSDKeQ4eiTUhAqa7GXFRkCx6KizEXFmE5tg/zkR1YTmRgrjRgMdQGHAYXaPJes3F54ckcTLgFq4sb7koVQ3z2Exzmijog0HaRCwhEExSIOiAAjb8/KtfmfXBYysup+GUV5T//TNXmzfVZE5UK8+DRbPCciMWqInlSLAPHxzrV5uYy1pj58V+7yMssx8Pblev/0h//sMbv3i6UpayM8pUrKV+6FP2WrWdmiTSa2kCtPmAr1UWzsSARo0WNlyeMnhRAUEII6oAAXNxbLjNTVWpgyf/tpjC7ErWrC1ff3Z0u/VqncLskt4q9q09wcHOufZ4MraeGHpdF0HNEJ3S+bpQfOEre1z9RvHYrBsUVo6sPJq9ASOiBNTQag8kFfbktwLCYm59tC+7sbR+pckZmS1HAWOmYPWnY3VNVgPHEKcp351F+qBpD6dlrQNSBgWgTEmxfiQm4JyTgFp+A2suT4pwqtv6UScaOfABULiq6DQ1j0ITYs2bcLBYrR3cWsOe3E+QerS8YDovzpfeoTsT1C0Z9Mcy3oii24cJb3sd6cBnllhCKzZ0pcu1HkecwiquDKS0woFgbv5R7+roREOlFYIQtGAmM9MQ/3PP8usHOQYKPP6rayaTY+w2Wnd9TvFNPcZoXVpPtD8S9sz8hDz+AbuLtbTKcS7Faqd61i/IlP1O+fLmtq6DpvXH3N6ELNeIZakAXZMTFtfZXz80bQrrVZkl62LMleAW3XFsVBWNmpq2mZNNmqrZuxVp22ggFjQbMTtY3qFSo/fzOftccEICLXwBbN1Wxd50tWOvUzZ/R9/bAw6tlAy4Ac2Eh5ctXUL50qX1I9amwIRzqdjsoClcm15BwUwpqr5YLDExGC0ve3s2p9FK0nhque6I/QZ1aNkVtraqiYvUayn/+mcr16x26S9z79MZ3wgQ8hw5FExyMi69vo1mG0jw9S97ZTVlBNVqdhnEP9CKya8sVUhedrGTJO7upLDHg4e3K+Bm9CYv1ddin0ljJ9rztbMndQmpuKl6uXsxKnkWi//nPtWPQmzi4MYe9a07UjyxRcV6jTF21ajy8XfHwdsPD2w1d3fc+bui83fDwdsUnyAOfIOcLdE15eZQvXUb50qXU7K3v2lO5uuI5PBmfEYNReQZSk5mNIT0dw+F0TNnZTY6Oco2MrA1IEqkKSWTvST+yM211LC4aFT0uj2TA2Gg8feuzBTWVJvavP8m+30/a10JyUauIHxhCn1FRF/dMsyVZsPVD2PEpGGo/v7Q+WHrfQUnM7RRV+lF8qoqiU5UUn6yioriJ7JMKfIM8uOnZQS067F2Cjz8SRYG8fbD3G9j3Hdai45Qc8aTooBcWgy1y1cZ2IvgvT+N15ZXt1uenmM1Ubd5C+dKlVPzyC9aKCtzi4uqLPwcORKOqaFDcehDyDkDhYbA20efuGVwfiNi/utUOTbzA9los1Bw6hH7zZtuomm3bUKptVfMqDWi0ZtRaKxp3K2oPFZqIGDQJA1EnDkUTElp/R+3vf86um6oyAys+2EdOhu3DYsDYaAZPimuTojHTyZOUL1tG2c9L2WXtR074cFyNFQzeO4+gYf3wmTAerxEjLqgWxmKy8vN7e8g+UIybu5prn+jXYh/gVqORqnXrKP/5ZypWr7H/HwFoExPxmTABnwnjcevUqdnHrK40svTdveQeLcNFrWLU7d3oOuTCJz07fqCI5R/sw1RjwS9UxzWP9ME32AODxcDu/N1sztnMltwt7C/cj0VxLJT20Hjwt6F/45q4ay6oDVarQtaeQvaszuZkWilgywR4eNUFE654eLmiKctHObgTjuzHzVSBm7ECz+hQwm66joBrx7dcbRRgLimhYsUKyn9ein7btvpAQq3Gc8gQ22i6q65E3cTnvVWvx5BxFMPhw7UBie3f0zOZdcr848nsOoVi9ygANGqFHoP96TI8lkObcknbmofFZMvweHi70uOKSHpeEekQoFz0DJWw+wvY8j4UHal9UAVdx8OQByHmclCpMFabKc6pouhkpT0oKTpZRU2lCQ9vV+7+5+Ut2iwJPv4oTDXw5VTI+BXFAqWZOgr3+2CutmU13KI7E/TYY/iMG3dRTVxjNRqxVlWh8W/GHaXFBEUZkL+/PiDJP2CL8Ju6bVO1fLpQsdhG7KndrKg0Ciq1BrpcaZv/ouu48wp4qsoM7F97kr1rTlJTZZtU6qq7uhPbp+UyOk61J+0IP7yXRmmNBz5lR+m/az4uigUXT0+8r7oKn2sm4DlkSLO7esCWul7+731k7SlEo1Uz6bG+hHfxbXxfq4V/7/k3Xx76kuGRw7kt6TZ6BPU4Yz/FbKZqyxbKf64PZOu4du6Mz4Tx+I4fjzYhwfk3oZbZZOHXBQc5st2Wqh88MZaB42POO3g/sOEUaz5LQ7EqhCf40nmKCzvLtrEldwu78ndhOG1USZR3FMnhyQwKHcT/Mv7HxlO2+Xdu7XYrTw58EtcLmI+iTlWZARe1Cneda5N1PjWHD1Py6SLKfvzRPpxdHRCA/y0343fLLbiGnF93kaWykspff6Xs55+p2rjJIaPoMWCAbR6hsWPRBAae1/HBFtQY0tPtGZK6760VFShAiX9XjsZOotwn5ozn+ijFxHueIjrMiDYwoLars7brMzDQ1uXZwt3AYMvAKg0Kz83FxZgLC21dvEW1jxUV2UdcWSsrazOrtrove9dsYADqwCA0/v6oq4+gOfoD6ty19YPIQnpA8gPQ+6ZGh5Dry41UlRoI7nzhN3INSfDxR2C1wNfTUfb/SFm2N4UHAzGV2iaG0kSEE/zww/hee61zBZOXEmMVFByqHXlzEPJswYlSkYdK1Vq/sirbEOReUyBp0nkPI26seC0w0pOx9/dyuhCupZUVVLN4TirGajPxvnnEbf0Qc06Ofbva3x/vMaPxnTABjwEDzhrUWq0Kv3y0nyPb81G7unDNw73p1K3x96y0ppRn1j3DhlMbHB7vG9yXad2ncWWnUZj27KP856W2LrwGo580ISH4jBuHzzUTbEXCLZTdU6wKm/+XwY4VtsLkbkPDSJnWzak5KBSrwuYfM9ix3HaMss7H+bnzfyi3OHbpBXsEkxyezOCwwSSHJxPhVT/azGK18O7ud/lgzwcA9Anuw9wRcwn1PP/5KZxlLimh9JtvKPnsc8y5ubYHXV3xGTuWgNtvw6N373Mew1pTQ+Xva23dYr//7jA3j3v37raRWOPGterEgIqiYM7NtQciNWmHOX7cxGHXflTpwggq2E3UyTX4lmWcs1rLxde3vsDbXot1egBgy4AqJlNtkXlTw7XrAwulpvmFuM5ycdegcTWg1pptWVtPVzRxfVD3GIWmU5ytliwoCE1AAC4+Pi1+0yrBx6VOUVB+foqK7z+hYK8PxnJbgKEOCiLowQfxu+nGFi/ObC+KomCssZxRWV9dYax9rPb7Ctt2Q5UZ/1A3Evv6Et/HB7+gFnwf3HS2IaXnwWqxktFo8ZoPvUdFXTzFa0Dm7gKWvmfrbx99T3ciXU7WX/Qb1u1oNLY7K3vRZu2/QYGo/QPZfMSPo1ng4gLj7ksipl/jXRf7i/bz59V/5lTVKdzV7jzS7xEOFh9kReZyInPNXHbAyuUHXQgoq++KUPv54T1mDD4TxqMbOLBVM3v7153k9y8Oo1iVZi1KpygKJypOsOnEZjL/Z8DrmO1iui1yOduiloEKvN287YFGclgysb6x5wya1mSv4dl1z1JhqiDAPYA3RrzBoLBBLflSz0kxm6lYtYriTz51WILBo08f/O+4HZ/Rox0yY4rJRNWmTbZusVW/Yq2qn1jMLTbW1i02fjzauNYpcj6bouoituZuZUvOFlJPbMaaV8DlXn24TNeTJFUEqtIKe/bBUlRYP49OcckZ8we1NJVWaxvBVhsIONaK1Y9uc/H0xFJaZm/fmXPRFNlfg9P1aq6udNux3alM57lI8HEJUxSFqvf/Qv4n/8NQYruwuvj6EnjvPQRMm3bG8LiLkaIoVJXWBwy2QMIxiKgLMPQVRqzm8/8VDIn2JmFQKPEDQvHyb/s+2yaL1waE0HtUFKExF+fv8Kbvj7BjxXFctWpunDUQ/zDP+rqdn3+2dXdUVjb6XAVIS7yFUxGXo1Is9Nz/H4IL9+Di6ekQoGgCAkknn6WlGyj2sKANDuHhkc8R492ZipW/ULLkRyxZ9cOh9W6wo6sGy5XDGHXD4yQGJ7XRuwHH9hex4oN9TS5Kl6/PZ0vOFvvFrLi0jDFp9xBREY9FZWFTwnf49lJs2Y3wwXTz74b6PCZSyy7P5ok1T5BWkoZapebx/o9zZ48726WWq3rffko+/YSypcvsxb2a0FD8b70V9149qfjlFyqWr8BSWmp/jiYiHN/xtrlmtN26tWm7K4wVbMvdxtbcrWzO2cyR0iNN7qvT6BjVeRTjYscxNGIori4NAiqrFUtZWf1ot7NNPlfbNYJKhdrf/7Qh8YH2gnONPciwbVPpdC363jhMllcXSBUVYjm8GfPhrVgK8mpH6KkxG12xGhRcfHzounXLuQ/uBAk+LlFVW7ZSMOd5qtNsH8guWlcC7rmPgLumo/Zu2b651qBYFTJ2FrD1p6OU5OrP/YQGXN3VtZX0tUVxDSrrbRX3tm2u7mpOHComfVs+Jw4W1xfBqyAi3o+EQaF06R/cKqNIGio6Wcme1Sc4vCUX8yVYvGa1WPnxX7s4ebiUgAhPpjw9EFdt/cVSMZsbpJGL7POWmIqK2HUyiAxDNCgKvU59T0iW4xwjzlC5uaEbcQWHB4bykfdO9lUctm9LDktmWtI0ruh0xXldyJ1VeKKCJe/soarUgLu3htAbjOxmC1tyt5BZlmnfz6cmkAkHH8K3JhjcLCTd6sPlg/u3SJ0GQLW5mhc3vchPR38C4Oroq3lx+It4urbOsOVzMRcUUPLlV5R8+eUZkwGCbRisz9ix+EwYj0ffvm1Wf1ZjrmFXwS5bUJizlX1F+7CeNltsV/+uDA4fzJDwIQS4B7Dy2EqWZy4np6q+q9FP68fV0VczLnYcA0IH4KJyrv1WgwGVRuP0vERtKmcPbPk37P0aLAYUC1hcw9A89ottOvoWIsHHJca26Nt8W2EWthlE/a9IIHDOJ80r2mxniqJwbF8RW348ap+K2MVFZQ8iPBoJIjy8XdHVbfNyRXMeY8715UYyduSTnppnH0lSd+5OSQEkDAohrk9wiw0lU6wKWfuK2PNbNicO1U9pHRTlRZ9RUcQPDEHj2nYfQIqiUGmqxNvt/ALTqjIDi+ekoi8zkjg4lKvu6n7Ou7HN/8tg+7JjAIy6oxtJwyJsd12VlfZAJf9kOl9v/g9VBTn46WGAazxRZh8stf3eVr0e3dAh+E6YgNeVV9qnwFcUhR35O/js4Gf8evxX+4Wkk1cnpiZN5fr46/Fya53p8vUmPTvzd7LlyA5MS0LxqgjC5GJgVcInHAvYhwoVSYFJDFGNxOPXRCx68ArQcs0jfVplxVlFUVictphXU1/FbDUT6xvL/JT5xPm17grEZ2M1GqlYtoziRZ9hysnB64or8JkwHs/k5DapPTNZTewv3G/PQO3K34XR6rhYY2fvzvbs0+CwwQS4n1mDZFWs7C7YzbLMZazIWkFxTX1XY4hHCGNixzA+djw9Anu064yhZquZKlMVvtrz6wpuVFUhbPsYUv9jW6dnxiacW6nw7CT4uETUpKVR8NbbDRZ9U/CP0xN4/WW43rXI1pl+kTuZVsKWH4/aL/6u7mr6XhlFn6s6n7HgUmuqKK4hfVse6al5DmsxqF1diKldXyK6Z+B5BTnGajMHN+awZ80Jygtqh+OqIK5vML1HRREe3/icEq0prTiNl7e8zM78nVzV+Soe7vsw8f7xTh/nVHopP7y507YA3a2J9BzR9JDVbUuz2PLjUQCuuCWRXiln7rv+5HqeWfcMZYYy/LR+vHbFawyLGOawj6Io53y/TlWe4stDX/JN+jdU1E5dr9PouC7+OqYmTSXaJ9rZl+rAZDGxp3APW3K2sCVnC3sK92CuXcPG1azl6vTpdC7tjoJC6NUw+pqBFOwzsOrjA1jMVoI7ezPh4d6tnuHaXbCbv6z5C3n6PDw0Hvxj+D8YGzO2Vc95sbAqVtJL0tmcs5mtuVvZnredKlOVwz4hHiH2YCM5LJlwL+eGTJutZrbmbmV55nJWHVtFRYNZb6O8oxgbM5bxsePP62+ruRRFIacqhyOlR0gvSSe9NJ0jJUc4WnYUk9VEUkAS42PHMzZ2LGGeYec+YHOYjVB+4vxnvW6CBB8XOUNmJoVvv0P5smW2Me8uLvjGWwhKzMetxxC47Ttwbbu1MM5HXmY5m/+XYc8AqF1d6J3SiX5jOrd6l8e5lORWkb7NlhEpzavv/nF1VxPXx7aUd6ck/3MWgJbm6dmz5gSHNubYV9PU6jR0vyyCXimd2nS9kjqVxkr+b9f/8cWhLxzmilChYkLcBGb0mUGUT5RTx9y58jgbvzuCi0bF5JkDGq1T2bXqOBu+sfWhD7shnn5XO6ZqrYqVD/Z8wLu73kVBoWdgT+alzHP6YnA6vUnPkqNL+OzgZxwtswU+KlRc3ulypiVNY2j40GYFfharhUMlh+wp+h35O6g+bTXUcM9w+4iUQSGDOfhjCQfW2Rali0ryJ/ug7Xc9pncQo+/p4dBN1ZqKqot4eu3TbMm19c/f3v12nhjwhEOdwh+BoigcrzhuDwhTc1MpMTgumufj5lNfyBueTIzP+Q+PPp3RYmT9yfUsy1zGmuw11FjqR6Uk+CcwLmYcY2PHEuXt3N9XQ8U1xRwpOUJ6aTrpJekcKT3CkdIjZwRVTekf0p9xseMYHTO60axOe5Pg4yJlOnWKgnffpez7H+zV1N5XjyI4ZAtaSzoEJ8Hdy2zpsItU4YlKtv50lMzdhYCtuLLHZREMGBeDp9/FVeegKAqF2ZWkp+aRvi3PXhAK4O7pSpf+tkAkIt7PPg+CoiicOFjC7tXZHNtXZJ9qxD/ck94jO9E1OazNLjqnv5blWcv5Z+o/Kai2Ta50dfTV3NL1Fr449AWrjq8CQKPScF3CdTzQ+4Fm3yUpisLyf+/j6K4CvAK03PzsYIe1LPb9foLfv7DVYjQ2PXuZoYxn1z/L2hNrAbgx8UaeGfwMbuqWC0IVRWFTziY+O/iZ/TwAXXy7MDVpKhO7TMRD4+Gwf2ZZpv2uOTU3lXJjucMx/bX+tjvm8GSGhA2hk3cnhwtZ3arDm77PsD/WK6UTl92U0OaripqtZt7Z+Q4f7fsIsF2E5qbMJcgjqE3b0dLyqvLsBaJbc7eSW5XrsN1D40H/0P4MCRtiK+QN6OZ0Tcb50Jv0rMlew7LMZaw/td6eFQPoHdSbcbHjGBMzhmBd43P26E16e2DRMJtRVHNmzQyAxkVDrG8s8X7xJPonEu8XT7xfPB4aD1YdW8XSzKXsyK8ffaRWqRkSPoSxsWO5svOV59312tIk+LjImAsKKPz3B5R+9ZW9MM9rxAiCH34Q9+3PwrEN4B0O964CX8dUtt6kJ7siG393f/zd/dvtbqc0T8/WJZmkb8sDxdbt0HWIbf2E85lmua0pVoXco2Wkp+ZxZEe+fcVKAE8/LfEDQ/AJdGff2lOU5NTfhUT3CqTPyCg6Jfm3W//v0bKjzNk8x37n29m7M88mP8vwyOH2ffYX7eftnW+z4aRtHg03Fzdu7nYz9/S8h0CPc0/kZKg28/WcVMpOW4Du0KYcfl14EID+Y6MZcm2cw/twqPgQT6x+ghOVJ9Cqtfx1yF+5Lv66Fnz1ZzpWfozPD37OD0d+QG+2ZbZ83Hy4IfEGYn1i2ZJry27UBWl1PF09GRg60J7dSPBPaNaFLH1bHtuXZdkzXu1ZB/DrsV95bsNzVJmqCPIIYu6IufQP7d9u7XFWmaHMPmJoS84WssqzHLZrXDT0Ce5jH6LcK6hXixXynq8yQxm/Hv+VpZlLSc1NtdciqVAxKGwQ42LHodPo7AFGemk6JytPNnosFSo6eXeyBxd1gUa0b/Q5P9tzq3JZkbWCpZlLOVB0wP64m4sbl3e6nHGx47ii0xUOQXhbk+DjImEpLaXoo48o/nSRfWIZXXIywY//CV2fPvDNXXDgB9D6wF3LIKynw/O35GzhqbVPORRE+Wp9CXAPcPgK9Agk0D3Q8XGPALxdvS/4g7KiuIbUnzM5tCnXvnBR/IAQBk+MbbVFw1qb1WLlZFoph7flcXRnAcZqx/Hxrlo1ScPC6ZXSqV0nBdOb9Hy490MW7F+A2WpGq9Zyb697uavnXWjVjWeZtudt5+2db7M9bztgu3O8Lek2pvecjo/b2f+eCk9U8M1r27GYrAyeaFsZ9JeP9qMo0HtUJy67McHh9+l/R/7Hi5tfxGAxEOkVyZspb5IU2HbDYyuMFfxw5Ac+P/g5JypPnLHdzcWNfiH97DUBPQJ7oHG59CflyyrL4ok1T3Ck9AgalYa/DPwL05Kmtfty6o3Rm/Rsz9tuDzgOFR9CaTBzsQoV3QO724ONfqH92vXieS6F1YWsyFrBssxl7C7YfdZ9gz2CbUGGfzwJfgkk+CcQ5xuHzvXCP1OyyrJYnrWcZZnL7N2RYKuLGtl5JONjxzM0fGibB24SfLQzS2UlxQsXUvzxAvtcCe59ehPy+ON4Dh1q22n5LNj8Lri4wu3f2ZaLrqUoCv/d91/e2vkWVsWKTqPDYDGcsR7Eubi6uDoEI3UBSqB7IAEepwUw7oEOv6hVZQa2Lz/G/nUn7fNwxPQKZPCkOIKjLo4UX0uwmKwc219k65YpriF+QChJw8JbdLElZymKwm/Zv/Ha1tfsQwKv6HQFzwx+pln9zYqisOnUJt7a+Rb7i/YDtkmv7upxF9OSpp31w+/gxhx+++QgqMBFpcJqVeh+eQQpU7vaL25Gi5HXtr7G4sOLAbg88nJeufyVlq3Kd4LFamHdyXUsTltMpanSnt3oE9wHd83FXTt1vvQmPbM3zWZZ5jIAxsaM5YVhL7TIhe1CGC1Gdhfstgcbewv2YlYcg/suvl3s3V0DQwe22+/NhTpZeZJlmctYfXw1LioXEvxtAUZdVsPfvfW7zxVF4XDJYZZmLmV55nJOVZ2yb/PV+nJV56sYHzueAaED2mS4ugQf7cRaU0PJZ59T9OGH9ol3tF27EvynP+E1MqX+zmTjO7DyOdv3N3xkm867VoWxguc3PM+vx20jYK7tci1/HfJX3NRulBnKKK4pprimmKKaIoqra/+tKaa4uti+rbimmEpT4xNEnY23mzdhLhEkZV9OWGYSLhbbBVgTZSQsxYXILvVBjI+bz0V5p3Wpy67I5tWtr9rrGsI9w3lm8DOMjBrp9PutKAq/Hf+Nd3a9Y59wKcA9gHt73ctNXW9qMnuy+tODHNhgC3q6Jodx5Z1J9pqYnMoc/vL7X9hbuBcVKh7q+xAP9H6gTfrhhSNFUfj80Oe8kfoGZsVMF98uvDnyTWJ92242UYvVwqHiQ7ZF83K2sDN/p0OhJkCEZ4S9QHRw2OAm6yTEhVEUxWEIccP6kmCPYMbEjGFc7Dh6BfVqtc9uCT7amKIolH75JYXvvmdfadEtJobgxx7Fe+xYx0l39n4D395j+/7qF2H4Y/ZN6SXpPLHmCY6VH8PVxZVZybOYkjDlvH5Rasw1lNSU1AcqNcUUVRc5BChF1UX2fVQmNb1zUuidMxKtxZb2zPXKZGvnnznlm37G8TUqjT2j0jB7UvdziC6EeL94gj2CJUhpBoPFwH/3/ZeP9n6EwWJA46Jheo/p3Nfrvgu+m7VYLSzLWsa7u94luyIbgFBdKA/0eYDr4q87o6/ZbLKw9ovDuHu6MuS6OFxqRwVtOrWJp9c+TYmhBB83H1674jUui7zsgtomLtzO/J38Zc1fKKguwNPVk5eGv8RV0Ve16DksVgtlxjL7TU56aTpbc7aSmpdqHwpdJ8A9gOSwZHt314WMDhHnx2K1kJqXyrLMZfxy7BeH/6NOXp0YF2sbuZPon9ii55Xgo42VfvMNOX99HgDXiAiCHn4Y32snnTnxTuY6WDQZLEZIfhDGvmqf4GXp0aXM3jSbanM1YZ5hvJnyJj2Dep5+qhZnNlrYsyabHSuOYaiydeu4BStoh5ZREZ7jEKzUBTKnf9icja/W94ziqnj/+HPWH3Qk60+u55Utr3C8wjazbXJYMs8OeZY435Ydg2+ymvjfkf/x/u73ydPnAba5DB7q8xDjY8c3mZa1Klb+u++/vL3zbayKlaSAJOalzKOTd/OXsRetq7C6kJm/z7TX+tzV8y4e6/fYWWtc9Cb9GX/fdTcl9oxqbVa1xFByxuyhdbxcvRgYNpAh4UMYHDaYeL94ueG4iJgsJjac2sDSzKWsyV7jMMR81ZRVLbqAoQQfbSz74Ueo/PVX/G6+mdDnnm180be8/fDfcWAos62YeuMCcFFjspiYu30unx38DICh4UN57YrX7P2Fpfl68jLLzzxeC9CXGdn163H0ZbZZAv1CdQyeGEt8/5Aml+AGW79uYx9YDQOUkxUnOV5xvMkPrFBdqEMhVrxfPHG+cW3aR19jrnF4DUaLkVjfWDr7dG6TUUW5Vbm8nvo6vxz7BbClRp8c9CRjY8a26oe3wWLg67Sv+XDvh/Zi5ni/eB7u+zBXdr7S4dzlxnKeW/8ca7LXAHB9/PU8N+S5JrtsRPsxWU3M3z6fTw58AsDgsMGMjBrp0E3bMBN6+jwnzeGn9SPAPYBwz3AGhg0kOSyZpMCkP0Qhb0egN+lZe2ItSzOXojfp+c+Y/7To8SX4aEOKopA+/DIsxcVEf/E5un79ztyp7CT85yqoOAWdh8LtP4CrO/n6fGb+PpOd+TsBuK/XfTzc92HULmqsVoWdK4+x9adM+7LsrcU7wJ1B18TQNTnMnmJvCQaLgcyyTIdx7uml6WeM5a/jonKhs3dn4v3i6wu3/OPp7N25WR9uVsVKqaHUof6lqS6n4priJif2cXVxtY+5T/BPIMEvgXj/eCI8I1okKDBZTSw6sIj3dr9HtbkatUrN1KSpzOgzo9WmD2+M3qTns4Of8fH+j+3ZrB6BPXi036MMixhGemk6T6x+guMVx3FzcePZ5Ge5IfGGNmufOD8rslbw/IbnmxVcuLm41Y+WO70I3aNBgbp7AH7ufn+4ic06MqtibfFaLQk+2pAhM5Oj48ajcnMjcVvqmVmP6lL4eBzkH4CgrnD3ctAFkJqbypO/P0lRTRFerl7MuWwOIzuPBKCsoJpfFxywT1ke3Nkbra7l7yxcXFTE9A6i+/AI1K5tVzBYYawgozSDwyWHHSbiKTWUNrq/m4sbcX5xxPvF08WvC4DTqeGmaFw09g9XF5ULR8uONvmh7enqSRe/Lg7ZmgT/BKdmGkzNTeXlzS+TUWabuKpfSD+eS36OrgFdnWp3Syo3lrNg3wIWHVxkf+29g3pzuOQwNZYawj3DeTPlTXoE9Wi3NgrnHC09yr/3/BuLYjmjHivQvT7Y0GladnVV0bFJ8NGGSr/9jpznnsNjwABiPlvkuNFsgE8nw7H14BUG965C8e3EJwc+4c3tb2JRLCT4JzA/ZT6dfTqjKAoHN+Sw/ut0TAYLru5qLr8pkW5Dw/7wHxCKolBUU2QLSErqA5KMsgyn08ONzoXiHmi/kzvbXChWxcqpylP2oKguQMosy3SY5bChAPcAh4Ak3t9W49JwFdLC6kLmbpvLkqNL7M95YsATTOoy6aIZKVJUXcRH+z7iq0Nf2RfsGhYxjFcvf7VNhg0KIS5tEny0oVN//Stl33xL4H33EvKXv9RvsFpto1r2fwdu3nD3MqoC43h+w/P2Pv5r4q7hb0P/hofGA325kdWLDpG1xzZteUSCH1femXRJzB7amqyKlZMVJ23dNqVHyCjNsM1f0sS8Jf5a/1aZWMdkNXGs7JhDQHKk9AgnKk44TJrUUKRXJPF+8YR7hrPk6BIqTZWoUHFT15t4tN+jF+38BrlVuSw6sIggjyBu7357m8wPcDFSFIXc8hrScis4Xqwn3NeDrqHedPL3aPPp1YW4FEjw0YYyxk/AePQond59F+9RI+s3rHgONr0DLhqY9g1HA6N5fM3jZJZlonHR8NSgp7il6y2oVCqO7ipgzWeHqK4w4aJRMWRSF/pcFSUfcJcAvUnP0bKjDnUtR0qPnDG1N9jqKf465K9tMopJOKdUbyQtt4LDeRUcqv03LbeC8pozs106NzUJod50C/UmMcybrqHedA3zJsjL7Q+foRTibJy5fkuJ8gUwl5RgPGqb2tajX9/6DZvetQUeANe+y0q1ied/vhW9WU+ILoS5I+bSN6Qvxmoz675O59BG24ROgZFeXHVXd4I6tV3RoYCSKiObjhaRVVSFv86NQE83gry1BHlqCfRyQ+embvKionPV0TOo5xkBRUlNiT07klmWSVJAEpO6TOqwWQSwZRJ2HC8lNasYT62GoNr3OdDTjUAvLT7umla/eFcbLaTnV5wRaOSVGxrdX+2iIjbIk5hAHSdLa8jIr0RvtLA7u5Td2aUO+wZ4utkDkUT7v154u0uRphCncyr4iImJ4dixY2c8PmPGDP7v//6P6dOns3DhQodtycnJbN68+cJaeZGq3mkbpeIWF4fGv7ZPfP/3sOJZAMxX/o35xmwW/v4SYBv69voVrxPoEcip9BJWLThIRVENqKD/6M4MviauTQs/O6oqg5mtmcVszChkw5EiDuaWc7b8n7urC4Ge2tqAxI1AL9vFMtDTjSAvLUFe2trH3AjQuaFRu+Dv7s+gsEEMChvUdi/sIqQoCgdzKvhpzyl+2n2KEyVN1++4qlW177MbgbWBX5BXfXASVPezlxsBnm5oNU0HcmaLlayiKltwkVsfZBwr1jf5fx3p50G3MMdsRlywp8N5TBYrx4qqSMutJC23nLS8Cg7nVZJVVEVxbRC76WjRGcetC0i61f7bJcTzrO0X4o/OqeAjNTUVi6V+fZF9+/Zx9dVXc+ONN9ofGzt2LB9//LH9Z7fG5rz4g6jeYVvi2KN/7fDarA3w3f2AQuGAO3iyah/bjm4D6if9UVlc2PjtEXauOg4KeAe6c9X07kQk+LXPi2gFqVnFbM0sJibQk65hXsQEeqJpwSG8zjKYLew8XsrGI4VszChiV3YpZqvjFSgx1IseEb5U1JgoqDRSVGmgsNJAjclKjcnKydJqTpY2r/DVX+dqv1gGetUFLPUX0yCv+ourl7b17/bbQ2ZhFT/tPsWPu09xJL9+qn+dm5orEoJRUCisfZ+LKo1UGMyYLLYai9zymrMcuZ63u+aM97PKYOZQbgVHC6owWhof+dQwQ1H3lRDSvAyFq9qF+BBv4kO8mdA73P54tdHCkfxKDuWW27ps8mzBSV65wf6789uhfPv+dRmVrqHexAZ52gIrb60t+Kr9vfHzcJWuV/GH5VTwERzsOCf/q6++SpcuXRgxYoT9Ma1WS1hYWMu07iKn32HLfOj6D4D8g/DlrWAxsjNxJH+p3k9BsW264xeHv8jV0VdTeKKSVR/vp+ikbX6JpGHhXHZjQrsuYtaSrFaFf/2azlu/pTvcXbqpXegS4mW/6+sa5kXXMB8ifN1b5cJrsSrsO1nGhoxCNmUUkZpVTI3J8ULUyd+D4V2CGBYfyNAugYR4Nz65md5opqjSSEHtRbKo0kBRlZHC2p/r/i2qMlBcZcSqQIneRInexJH8Rg/pwE3jckb3Q6CXm73LJ6hB0BLg6YZrOwZx55JTVs2S3Tn8uPsUe0+W2R9307gwsmswk/pEMqpbCB5uZ97x15gsFFcZ7e9pYe37XBecFDR4n4sqjZitChU1ZipqzGQWNj5fi85N7ZBtqAs0grxafoI0Dzc1vTr50quTYxFxqd7I4bwGWZJcW4BSXmPmSH6lQ2B2OrWLCn+dm0PGpz4jVPdYfQausfdViIvVeV/1jEYjixYt4s9//rPDBWTNmjWEhITg5+fHiBEjePnllwkJCWnyOAaDAYOhvr+1vLx1ZvNsaVajkZp9+wDQdYuGRVNQasr4vHNP3jBnYTZZ7As9RXvHsGPFMbb8dBSrWcHD25WUad2I6/vHWWCpVG/k8a92sSbNVmh5RWIw5dUmDudVoDdaOJhTzsEcx/9bL62GxFBbINI11IvEMG+6hfkQ4OlctkxRFNLzK9lQm9nYfLSIitMKBYO8tAzrEsiwLoEMjw8iKqB566Xo3DToAjTN2t9iVSjVG23BSYWBwqr6DEpxlZHChsFKpYEqowWj2cqpshpOlTXvbt/Xw7U+o+LlRrCXlvjQ2m6CUG98dW1bX1BUaWDpvlx+2nWKrVnF9sfVLiqGxwcxqU8Eo3uE4nOOrIK7q5oIPw8i/M49uktRFMqrzRRWGSisqA9SCiqNaDUu9qxGpF/7j0rx07kxODaAwbH1c8EoikJeuYG0vArScss5UVLdIMCyvZ5SvQmLVbEHYnDuJQ10buoGXVW234/YIM/avytvwnxaJ9i/UFUGM4fzbN1iRwur6B7uw7ie4bhpLt5AW1y48x7tsnjxYqZOncrx48eJiIgA4KuvvsLLy4vo6GgyMzN5/vnnMZvNbN++Ha228buN2bNn88ILL5zx+MU+2kW/YyfHpk5FHRBAwms3UP3bP5gdGc0yV9sddt0S1+YyFasWHCDniO1OMKZ3ECNv64bO54/THbXvZBkPLtrOiZJqtBoX5lzfixsG2Nb9sFoVTpZWk5ZbUftha/uQySioxNTEzK1BXlpbdiTUh65hXiSG2u5cPbX1sXJ2sd5es7Exo6j2A7qet7uGIXH1wUZCiNdF98FbbbRQVGVw6H4orKoPTuzBSpWR4iojFuu5/1TDfNztF5vE2oAkIdQLd9eWuyuuqDGxYn8eP+0+xfojhQ7tGhwTwMS+EYzrGdYqGYaOwmSxUlJ1ZsansEGXYFFtpqig0oDRfO7J9bzdNQ7dTXVZIT9d23wWGc1WjhZW2j4L6kYU5VWQXXxmd2aIt5bbhkQzNbmz/B5dQtpkqO2YMWNwc3Pjp59+anKfnJwcoqOj+fLLL5k8eXKj+zSW+YiKirrog4+ijz4i/59v4HXVlbiPVXN/znKOuLmhUWn4y8C/MLXbVNI257Luq9oJw7RqLrspgaRh4RfdRfBCLN6WzV9/2IfRbKVzgI73bxtA94hz/78ZzbaCwLoPorrA5HixvsnnRAV4EBfkxdHCyjM+sLQaFwbFBDAsPpDhXYLoEeHTrnUmLc1qVSirNtXeCddfjE6VVZOeZ/tAb6omxUUF0YGeDbJM3k7X4tSYLPx2KJ8fd53it7R8h4tdr0hfJvYJ55reEc3KXIiWpSgKVUaLPSixBShG8itqOJJv+904WljVZPAa4q21BSS1Q4e7hXkTH+KFzu38EuNWq0J2id7hb/twnq0O5/Raq9Pb0Mnfg18P5pNfYbsmuKldmNQ3gunDYugZeXHOiyPqtXrwcezYMeLi4vjuu++49tprz7pvQkIC9957L08//XSzjn2pzPNRt5hcyJNP8h+PL1hkyiVI7cHcq9+nm0dP1nx2iMzdtgnDwuN9ufLO7vgG/3E+mGtMFl74aT9fbLUt0X5ltxDm3dT3gtP+eqPZfjGt+9A6lFtBQYVjZkPjoqJPlB/DuwQytEsQ/aP9OvzogfIak/29s71v5aTlVlCiNzW6/7lqcUwWK+vTC/lx9ylW7s+lylhfbN4l2JNJfSKZ2CecuGAZGn6xM5gtZBZWOWQdDuVWNDn6SKWCzgE6h6HD3cK8iQnytNcdKYpCQYXBfuNQd9zDeZVUmyyNHrcu+9IwO5cY6u3Q1Wo0W1m2L4f/bshyGM48OCaAu4bHcHX30D/UjcUfSasHH7Nnz+bf//432dnZaE5fNr6BoqIiIiMj+eCDD7jjjjuadexLIfhwWEzu88+ZvPduTmJmfsxk4vweYfWi2gnD1CqSJ8XR9+rO7d733JJOlOiZ8dkO9pwoQ6WCP1+VyMMj41v1NRZXGe3dNRG+HgyKDcBL+8co1G1NimIbVWIP5nIrOJRXQXptLU5jvLQa4kO8yCqqorRB4BLp58HEPhFM6hNBUrj3HyqD11FVGsyk5zlmKNJyKyisNDa6v6taRZdgL3zcXUnPP0tgq3EhIcSrPnipzayEO1lkvuN4CR9vyGLZ3hx71iTSz4M7hkZzy6DObV7jJM6uVYMPq9VKbGwst956K6+++qr98crKSmbPns0NN9xAeHg4WVlZPPvssxw/fpyDBw/i7e3d4o1vLw0Xk2PlZ9y44la8TK78o/oVDu+3/TEERHhy9d3dCerUvNd9qVh7uIA/fbmTEr0JP50r/7qlHyMS/ziFsx1FXS1Ow9k803JtwV3D1HiQl5ZreoczsU8E/Tv7ScDRQRRWGuy/F/bJ2HIrHLJfYOvSi6kdMtyw6yY6QNei2YncshoWbT7G51uPU1xlC4w8XNVM7h/JXcNjiA/5Y33OXqpaNfhYuXIlY8aMIS0tjcTERPvj1dXVXHfddezcuZPS0lLCw8MZOXIkL774IlFRUa3S+PbScDG5ZU8N59PNX3Lj/hm4GYNBBX2v6kzypFg0LVjk196sVoV31xxh7i+HURTo3cmXd6f1p5N/80aNiEtDXS3O4bwKAjzdGBwTICluAdiyaCdLqzmcV0F5tZn4EC/iQ1q2mPlcakwWftx1iv9uyORQbv0IoMsTgrhreAwpiSF/qCzzpUbWdmllDReTeyRxC0HbetEjbzhefmquurs3kYl/rBVAy6pN/GXxLlYdtE1ccevgKP4+sUebfugIIUQdRVHYfLSYjzdk8svBPPu8QrFBntw5NJopA6OkW7YdyNourax6u21m0+qkaA7lL+T24qkAjLw59g8XeBw4Vc5Dn23nWJEeN40LL13bk5sGNT+TJYQQLU2lUjG0i22CwOxiPQs3ZvHVtmwyC6uY/dMB5q48zI0Do5g+LIbOgZKdrZuBt25umbS8SgwmC189MLTd2iTBh5PMJSUYMzMB2BRYQmhGLDqTD1pVJZE9I9u5dS3rux0nePb7vdSYrHTy9+D92wbIcDchxEUlKkDHX6/pzhNXJ/LtjhMs2JDF0cIq/rshk483ZnJlt1DuHh7D0C6Bf/iapbo1jRpbe+j0Pg4Xla0bq70y2BJ8OMm+mFyXLvxSupnY4t4AxHjuRe06qT2b1mKMZisvLjnAp5ttiwiOSAxm/s198Xdy5lEhhGgrnloNdwyN4bbkaH5PL+DjDVmsPVzAqoN5rDqYR3Sgjs4BOoLtC0HWT03fcAmDS6E7uWH9TX2gUUlGfmWTaxr561zpWjuLdN3Qek071sdI8OGkusXkNH16sD13BbcUjwUgLuBIezarxeSUVTPjsx3sPF6KSgWPjUrgsSsTUEsRlxDiEuDiomJk1xBGdg3hSH4lCzZm8u32kxwr0nOsqOlJDOt4azWOwYm34+KQdVPXB3lp8W2Dxf+Kq4y2BQsdZomupNJgbnR/nZuahFBvujaYVDAxzItgL+1FlfmR4MNJdYvJZXZ2x78yHG9DIBpVDVHBhe3csgu38Ughj36xk6IqIz7uGv51Sz9Gdmt6XR4hhLiYxYd48dJ1vXhyTDd2Hi+pX8qgicUhTRaFCoOZCoOZrGYEKmoXVW22pHVGhOkNFoqqGp9zReNim3PFNodKfaDRyb/91zRqDgk+nNBwMbnV/jnEHrZ1uUS77cDV++IcmdMciqLw77VHeX35IawKdA/34f3bBkihlhDiD8HXw5WUrme/kVIUhfIas8Pijw0Xhyyqnba+bkHD8hozFqtyxuzLrSEqwMO+1lVdkBEb5HlJL74nwYcTavbtRzEacQnwZ5lpJxOLnwAgzn0z6OLauXXnp6LGxMyvd7Nifx4AUwZ04qXrel4S/Z5CCNFSVCoVvh6u+Hq40qUZ8yYazdba1aoNTdZZXCg3tQuxQZ4Oi2r+UfzxXlErqt5pq/fQJ3VGW1FIQHUYLi5WYrTbQDeonVvnvMN5FTz46XaOFlbhpnZh9qQe3Do46qLqFxRCiIuRm8aFMF93wnzd27splyQJPpxQV+9xMEKxj3KJCsjBzaUadAHt2TSn/bj7FE9/s4dqk4UIX3feu20AfaL82rtZQgghOgAJPppJURT7MNuVvtnE5YwHIM7vEFQCusB2bF3zmSxW5iw9yMcbsgC4LD6It27t57CqpBBCCNGaJPhoJmNWFpbiYhQ3VzJ0GgZVRaFSQaz7jksm+Mgvr2HGZzvYdqwEgEdGxvPE1YkyjFYIIUSbkuCjmerm9yiJCaBzWS8AIhL88DCdsO3gcXF3u2w5WsTDn++ksNKAt1bDvJv7cnX30PZulhBCiA5Igo9m0tcGHzvDa+z1HnH9gmFTsW2Hi7TmQ1EUPlqfySvLDmGxKnQL8+a92wYQG+TZ3k0TQgjRQUnw0UzVtcWmuwKhb3EsALE9fOD3StsOF2G3S6XBzNPf7OHnvTkAXNc3gjmTe6Fzk/92IYQQ7UeuQs3QcDE5g643qmIXQmJ88Havsu2gUoP7xbXg2pH8Sh5ctJ0j+ZW4qlU8f013bh8SLcNohRBCtDsJPpqhbpRLQYiWyIo+AHTpFwz6Bl0uF9FFfdneHGZ+vZsqo4VQHy3vThvAgGj/9m6WEEIIAUjw0Sx1xaZ7ItRElCcCENc3GKqybDtcJF0uZouV11ek8cHaowAMiQvg7Vv7E+ytbeeWCSGEEPUk+GiGusnFToX0ItKoJiDCE79QHewrsu1wEQQfBRUGHvl8B1sybdmYB66I48kxXdGoL925/4UQQvwxSfBxDg0Xk1O59QFj7SgXAH1t8OHRvl0a248VM+OzHeSVG/DSavjnlN6M6xXerm0SQgghmiLBxznULSZX7OVGSFUSUFvvAVBtm6xL0QVyJK+CcD8PvNpwASBFUfhk0zFeXHIAs1UhIcSL928fQJdgrzZrgxBCCOEsCT7OoW4xuUNxPdAobvgEuRMYWXtxr8187CvVMPHNtWhcVPSJ8mNYl0CGdQmiX2e/VlsdVm80M+u7vfxv1ykArukdzms39P5Drn4ohBDij0WuVOdQV+9RHNAHHRDXL6R+uGpt8LHmuAUAs1Vh+7ESth8r4e3fjqDVuDAoJoChXQIZHh9EzwifFqnByCys4sFPt5OWV4HaRcWz45O4e3iMDKMVQghxSZDg4ywURUG/YwdWlQY3VU9QGnS5gH2obZbenQBPNxY/MJQdx0rYkFHIxowiCioMrD9SyPojhfxzRRre7hqSYwMZHm/LjCSGejkdMKzcn8tfFu+mwmAm2FvL/03tz+DYi3N2VSGEEKIxEnychTEzC2tJCfnB3dEoHnj6uhEa41O/Q23moxhvbhsSTXyIF/EhXtw0KApFUTiSX8nGjCI2HClk89EiymvMrDqYx6qDeQAEebkxtEsQw2u7aToH6ppsi8WqMHdlGu+uyQBgUIw//ze1PyE+7q33BgghhBCtQIKPs6ir98iM6gvY5vZQNVgB1lBRgBaodPHh9iHRDs9VqVQkhHqTEOrNncNisFgV9p8qY8ORIjZmFJKaVUxhpZGfdp/ip922uo1O/h4Mq+2iGRoXaA8siioN/OnLXaw/UgjAPZfF8sy4brjKMFohhBCXIAk+zqJq+3asKheqvHvjQoMhtrWUKlu3y4BuXc45kZfaRUXvTn707uTHQyldMJgt7DpeyoaMIjZlFLLzeCknSqpZvO0Ei7fZVspNCPFiSFwgvx7M41RZDTo3Na/d0JuJfSJa5fUKIYQQbUGCj7Mo3baFMt8uuKi80XpqiEjws2/Lzi8mSqkGYPLlvZ0+tlajJjkukOS4QLg6kSqDmdSsYjZm2DIj+0+Vk55fSXq+beG6uGBP3r9tAImh3i3y2oQQQoj2IsFHE8wlJbgcP0VB/I0AxPYJxqVBN8e36/fwOGDBhcTOnS74fJ5aDSldQ0jpGgJASZWRLZlFbMwowttdw4MjuuDt7nrB5xFCCCHamwQfTdDv2IGCitzQ2oXk+tZ3uVTUmFi7+zCPq8Ci9UfdCkNc/T3dGNsznLE9ZaZSIYQQfyxSsdiEnE2rqfDujNnVH43WhU5J9VOof5WajdZUCoCrT3ATRxBCCCFEYyT4aEJJ6ibyg/sCENMrCE3tTKVmi5WPN2QRQAUAqotgUTkhhBDiUiLBRyOsRiO6jBwKgvoCtiG2dVYeyONkaTWRWr3tgXZeVE4IIYS41Ejw0YhT29Zh0IZRrQvBRaMiumd9duM/644CMKJT7ZotkvkQQgghnCLBRyMO//4/e9ajc/dA3Nxtdbk7jpew43gpbmoX+gXZ1nOR4EMIIYRwjgQfjdDv2EFBbb1Hwy6Xj9ZnAjCpbwQ6c5ntQZ2sqyKEEEI4Q4KP05QZyvDOVlHp1QmVSiG2dxAAJ0r0LNubA8Ddw2Pti8pJ5kMIIYRwjlPBR0yMbdn2078efvhhwLYK7OzZs4mIiMDDw4OUlBT279/fKg1vLVu3fE+Nt21uj8gEP9y9bBN7LdyYhVWB4fGBdI/wsS8qJ8GHEEII4Ryngo/U1FRycnLsX7/88gsAN95omwX09ddfZ968ebzzzjukpqYSFhbG1VdfTUVFRcu3vJVkrl9Gft0ol/6hAFQazHy5NRuAey+Ls+1Yl/nwkG4XIYQQwhlOBR/BwcGEhYXZv5YsWUKXLl0YMWIEiqIwf/58nnvuOSZPnkzPnj1ZuHAher2ezz//vLXa36KMFiPWvScp940DFPtCcotTs6kwmIkL9mREYm0NSHVdt4sEH0IIIYQzzrvmw2g0smjRIu6++25UKhWZmZnk5uYyevRo+z5arZYRI0awcePGJo9jMBgoLy93+GovW3O3EljZA4DgQBWevlosVoX/brAVmt5zWSwuLiow1YDRtuCbdLsIIYQQzjnv4OOHH36gtLSU6dOnA5CbmwtAaGiow36hoaH2bY155ZVX8PX1tX9FRUWdb5Mu2Pr9yzDp+gIQnxwJwMr9uZwoqcZf58rkfrULyNVlPVRqcPdth5YKIYQQl67zDj4++ugjxo0bR0REhMPjqtMWWVMU5YzHGpo1axZlZWX2r+zs7PNt0gWxKlbyNmym1C8egC5DOwP1w2unJUfj4VY7sZi92DQAWmFROSGEEOKP7LxWtT127BirVq3iu+++sz8WFhYG2DIg4eH1K7Hm5+efkQ1pSKvVotVqz6cZLWpv4V4iT0SjuKrxVVfgG+zBzuMlbDtWgqtaxR1Do+t3lmG2QgghxHk7r8zHxx9/TEhICBMmTLA/FhsbS1hYmH0EDNjqQn7//XeGDRt24S1tZb8d/w0/o22IbXSMbXitfVKxPpGE+LjX7yzDbIUQQojz5nTwYbVa+fjjj7nzzjvRaOoTJyqViscff5w5c+bw/fffs2/fPqZPn45Op2Pq1Kkt2ujWsCF9LUaPbgAkpMRzsrSaZftstSr3XBbruHNd8CGLygkhhBBOc7rbZdWqVRw/fpy77777jG1PPfUU1dXVzJgxg5KSEpKTk1m5ciXe3t4t0tjWcrTsKMF7vVBcXNEZCgkZkMKry9OwWBWGdamdVKyh6hLbv5L5EEIIIZzmdPAxevRoFEVpdJtKpWL27NnMnj37QtvVplYfX02XotpZTT2K0JusfLH1ONBI1gOk20UIIYS4ALK2C7A663c8rLb5PWK6+fD1tmwqaszEBXkysmvImU9oONpFCCGEEE7p8MFHgb6A0iNGcNGirSkh/LLu9knF7q6bVOx0MtpFCCGEOG8dPvhYnb2aHjm9AQgu2csGVRDZxdX46Vy5oX+nxp8k3S5CCCHEeevwwcdvx1YTWd4LgE6+lXy0xTbJ2bTkzvWTip1OFpUTQgghzluHDj4qjZVkpxWhxhNXYwVusQGkZtVNKhbT9BOl5kMIIYQ4bx06+Fh/aj2dC22FpkGFe1jpYpuJdWLvCEIbTirWkKkGTFW276XbRQghhHBahw4+fjv2G3FFtnqPkMJdfFJhWyTu7saG19aRReWEEEKIC9Jhgw+TxcTBA8fQmX1Rm6tRu1ZSpvFgaFwgPSPPElTIonJCCCHEBemwwUdqbiph+QkABBXtZbunbXXeRicVa0iG2QohhBAXpMMGH78d/4244tohtgW72OUXTWyQJ6O6NTKpWEMyzFYIIYS4IB0y+LAqVnYcOICPIQgXi5HA4oPsD4jl7uExjU8q1pAsKieEEEJckA4ZfBwoOoDvKdsEYoHFByh3dUMfFMYNA5qYVKwhWVROCCGEuCAdMviwdbnYFpILLtzFgYAYpg2JRufWjHX2pNtFCCGEuCAdMvjYfGAHAdXhoFgILNrHoaDYs08q1pBMMCaEEEJckA4XfGSVZaE+bqvX8C07gqu5msDkQYT5NjGp2OlktIsQQghxQTpc8LE6e7V9YrGwvO0YXDRMnDKy+QeQbhchhBDignS44GP9oS2EVEWjoBBcuIfc8Dh6xgQ1/wCS+RBCCCEuSIcKPgqrC6k+YisqVdXk4maqwD95oHMHkaG2QgghxAXpUMHH79m/E1vcC4Cg/N0AJF11WfMPIIvKCSGEEBesYwUf6RsIK+8CQMLJDQB49u/X/APIonJCCCHEBeswwUeVqYqCA9W44EKFtRwPQzGuXbqg9vNr/kFkUTkhhBDignWY4ENv0jOoehQA6opjAHj27+/kQaTYVAghhLhQHSb48MEP91zbqJbOuVsA8HA6+JBhtkIIIcSF6jDBh9lkpSTCjWNqMz3y9wKgc6beA2SkixBCCNECmrGYyR9DBQof68tINB5FbTGjDgzEtXNn5w4ii8oJIYQQF6zDZD7cXV2YMTKe611t2Qtd/36onC0alW4XIYQQ4oJ1mODDT+fGn69OZCz5AHj0c7LeA2RROSGEEKIFdJjgA0BRFKp37gTOo94DJPMhhBBCtIAOFXwYM7OwlJSg0mpx797d+QPIUFshhBDignWo4KN65w4A3Hv1ROXm5vwBJPgQQgghLliHCj70O2zBh67/gPM8gAy1FUIIIS5Uhwo+qnfY6j08zqfeQxaVE0IIIVpEhwk+zCUlGDMzAdD17ev8AWRROSGEEKJFdJhJxlzc3Ij45+sYs445t5hcHVlUTgghhGgRHSf48PTEd+LE8z+AFJsKIYQQLaLDdLtcMJnjQwghhGgRTgcfJ0+e5LbbbiMwMBCdTkffvn3Zvn27ffv06dNRqVQOX0OGDGnRRrcLGekihBBCtAinul1KSkoYPnw4I0eOZNmyZYSEhJCRkYHfaTUUY8eO5eOPP7b/7HY+c2pcbGRROSGEEKJFOBV8vPbaa0RFRTkEFjExMWfsp9VqCQsLu+DGXVSk20UIIYRoEU51u/z4448MHDiQG2+8kZCQEPr168eHH354xn5r1qwhJCSExMRE7rvvPvLz85s8psFgoLy83OHroiTBhxBCCNEinAo+jh49ynvvvUdCQgIrVqzgwQcf5LHHHuOTTz6x7zNu3Dg+++wzfvvtN+bOnUtqaiqjRo3CYDA0esxXXnkFX19f+1dUVNSFvaLWIivaCiGEEC1CpSiK0tyd3dzcGDhwIBs3brQ/9thjj5GamsqmTZsafU5OTg7R0dF8+eWXTJ48+YztBoPBITApLy8nKiqKsrIyfHx8nHktrevfIyBnF0xdDIlj2rs1QgghxEWlvLwcX1/fZl2/ncp8hIeH0/201WCTkpI4fvz4WZ8THR1Nenp6o9u1Wi0+Pj4OXxclmedDCCGEaBFOBR/Dhw8nLS3N4bHDhw8THR3d5HOKiorIzs4mPDz8/Fp4sZChtkIIIUSLcCr4eOKJJ9i8eTNz5szhyJEjfP7553zwwQc8/PDDAFRWVjJz5kw2bdpEVlYWa9asYeLEiQQFBXH99de3ygtoE7KonBBCCNFinAo+Bg0axPfff88XX3xBz549efHFF5k/fz7Tpk0DQK1Ws3fvXq699loSExO58847SUxMZNOmTXh7e7fKC2gTsqicEEII0WKcXtvlmmuu4Zprrml0m4eHBytWrLjgRl10ZFE5IYQQosXI2i7NIcWmQgghRIuR4KM5ZIIxIYQQosVI8NEcMtJFCCGEaDESfDSHdLsIIYQQLUaCj+aoluBDCCGEaCkSfDSH1HwIIYQQLUaCj+aQReWEEEKIFiPBR3NIzYcQQgjRYiT4aA4JPoQQQogWI8FHc8hQWyGEEKLFSPBxLrKonBBCCNGiJPg4F1lUTgghhGhREnyciywqJ4QQQrQoCT7OReb4EEIIIVqUBB/nIiNdhBBCiBYlwce5yARjQgghRIuS4ONc6jIfHhJ8CCGEEC1Bgo9zkUXlhBBCiBYlwce5SMGpEEII0aIk+DgXqfkQQgghWpQEH+cio12EEEKIFiXBx7lI8CGEEEK0KAk+zkUWlRNCCCFalAQfZyOLygkhhBAtToKPs5FF5YQQQogWJ8HH2TQcZiuLygkhhBAtQoKPs5FhtkIIIUSLk+DjbGSkixBCCNHiJPg4G8l8CCGEEC1Ogo+zkUXlhBBCiBYnwcfZyKJyQgghRIuT4ONsZFE5IYQQosVJ8HE2UvMhhBBCtDgJPs5GRrsIIYQQLU6Cj7OR4EMIIYRocRJ8nI0sKieEEEK0OKeDj5MnT3LbbbcRGBiITqejb9++bN++3b5dURRmz55NREQEHh4epKSksH///hZtdJuQReWEEEKIVuFU8FFSUsLw4cNxdXVl2bJlHDhwgLlz5+Ln52ff5/XXX2fevHm88847pKamEhYWxtVXX01FRUVLt711yaJyQgghRKvQOLPza6+9RlRUFB9//LH9sZiYGPv3iqIwf/58nnvuOSZPngzAwoULCQ0N5fPPP+eBBx5omVa3BVlUTgghhGgVTmU+fvzxRwYOHMiNN95ISEgI/fr148MPP7Rvz8zMJDc3l9GjR9sf02q1jBgxgo0bNzZ6TIPBQHl5ucPXRUGG2QohhBCtwqng4+jRo7z33nskJCSwYsUKHnzwQR577DE++eQTAHJzcwEIDQ11eF5oaKh92+leeeUVfH197V9RUVHn8zpanox0EUIIIVqFU8GH1Wqlf//+zJkzh379+vHAAw9w33338d577znspzqtm0JRlDMeqzNr1izKysrsX9nZ2U6+hFYimQ8hhBCiVTgVfISHh9O9e3eHx5KSkjh+/DgAYWFhAGdkOfLz88/IhtTRarX4+Pg4fF0UZFE5IYQQolU4FXwMHz6ctLQ0h8cOHz5MdHQ0ALGxsYSFhfHLL7/YtxuNRn7//XeGDRvWAs1tQ7KonBBCCNEqnBrt8sQTTzBs2DDmzJnDTTfdxNatW/nggw/44IMPAFt3y+OPP86cOXNISEggISGBOXPmoNPpmDp1aqu8gFYji8oJIYQQrcKp4GPQoEF8//33zJo1i3/84x/ExsYyf/58pk2bZt/nqaeeorq6mhkzZlBSUkJycjIrV67E29u7xRvfqqTmQwghhGgVKkVRlPZuREPl5eX4+vpSVlbWvvUf/x4BObtg6mJIHNN+7RBCCCEuAc5cv2Vtl6bIUFshhBCiVUjw0RTpdhFCCCFahQQfjWm4qJwMtRVCCCFalAQfjZFF5YQQQohWI8FHY2RROSGEEKLVSPDRGKn3EEIIIVqNBB+NkZEuQgghRKuR4KMxkvkQQgghWo0EH42RReWEEEKIViPBR2NkUTkhhBCi1Ujw0RhZVE4IIYRoNRJ8NEZqPoQQQohWI8FHYyTzIYQQQrQaCT4aoy+x/SvBhxBCCNHiJPhojHS7CCGEEK1Ggo/TyaJyQgghRKuS4ON0sqicEEII0aok+DidLConhBBCtCoJPk4n9R5CCCFEq5Lg43SyqJwQQgjRqiT4OJ1kPoQQQohWJcHH6WRROSGEEKJVSfBxOpndVAghhGhVEnycTla0FUIIIVqVBB+nk8yHEEII0aok+DidFJwKIYQQrUqCj9PJonJCCCFEq5Lg43SS+RBCCCFalQQfDcmickIIIUSrk+CjIVlUTgghhGh1Enw0JIvKCSGEEK1Ogo+GpN5DCCGEaHUSfDQki8oJIYQQrU6Cj4Yk8yGEEEK0Ogk+GpJF5YQQQohWJ8FHQzK1uhBCCNHqnAo+Zs+ejUqlcvgKCwuzb58+ffoZ24cMGdLijW41sqicEEII0eo0zj6hR48erFq1yv6zWq122D527Fg+/vhj+89ubm4X0Lw2JpkPIYQQotU5HXxoNBqHbMfptFrtWbdf1KTgVAghhGh1Ttd8pKenExERQWxsLLfccgtHjx512L5mzRpCQkJITEzkvvvuIz8/v8Ua2+pkUTkhhBCi1TmV+UhOTuaTTz4hMTGRvLw8XnrpJYYNG8b+/fsJDAxk3Lhx3HjjjURHR5OZmcnzzz/PqFGj2L59O1qtttFjGgwGDAaD/efy8vILe0UXQjIfQgghRKtTKYqinO+Tq6qq6NKlC0899RR//vOfz9iek5NDdHQ0X375JZMnT270GLNnz+aFF1444/GysjJ8fHzOt2nOM9XAy6G2758+Bh5+bXduIYQQ4hJXXl6Or69vs67fFzTU1tPTk169epGent7o9vDwcKKjo5vcDjBr1izKysrsX9nZ2RfSpPMni8oJIYQQbcLpgtOGDAYDBw8e5PLLL290e1FREdnZ2YSHhzd5DK1W22SXTJuSReWEEEKINuFU5mPmzJn8/vvvZGZmsmXLFqZMmUJ5eTl33nknlZWVzJw5k02bNpGVlcWaNWuYOHEiQUFBXH/99a3V/pYj9R5CCCFEm3Aq83HixAluvfVWCgsLCQ4OZsiQIWzevJno6Giqq6vZu3cvn3zyCaWlpYSHhzNy5Ei++uorvL29W6v9LUfm+BBCtCCLxYLJZGrvZgjRolxdXc+Y3+t8OBV8fPnll01u8/DwYMWKFRfcoHZjX9FWMh9CiPOnKAq5ubmUlpa2d1OEaBV+fn6EhYWhuoAShQuq+fhD0cvU6kKIC1cXeISEhKDT6S7oA1qIi4miKOj1evv8XWer5zwXCT7q1HW7yIq2QojzZLFY7IFHYKDcyIg/Hg8PDwDy8/MJCQk57y4YWdW2jiwqJ4S4QHU1Hjqdrp1bIkTrqfv9vpCaJgk+6kjBqRCihUhXi/gja4nfbwk+6shQWyGEEKJNSPBRRxaVE0J0YEVFRYSEhJCVldXo9gULFuDn59embRJt65133mHSpEltci4JPupI5kMI0YG98sorTJw4kZiYGACysrIc0us333wzhw8fbqfWOS8mJoY1a9Y0e/81a9bYX3tHcfprvu+++0hNTWX9+vWtfm4JPgBM1WCqsn0vo12EEB1MdXU1H330Effee2+T+3h4eBASEtKGreqY2nNiOq1Wy9SpU3n77bdb/VwSfED9HB+yqJwQogNatmwZGo2GoUOHNrnP6d0us2fPpm/fvvz3v/+lc+fOeHl58dBDD2GxWHj99dcJCwsjJCSEl19+2eE48+bNo1evXnh6ehIVFcWMGTOorKx02OfDDz8kKioKnU7H9ddfz7x5887o8vnpp58YMGAA7u7uxMXF8cILL2A2my/4vaiTkZHBtddeS2hoKF5eXgwaNIhVq1bZt8+aNYshQ4ac8bzevXvz97//3f7zxx9/TFJSEu7u7nTr1o13333Xvq0uu7R48WJSUlJwd3dn0aJFFBUVceutt9KpUyd0Oh29evXiiy++cDhPRUUF06ZNw9PTk/DwcN58801SUlJ4/PHH7fsYjUaeeuopIiMj8fT0JDk5+ZzZoEmTJvHDDz9QXV3t5DvmHJnnAxyH2UqVuhCihSiKQrXJ0i7n9nBVN3tUwtq1axk4cKDT58jIyGDZsmUsX76cjIwMpkyZQmZmJomJifz+++9s3LiRu+++myuvvNJ+oXZxceGtt94iJiaGzMxMZsyYwVNPPWW/KG/YsIEHH3yQ1157jUmTJrFq1Sqef/55h/OuWLGC2267jbfeeovLL7+cjIwM7r//fgCHC/+FqKysZPz48bz00ku4u7uzcOFCJk6cSFpaGp07d2batGm8+uqrZGRk0KVLFwD279/P3r17+eabbwBbEPX3v/+dd955h379+rFz507uu+8+PD09ufPOO+3nevrpp5k7dy4ff/wxWq2WmpoaBgwYwNNPP42Pjw8///wzt99+O3FxcSQnJwPw5z//mQ0bNvDjjz8SGhrK3/72N3bs2EHfvn3tx73rrrvIysriyy+/JCIigu+//56xY8eyd+9eEhISGn3dAwcOxGQysXXrVkaMGNEi72VjVIqiKK129PNQXl6Or68vZWVl+Pj4tM1Jj66BT66F4G7w8Ja2OacQ4g+npqaGzMxMYmNjcXd3R2800/1v7bPsxIF/jEHn1rz7y+uuu47AwEA++uijJvdZsGABjz/+uH3a+NmzZ/PPf/6T3Nxc+/pdY8eOJS0tjYyMDFxcbIn1bt26MX36dJ555plGj/v111/z0EMPUVhYCMAtt9xCZWUlS5Ysse9z2223sWTJEvu5r7jiCsaNG8esWbPs+yxatIinnnqKU6dONes1n48ePXrw0EMP8cgjjwDQp08fpkyZYg+Onn32WVatWsXWrVsB6Ny5M6+99hq33nqr/RgvvfQSS5cuZePGjWRlZREbG8v8+fP505/+dNZzT5gwgaSkJN544w0qKioIDAzk888/Z8qUKQCUlZURERHBfffdx/z588nIyCAhIYETJ04QERFhP85VV13F4MGDmTNnTpPnCggI4M0333QIkBo6/fe8jjPXb8l8gMzxIYTo0Kqrqx0uIs0VExPjsHBoaGgoarXaHnjUPVY3HTfA6tWrmTNnDgcOHKC8vByz2UxNTQ1VVVV4enqSlpZ2xkrogwcPdghGtm/fTmpqqkOXjsVioaamBr1e3yKTvFVVVfHCCy+wZMkSTp06hdlsprq6muPHj9v3mTZtGv/97395/vnnURSFL774wt7tUVBQQHZ2Nvfccw/33Xef/TlmsxlfX8fu/dOzThaLhVdffZWvvvqKkydPYjAYMBgMeHp6AnD06FFMJhODBw+2P8fX15euXbvaf96xYweKopCYmOhwbIPBcM7Zdz08PNDr9c14l86fBB8gi8oJIVqFh6uaA/8Y027nbq6goCBKSkqcPoerq6vDzyqVqtHHrFYrAMeOHWP8+PE8+OCDvPjiiwQEBLB+/Xruuecee6GloihndBednqC3Wq288MILTJ48+Yw2nU8Q1Zgnn3ySFStW8MYbbxAfH4+HhwdTpkzBaDTa95k6dSrPPPMMO3bsoLq6muzsbG655RZ7G8HW9VLXVVLn9CnJ64KKOnPnzuXNN99k/vz59vqYxx9/3H7uuvfjbO+T1WpFrVazffv2M87n5eV11tdeXFxMcHDwWfe5UBJ8gCwqJ4RoFSqVqtldH+2pX79+LFq0qNXPs23bNsxmM3PnzrVnRxYvXuywT7du3ezdFg2f11D//v1JS0sjPj6+1dq6bt06pk+fbs/CVFZWnjEHSqdOnbjiiiv47LPPqK6u5qqrriI0NBSwZXwiIyM5evQo06ZNc/rc1157LbfddhtgCyTS09NJSkoCoEuXLri6urJ161aioqIAW5dHenq6vU6jX79+WCwW8vPzufzyy5t97oyMDGpqaujXr59TbXbWxf9X0RZkUTkhRAc2ZswYZs2aRUlJCf7+/q12ni5dumA2m3n77beZOHEiGzZs4P3333fY59FHH+WKK65g3rx5TJw4kd9++41ly5Y53OX/7W9/45prriEqKoobb7wRFxcX9uzZw969e3nppZdapK3x8fF89913TJw4EZVKxfPPP2/PZjQ0bdo0Zs+ejdFo5M0333TYNnv2bB577DF8fHwYN24cBoOBbdu2UVJSwp///Oeznvvbb79l48aN+Pv7M2/ePHJzc+3Bh7e3N3feeSdPPvkkAQEBhISE8Pe//x0XFxf7+5SYmMi0adO44447mDt3Lv369aOwsJDffvuNXr16MX78+EbPvW7dOuLi4uxFtK1FhtqCLConhOjQevXqxcCBA8/IQrS0vn37Mm/ePF577TV69uzJZ599xiuvvOKwz/Dhw3n//feZN28effr0Yfny5TzxxBMO3SljxoxhyZIl/PLLLwwaNIghQ4Ywb948oqOjmzx3SkoK06dPb3Zb33zzTfz9/Rk2bBgTJ05kzJgx9O/f/4z9brzxRoqKitDr9Vx33XUO2+69917+85//sGDBAnr16sWIESNYsGABsbGxZz33888/T//+/RkzZgwpKSmEhYWdcex58+YxdOhQrrnmGq666iqGDx9uH9Jb5+OPP+aOO+7gL3/5C127dmXSpEls2bLFni1pzBdffOFQo9JaZLQLwKfXQ8ZvcN370PfWc+8vhBCNaGoUwKVg6dKlzJw5k3379jkUjF4M7rvvPg4dOsS6devO+xgxMTHMnj3bqQDkUlJVVUVkZCRz587lnnvuOa9j7Nu3jyuvvJLDhw+fURTbkIx2aSkytboQooMbP3486enpnDx58qx3xm3hjTfe4Oqrr8bT05Nly5axcOFCh8m5nHXo0CG8vb254447WrCV7Wvnzp0cOnSIwYMHU1ZWxj/+8Q8Arr322vM+5qlTp/jkk0/OGni0FAk+QBaVE0IIOOdcE21l69atvP7661RUVBAXF8dbb7111qnfz6Vbt27s3bu3BVt4cXjjjTdIS0vDzc2NAQMGsG7dOoKCgs77eKNHj27B1p2dBB8gmQ8hhLiItHbtyR9Bv3792L59e3s347xdXB177UEWlRNCCCHalAQfsqicEEII0aYk+JBF5YQQQog2JcGHrOsihBBCtCkJPqTYVAghhGhTEnzIonJCCCFEm5LgQxaVE0IIioqKCAkJOWPxtNaWlZWFSqVi165dZ90vJSXFvly9cDRo0CC+++679m6GUyT4kEXlhBCCV155hYkTJxITEwPUBwVtbc2aNahUKkpLSx0e/+6773jxxRftP6ekpLBgwYJmH7e9Xs+FaOq9mD59OrNnz7b//Pzzz/PMM880uvDdxUqCD1lUTgjRwVVXV/PRRx9d0CyirS0gIABvb+/2bsZFacKECZSVlbFixYr2bkqzSfAho12EEB3csmXL0Gg0DB06tMl9FixYgJ+fH0uWLKFr167odDqmTJlCVVUVCxcuJCYmBn9/fx599FEsFov9eSqVih9++MHhWH5+fo1mLbKyshg5ciQA/v7+qFQq+0JwrdHtsmHDBkaMGIFOp8Pf358xY8ZQUmJbbsNgMPDYY48REhKCu7s7l112Gampqfbn1mUlfv31VwYOHIhOp2PYsGGkpaU5nOPHH39k4MCBuLu7ExQUxOTJk+3bFi1axMCBA/H29iYsLIypU6eSn59/zvfidGq1mvHjx/PFF1+05NvTqiT4kNEuQojWoihgrGqfLycWLF+7di0DBw485356vZ633nqLL7/8kuXLl7NmzRomT57M0qVLWbp0KZ9++ikffPAB33zzzXm9XVFRUXz77bcApKWlkZOTw7/+9a/zOta57Nq1iyuvvJIePXqwadMm1q9fz8SJE+2B01NPPcW3337LwoUL2bFjB/Hx8YwZM4bi4mKH4zz33HPMnTuXbdu2odFouPvuu+3bfv75ZyZPnsyECRPYuXOnPVCpYzQaefHFF9m9ezc//PADmZmZ9gDD2fdi8ODBF7Tqb1uTtV2k4FQI0VpMepgT0T7nfvYUuHk2a9esrCwiIhzbGRMTg3JaAGMymXjvvffo0qULAFOmTOHTTz8lLy8PLy8vunfvzsiRI1m9ejU333yz001Wq9UEBNhuBENCQvDz82ty3zVr1jh17NNfz+uvv87AgQMdVsvt0aMHYFue/r333mPBggWMGzcOgA8//JBffvmFjz76iCeffNL+nJdffpkRI0YA8MwzzzBhwgRqampwd3fn5Zdf5pZbbuGFF16w79+nTx/79w0DlboF9AYPHkxlZSVeXl5NvheNZY0iIyM5fvw4VqsVF5eLP69w8bewtclQWyFEB1ddXY27u/s599PpdPbAAyA0NJSYmBi8vLwcHqvrOriY1WU+GpORkYHJZGL48OH2x1xdXRk8eDAHDx502Ld3797278PDwwHsr/9s5wDYuXMn1157LdHR0Xh7e5OSkgLA8ePHnX49Hh4eWK1WDAaD089tDx078yGLygkhWpOrzpaBaK9zN1NQUJC91uGsh3R1dfhZpVI1+ljDURcqlarRDEp78/DwaHJbXXtPHx2jKMoZjzV8/XXb6l7/2c5RVVXF6NGjGT16NIsWLSI4OJjjx48zZswYjEajcy8GKC4uRqfTnfWcF5OOnfmQReWEEK1JpbJ1fbTHlxPDSvv168eBAwda5S0IDg4mJyfH/nN6ejp6vb7J/d3c3AAcilZbQ+/evfn1118b3RYfH4+bmxvr16+3P2Yymdi2bRtJSUktco5Dhw5RWFjIq6++yuWXX063bt3OyBg5817s27eP/v37N7tt7a1jBx+yqJwQQjBmzBj279/frOyHs0aNGsU777zDjh072LZtGw8++OAZ2ZKGoqOjUalULFmyhIKCAiorK1u8TQCzZs0iNTWVGTNmsGfPHg4dOsR7771HYWEhnp6ePPTQQzz55JMsX76cAwcOcN9996HX67nnnnuafY6///3vfPHFF/z973/n4MGD7N27l9dffx2Azp074+bmxttvv83Ro0f58ccfHeYxAefei3Xr1jF69OjzezPaQccOPmSYrRBC0KtXLwYOHMjixYtb/Nhz584lKiqKK664gqlTpzJz5kx0uqa7hCIjI3nhhRd45plnCA0N5ZFHHmnWeaZPn26vmWiOxMREVq5cye7duxk8eDBDhw7lf//7HxqNrRrh1Vdf5YYbbuD222+nf//+HDlyhBUrVuDv79/sc6SkpPD111/z448/0rdvX0aNGsWWLVsAW0ZowYIFfP3113Tv3p1XX32VN954w+H5zX0vTp48ycaNG7nrrrua3bb2plJO74w7i9mzZztU7YKtuCg3Nxew9Ye98MILfPDBB5SUlJCcnMz//d//2SuIm6O8vBxfX1/Kysrw8fFp9vPOy75v4Zu7IXo43LW0dc8lhPjDq6mpITMzk9jY2GYVcF5Mli5dysyZM9m3b98lMVridCkpKaSkpDjM/NlRPPnkk5SVlfHBBx+0yfma+j135vrtdMFpjx49WLVqlf1ntVpt//71119n3rx5LFiwgMTERF566SWuvvpq0tLSLs6Z6WSkixBCADB+/HjS09M5efIkUVFR7d0cp1RUVJCRkcGSJUvauyntIiQkhJkzZ7Z3M5zidPCh0WgICws743FFUZg/fz7PPfecfQa3hQsXEhoayueff84DDzxw4a1taTLHhxBC2P3pT39q7yacF29vb7Kzs9u7Ge2m4bwjlwqnc2vp6elEREQQGxvLLbfcwtGjRwHIzMwkNzfXoeBFq9UyYsQINm7c2OTxDAYD5eXlDl9tRhaVE0IIIdqcU8FHcnIyn3zyCStWrODDDz8kNzeXYcOGUVRUZK/7CA0NdXhOw5qQxrzyyiv4+vrav9o03SeLygkhhBBtzqngY9y4cdxwww306tWLq666ip9//hmwda/Uac6kLA3NmjWLsrIy+1ebps5ktIsQQgjR5i6opNnT05NevXqRnp5urwM5PcuRn59/RjakIa1Wi4+Pj8NXm5FF5YQQQog2d0HBh8Fg4ODBg4SHhxMbG0tYWBi//PKLfbvRaOT3339n2LBhF9zQViEFp0IIIUSbc2q0y8yZM5k4cSKdO3cmPz+fl156ifLycu68805UKhWPP/44c+bMISEhgYSEBObMmYNOp2Pq1Kmt1f4LI0NthRBCiDbnVPBx4sQJbr31VgoLCwkODmbIkCFs3ryZ6OhoAJ566imqq6uZMWOGfZKxlStXXpxzfDRcVE4yH0IIIUSbcarb5csvv+TUqVMYjUZOnjzJt99+S/fu3e3bVSoVs2fPJicnh5qaGn7//Xd69uzZ4o1uEXVZDxcNaNuwzkQIIS5CRUVFhISEkJWV1abnzcrKQqVSsWvXrrPul5KSwuOPP94mbfqjWrJkCf369XNYdbi9XHpz6LaUumG2HgGyqJwQosN75ZVXmDhxIjExMUB9UNDW1qxZg0qlorS01OHx7777zmHhtZSUFBYsWNDs47bX62lPp7/ma665BpVKxeeff96OrbLpuMGHDLMVQggAqqur+eijj7j33nvbuylNCggIuDi78J1kMpna9fx33XUXb7/9dru2AST4kGJTIUSHt2zZMjQaDUOHDm1ynwULFuDn58eSJUvo2rUrOp2OKVOmUFVVxcKFC4mJicHf359HH30Ui8Vif55KpeKHH35wOJafn1+jWYusrCxGjhwJgL+/PyqViunTpwMt3+1SVFTErbfeSqdOndDpdPTq1YsvvvjCvv3f//43kZGRZ3RRTJo0iTvvvNP+808//cSAAQNwd3cnLi6OF154AbPZbN+uUql4//33ufbaa/H09OSll17CYrFwzz33EBsbi4eHB127duVf//qXw3nMZjOPPfYYfn5+BAYG8vTTT3PnnXdy3XXX2fdRFIXXX3+duLg4PDw86NOnD998881ZX/ekSZPYunWrfXby9uL02i5/GDLSRQjRyhRFodpc3S7n9tB4NLubYe3atQwcOPCc++n1et566y2+/PJLKioqmDx5MpMnT8bPz4+lS5dy9OhRbrjhBi677DJuvvlmp9scFRXFt99+yw033EBaWho+Pj54eHg4fZzmqKmpYcCAATz99NP4+Pjw888/c/vttxMXF0dycjI33ngjjz32GKtXr+bKK68EoKSkhBUrVvDTTz8BsGLFCm677TbeeustLr/8cjIyMrj//vsB+Pvf/24/19///ndeeeUV3nzzTdRqNVarlU6dOrF48WKCgoLYuHEj999/P+Hh4dx0000AvPbaa3z22Wd8/PHHJCUl8a9//YsffvjBHpwB/PWvf+W7777jvffeIyEhgbVr13LbbbcRHBzMiBEjGn3d0dHRhISEsG7dOuLi4lrlvW0OCT6k20UI0UqqzdUkf57cLufeMnULOldds/bNysoiIiLC4bGYmBgURXF4zGQy8d5779GlSxcApkyZwqeffkpeXh5eXl50796dkSNHsnr16vMKPtRqNQEBthvCkJAQ/Pz8mtx3zZo1Th379NcTGRnpsBLso48+yvLly/n6669JTk4mICCAsWPH8vnnn9uDj6+//pqAgAD7zy+//DLPPPOMPRMSFxfHiy++yFNPPeUQfEydOpW7777boT0vvPCC/fvY2Fg2btzI4sWL7cHH22+/zaxZs7j++usBeOedd1i6dKn9OVVVVcybN4/ffvvNnrGKi4tj/fr1/Pvf/2bEiBGN/h/Wvfa2Liw+XQcOPmRROSGEAFvNh7u7+zn30+l09sADbGt3xcTE4OXl5fBYfn5+q7SzJVksFl599VW++uorTp48icFgwGAw4Onpad9n2rRp3H///bz77rtotVo+++wzbrnlFtRqNQDbt28nNTWVl19+2eG4NTU16PV6dDpb8NdYVun999/nP//5D8eOHaO6uhqj0Ujfvn0BKCsrIy8vj8GDB9v3V6vVDBgwwN4NdODAAWpqarj66qsdjms0GunXr99ZX7uHhwd6vd6Jd6vlSfAhmQ8hRCvx0HiwZeqWdjt3cwUFBVFSUnLO/VxdXR1+VqlUjT7WsE5CpVI1mkFpb3PnzuXNN99k/vz59OrVC09PTx5//HGMRqN9n4kTJ2K1Wvn5558ZNGgQ69atY968efbtVquVF154gcmTJ59x/IbBXMOABmDx4sU88cQTzJ07l6FDh+Lt7c0///lPtmxx/F1pbK20hucG+Pnnn4mMjHTYT6vVnvW1FxcXExwcfNZ9WlvHDT5kRVshRCtTqVTN7vpoT/369WPRokWtcuzg4GBycnLsP6enp5/1rtvNzQ3AoWi1Naxbt45rr72W2267DbBdzNPT00lKSrLv4+HhweTJk/nss884cuQIiYmJDBgwwL69f//+pKWlER8f7/S5hw0bxowZM+yPZWRk2L/39fUlNDSUrVu3cvnllwO292Pnzp327Ej37t3RarUcP368yfqOxtTU1JCRkXHO7Ehr67jBh4x2EUIIAMaMGcOsWbMoKSnB39+/RY89atQo3nnnHYYMGYLVauXpp58+I1vSUHR0NCqViiVLljB+/Hg8PDwcunVaSnx8PN9++y0bN27E39+fefPmkZub6xB8gK3rZeLEiezfv98eqNT529/+xjXXXENUVBQ33ngjLi4u7Nmzh7179/LSSy+d9dyffPIJK1asIDY2lk8//ZTU1FRiY2Pt+zz66KO88sorxMfH061bN95++21KSkrs2RBvb29mzpzJE088gdVq5bLLLqO8vJyNGzfi5eXlMCKnoc2bN6PVas86sqktdOChtpL5EEIIgF69ejFw4EAWL17c4seeO3cuUVFRXHHFFUydOpWZM2faayEaExkZyQsvvMAzzzxDaGgojzzySLPOM336dFJSUprdrueff57+/fszZswYUlJSCAsLcxjGWmfUqFEEBASQlpZ2xjplY8aMYcmSJfzyyy8MGjSIIUOGMG/ePPuSI0158MEHmTx5MjfffDPJyckUFRU5ZEEAnn76aW699VbuuOMOhg4dipeXF2PGjHHoznnxxRf529/+xiuvvEJSUhJjxozhp59+cghiTvfFF18wbdq0s/4ftAWV0lgpbDsqLy/H19eXsrIyfHxacdrzlyNsa7s8thMC2m+4kRDij6OmpobMzExiY2ObVcB5MVm6dCkzZ85k3759uLhcevelKSkppKSkMHv27PZuSquwWq0kJSVx0003Ocz06oyCggK6devGtm3bzhqgnEtTv+fOXL87ZreLLConhBAOxo8fT3p6OidPniQqKqq9m+OUiooKMjIyWLJkSXs3pcUcO3aMlStXMmLECAwGA++88w6ZmZkXtEp8ZmYm77777gUFHi2lYwYfsqicEEKc4U9/+lN7N+G8eHt7k52d3d7NaFEuLi4sWLCAmTNnoigKPXv2ZNWqVWfUpDhj8ODBDsN321PHDD5kUTkhhBAXsaioKDZs2NDezWg1l17HXkuQOT6EEEKIdtPBgw8ZZiuEEEK0tQ4afMiickIIIUR76eDBh3S7CCGEEG2tgwYfsqicEEII0V46dvAhmQ8hhBCizXXM4EMWlRNCCAdFRUWEhISQlZXVpufNyspCpVKxa9eus+6XkpLC448/3iZtam8xMTHMnz+/1Y4/c+ZMHnvssVY7fnN0zOBDMh9CCOHglVdeYeLEicTExAD1QUFbW7NmDSqVitLSUofHv/vuO4dpxVNSUliwYEGzj9ter+disGDBAod1b5566ik+/vhjMjMz261NHTT4kNEuQghRp7q6mo8++oh77723vZvSpICAALy9vdu7GX8IISEhjB49mvfff7/d2iDBhxBCdHDLli1Do9GcdZn1BQsW4Ofnx5IlS+jatSs6nY4pU6ZQVVXFwoULiYmJwd/fn0cffRSLxWJ/nkql4ocffnA4lp+fX6NZi6ysLEaOHAmAv78/KpWK6dOnA63T7bJhwwZGjBiBTqfD39+fMWPGUFJSAoDBYOCxxx4jJCQEd3d3LrvsMlJTU+3PrcvQrFixgn79+uHh4cGoUaPIz89n2bJlJCUl4ePjw6233oper7c/LyUlhUceeYRHHnkEPz8/AgMD+etf/8rZ1ngtKyvj/vvvJyQkBB8fH0aNGsXu3bsB22JxYWFhzJkzx77/li1bcHNzY+XKlU0ec9KkSXzxxRfn/d5dqI43vbosKieEaCOKoqBUV7fLuVUeHs3uZli7di0DBw485356vZ633nqLL7/8koqKCiZPnszkyZPx8/Nj6dKlHD16lBtuuIHLLruMm2++2ek2R0VF8e2333LDDTeQlpaGj48PHh4eTh+nOXbt2sWVV17J3XffzVtvvYVGo2H16tX2wOmpp57i22+/ZeHChURHR/P6668zZswYjhw5QkBA/Y3r7Nmzeeedd9DpdNx0003cdNNNaLVaPv/8cyorK7n++ut5++23efrpp+3PWbhwIffccw9btmxh27Zt3H///URHR3Pfffed0U5FUZgwYQIBAQEsXboUX19f/v3vf3PllVdy+PBhgoOD+e9//8t1113H6NGj6datG7fddhszZsxg9OjRTb7+wYMHk52dzbFjx4iOjm7Bd7Z5Ol7wIYvKCSHaiFJdTVr/Ae1y7q47tqPS6Zq1b1ZWFhEREQ6PxcTEnHE3bjKZeO+99+jSpQsAU6ZM4dNPPyUvLw8vLy+6d+/OyJEjWb169XkFH2q12n5hDwkJwc/Pr8l916xZ49SxT389r7/+OgMHDuTdd9+1P9ajRw8AqqqqeO+991iwYAHjxo0D4MMPP+SXX37ho48+4sknn7Q/56WXXmL48OEA3HPPPcyaNYuMjAzi4uIA23u0evVqh+AjKiqKN998E5VKRdeuXdm7dy9vvvlmo8HH6tWr2bt3L/n5+Wi1WgDeeOMNfvjhB7755hvuv/9+xo8fz3333ce0adMYNGgQ7u7uvPrqq/ZjTJ8+3Z5BqhMZGQnY/u/bI/joeN0usqicEEI4qK6uxt3d/Zz76XQ6e+ABEBoaSkxMDF5eXg6P5efnt0o7W1Jd5qMxGRkZmEwme1AB4OrqyuDBgzl48KDDvr1797Z/Hxoaik6nswcedY+d/n4MGTLEISs1dOhQ0tPTHbqr6mzfvp3KykoCAwPx8vKyf2VmZpKRkWHf74033sBsNrN48WI+++yzc/5/1mWUGnYJtaUOmPmQkS5CiLah8vCg647t7Xbu5goKCrLXOpyNq6ur4zlUqkYfs1qtDj83lkFpb2frzqlr7+ndVoqinPFYw9ffnPfDWVarlfDw8EYzPQ0zQ0ePHuXUqVNYrVaOHTvmEBQ1prjYdiMeHBx83m27EB04+JBiUyFE61KpVM3u+mhP/fr1Y9GiRa1y7ODgYHJycuw/p6enn/Vu283NDaDRLEBL6t27N7/++isvvPDCGdvi4+Nxc3Nj/fr1TJ06FbAFTNu2bWuRotfNmzef8XNCQgJqtfqMffv3709ubi4ajcY+DPp0RqORadOmcfPNN9OtWzfuuece9u7dS2hoaJNt2LdvH66urvauprbW8bpdZKSLEEI4GDNmDPv3729W9sNZo0aN4p133mHHjh1s27aNBx988IzsQEPR0dGoVCqWLFlCQUEBlZWVLd4mgFmzZpGamsqMGTPYs2cPhw4d4r333qOwsBBPT08eeughnnzySZYvX86BAwe477770Ov13HPPPRd87uzsbP785z+TlpbGF198wdtvv82f/vSnRve96qqrGDp0KNdddx0rVqwgKyuLjRs38te//pVt27YB8Nxzz1FWVsZbb73FU089RVJS0jnbuW7dOi6//PJWK+g9lw4cfEi3ixBCAPTq1YuBAweyePHiFj/23LlziYqK4oorrmDq1KnMnDkT3VmyQZGRkbzwwgs888wzhIb+f3t3HtXUnf4P/B0gCYsELYuAsqkFtHAoohTc4GiLoiLWHhV1FJdq7ahFK+PSmRZ7xhkZrbu1VIdCq9Y6rUs7gusoaOtGNa24DGIMiopSqQLKTp7fH/MjXwNJIEA2fF7n5BxJPveTz8OT6324uZ/76Yr58+e36H2mT5+uciOt5vj6+uLo0aP49ddfERoaivDwcHz//fewsvrfFwLJycl46623MHXqVPTt2xc3b97EkSNH0KVLlxa/hybTpk1DZWUlQkNDMW/ePCxYsABz5sxR21YgECAzMxNDhgzBzJkz4evri7i4OBQUFKBr167IysrChg0bsGPHDkgkElhYWGDHjh348ccf8dlnn2kcw+7du9Ve4GooAtI2udgIysrK4ODggNLSUkgkepiNkrkEuPA5MHgxMOyj9u+fMfbCqqqqglwuh4+PT4su4DQlmZmZSExMxJUrV2BhYX5/l0ZGRiIyMhIrVqww9lC0ioyMxKuvvqrX26c3JyMjA3/6059w+fJlZbGlC02fc12O3y/uNR+8oi1jjCmNHDkS+fn5uHfvHjw8PIw9HJ2Ul5dDJpPh4MGDxh6KWXj27BnS0tJaVXi0lxev+OBF5RhjTC1N1x2YOnt7exQWFhp7GGZjwoQJxh7CC1h88FRbxhhjRqLrzdE6KvP7Yq+teLYLY4wxZlRtKj5WrVoFgUCgMu95+vTp/5vb/twjLCysreNsP1x8MMYYY0bV6q9dcnJysG3bNrV3URsxYgTS0tKUPzfcNMboeFE5xhhjzOhadebj6dOnmDJlCrZv3652zrNYLIarq6vy8fwKgEbFi8oxxhhjRteq4mPevHkYNWoUXn/9dbWvZ2VlwcXFBb6+vpg9e7bWRYaqq6tRVlam8tCb56fZ8qJyjDHGmFHo/LXLN998g0uXLiEnJ0ft69HR0Rg/fjy8vLwgl8vx4YcfYujQobh48aJyOeDnrVq1Su299fWCp9kyxhhjRqfTmY/CwkIkJCRg586dGu/eN3HiRIwaNQoBAQGIiYnBoUOHcOPGDWRkZKhtv3z5cpSWliofep2rzYvKMcaYWiUlJXBxcUFBQYGxh8Ja6eDBgwgODm7TKrqGolPxcfHiRRQXFyMkJARWVlawsrJCdnY2Nm3aBCsrK7WrELq5ucHLywv5+flq+xSLxZBIJCoPveGZLowxptaqVasQExOjXDm1oKBAZfn49PR0lSXc24umfiMjI5Gent7ifhqP90XQOObRo0dDIBDg66+/NuKoWkanr12GDRuG3NxcledmzJgBf39/LF26VO1ywCUlJSgsLISbm1vbRtoeeFE5xhhrorKyEqmpqcjMzDT2UMxebW2t1lV79W3GjBnYvHkz/vCHPxhtDC2h05kPe3t7BAQEqDzs7Ozg6OiIgIAAPH36FImJiTh79iwKCgqQlZWFmJgYODk54c0339RXDC3HdzdljLEmDh06BCsrK4SHh6t9PSsrCzNmzEBpaany/k0NC7jV1NRgyZIl6NatG+zs7PDaa68p7+JZVVWFV155RWXFVrlcDgcHB2zfvl1rv21VUlKCSZMmoXv37rC1tUVgYCB2796tfP3zzz9Ht27dmnxFMWbMGMTHxyt//ve//42QkBBYW1ujR48e+Pjjj1FXV6d8XSAQICUlBbGxsbCzs8PKlStRX1+PWbNmwcfHBzY2NvDz88PGjRtV3qeurg7vvfceOnfuDEdHRyxduhTx8fEYO3assg0RYfXq1ejRowdsbGwQFBSE7777TmvcY8aMwYULF3Dr1q3W/NoMh9ooIiKCEhISiIiooqKCoqKiyNnZmYRCIXl6elJ8fDzduXOnxf2VlpYSACotLW3r0Jr6diZRkoTop83t3zdj7IVXWVlJ165do8rKSiIiUigUVFNVZ5SHQqFo8bgTEhJoxIgRKs/J5XJqOERUV1fThg0bSCKRUFFRERUVFVF5eTkREU2ePJkGDBhAp06dops3b9KaNWtILBbTjRs3iIhIKpWSSCSi/fv3U11dHQ0cOJBiY2Ob7TciIoLS0tJaHMPz4yUiunv3Lq1Zs4akUinJZDLatGkTWVpa0rlz54iIqKSkhEQiER0/fly5ze+//04ikYiOHDlCRESHDx8miURC6enpJJPJ6OjRo+Tt7U0rVqxQbgOAXFxcKDU1lWQyGRUUFFBNTQ199NFHdOHCBbp16xbt3LmTbG1tac+ePcrtVq5cSS+99BLt27ePrl+/TnPnziWJRKL83RARffDBB+Tv70+HDx8mmUxGaWlpJBaLKSsrS23MDVxcXCg9Pb3FvztdNf6cN9Dl+C0gIjJm8dOYLkvy6mzHm4DsBDA2BXh1Uvv2zRh74TVeary2uh7bErKNMpY5GyMgFDf9KlydsWPHwtHREampqRrbpKenY+HChXjy5InyOZlMhpdffhl3796Fu7u78vnXX38doaGh+Pvf/w4AWLNmDVavXo1Jkybh22+/RW5uLpycnDT2qy+jRo1C79698cknnwAAYmNj4eTkpIx727ZtSEpKwt27d2FpaYkhQ4YgOjoay5cvV/axc+dOLFmyBPfv3wcA5V2+169fr/W9582bh4cPHyrPXLi6uiIxMRGJiYkAgPr6evTo0QPBwcE4cOAAnj17BicnJ5w4cULljNTbb7+NiooKrdd19O3bF7GxsUhKSmrFb6l5jT/nDXQ5fr9YC8vx1y6MMdZEZWWlxhmM2ly6dAlEBF9fX5Xnq6ur4ej4f//PLl68GN9//z02b96MQ4cOKQsPfaqvr0dycjL27NmDe/fuobq6GtXV1bCzs1O2mTJlCubMmYOtW7dCLBZj165diIuLU16/ePHiReTk5OBvf/ubSr9VVVWoqKiAra0tAKBfv35N3j8lJQX//Oc/cfv2bVRWVqKmpgavvvoqAKC0tBQPHz5EaGiosr2lpSVCQkKUXwNdu3YNVVVVeOONN1T6rampQXBwsNbYbWxsUFFRocNvy/BesOKDZ7swxgzHSmSBORsjjPbeLeXk5ITHjx/r/B4KhQKWlpa4ePFikwkHnTp1Uv67uLgYeXl5sLS0RH5+PkaMGKHze+lq7dq1WL9+PTZs2IDAwEDY2dlh4cKFqKmpUbaJiYmBQqFARkYG+vfvj9OnT2PdunUq8X388ccYN25ck/6fL9aeL2gA4F//+hcWLVqEtWvXIjw8HPb29lizZg3Onz+v0q7x7Jznv4hoKEIyMjLQrVs3lXbq7pn1vN9//x3Ozs5a2xgbFx+MMaYnAoGgxV99GFNwcDB27typtY1IJGpyO4Xg4GDU19ejuLgYgwcP1rjtzJkzERAQgNmzZ2PWrFkYNmwY+vTpo7Hf9nD69GnExsYqZ30oFArk5+ejd+/eyjY2NjYYN24cdu3ahZs3b8LX1xchISHK1/v27Yu8vDz06tVL5/ceMGAA/vjHPyqfk8lkyn87ODiga9euuHDhgvL3Vl9fD6lUqjw70qdPH4jFYty5cwcRES0vYKuqqiCTyZo9O2JsL07xwYvKMcaYWsOHD8fy5cvx+PFjtet1AYC3tzeePn2K//znPwgKCoKtrS18fX0xZcoUTJs2DWvXrkVwcDAePXqEEydOIDAwECNHjsSnn36Ks2fP4vLly/Dw8MChQ4cwZcoUnD9/HiKRSG2/DV9ntEWvXr2wd+9enDlzBl26dMG6devw4MEDleID+N9XLzExMbh69WqT6akfffQRRo8eDQ8PD4wfPx4WFha4fPkycnNzsXLlSq3v/dVXX+HIkSPw8fHBjh07kJOTAx8fH2WbBQsWYNWqVejVqxf8/f2xefNmPH78WHk2xN7eHomJiVi0aBEUCgUGDRqEsrIynDlzBp06dVKZkfO8c+fOQSwWa5y5ZDL0cSVsW+httkv1M6Ks1USHlhHpcBU4Y4y1lKZZAOYgLCyMUlJStLaZO3cuOTo6EgBKSkoiIlLO7PD29iahUEiurq705ptv0uXLl+n69etkY2NDX3/9tbKP0tJS8vb2piVLlmjtt7H4+HiKiIhocTwlJSUUGxtLnTp1IhcXF/rLX/5C06ZNU5lNQkRUV1dHbm5uBIBkMlmTfg4fPkwDBgwgGxsbkkgkFBoaStu2bVO+DoD279+vsk1VVRVNnz6dHBwcqHPnzvTuu+/SsmXLKCgoSNmmtraW5s+fTxKJhLp06UJLly6l8ePHU1xcnLKNQqGgjRs3kp+fHwmFQnJ2dqbhw4dTdna2xrjnzJlD77zzTot/T63Bs10YY8yEaJoFYA4yMzORmJiIK1euwMKiVWuO6lVkZCQiIyPb7T4gpkahUKB3796YMGEC/vrXv7aqj99++w3+/v74+eefVc6ytDee7cIYY6xdjBw5Evn5+bh37x48PDyMPRwV5eXlkMlkOHjwoLGH0m5u376No0ePIiIiAtXV1diyZQvkcjkmT57c6j7lcjm2bt2q18KjvXDxwRhjDACQkJBg7CGoZW9vr99FR43AwsIC6enpSExMBBEhICAAx48fb3JNii5CQ0NVpu+aMi4+GGOMMQPz8PDATz/9ZOxhGI3pfbHHGGOMsQ6Niw/GGGOMGRQXH4wx1s4ar5TKWEfSHp9vvuaDMcbaiUgkgoWFBe7fvw9nZ2eIRKImt9BmzFwREWpqavDbb7/BwsICIpGo1X1x8cEYY+3EwsICPj4+KCoqUq56ylhHY2trC09PzzbdD4aLD8YYa0cikQienp6oq6vTy5oljBmTpaUlrKys2nxGj4sPxhhrZwKBAEKhEEKh0NhDYcwk8QWnjDHGGDMoLj4YY4wxZlBcfDDGGGPMoEzumo+GRXbLysqMPBLGGGOMtVTDcbvhOK6NyRUf5eXlAGByqyoyxhhjrHnl5eVwcHDQ2kZALSlRDEihUOD+/fuwt7dv95vzlJWVwcPDA4WFhZBIJO3atyno6PEBHT9Gjs/8dfQYO3p8QMePUV/xERHKy8vh7u7e7D1ATO7Mh4WFBbp3767X95BIJB3yA9Wgo8cHdPwYOT7z19Fj7OjxAR0/Rn3E19wZjwZ8wSljjDHGDIqLD8YYY4wZ1AtVfIjFYiQlJUEsFht7KHrR0eMDOn6MHJ/56+gxdvT4gI4foynEZ3IXnDLGGGOsY3uhznwwxhhjzPi4+GCMMcaYQXHxwRhjjDGD4uKDMcYYYwbV4YqPrVu3wsfHB9bW1ggJCcHp06e1ts/OzkZISAisra3Ro0cPpKSkGGikulm1ahX69+8Pe3t7uLi4YOzYscjLy9O6TVZWFgQCQZPHf//7XwONWjcrVqxoMlZXV1et25hL/gDA29tbbT7mzZuntr055O/UqVOIiYmBu7s7BAIBDhw4oPI6EWHFihVwd3eHjY0NIiMjcfXq1Wb73bt3L/r06QOxWIw+ffpg//79eopAO23x1dbWYunSpQgMDISdnR3c3d0xbdo03L9/X2uf6enpavNaVVWl52iaai5/06dPbzLOsLCwZvs1lfwBzceoLhcCgQBr1qzR2Kep5LAlxwVT3Qc7VPGxZ88eLFy4EH/+858hlUoxePBgREdH486dO2rby+VyjBw5EoMHD4ZUKsUHH3yA9957D3v37jXwyJuXnZ2NefPm4dy5czh27Bjq6uoQFRWFZ8+eNbttXl4eioqKlI+XX37ZACNunVdeeUVlrLm5uRrbmlP+ACAnJ0cltmPHjgEAxo8fr3U7U87fs2fPEBQUhC1btqh9ffXq1Vi3bh22bNmCnJwcuLq64o033lCu4aTO2bNnMXHiREydOhW//vorpk6digkTJuD8+fP6CkMjbfFVVFTg0qVL+PDDD3Hp0iXs27cPN27cwJgxY5rtVyKRqOS0qKgI1tbW+ghBq+byBwAjRoxQGWdmZqbWPk0pf0DzMTbOwxdffAGBQIC33npLa7+mkMOWHBdMdh+kDiQ0NJTmzp2r8py/vz8tW7ZMbfslS5aQv7+/ynPvvPMOhYWF6W2M7aW4uJgAUHZ2tsY2J0+eJAD0+PFjww2sDZKSkigoKKjF7c05f0RECQkJ1LNnT1IoFGpfN7f8AaD9+/crf1YoFOTq6krJycnK56qqqsjBwYFSUlI09jNhwgQaMWKEynPDhw+nuLi4dh+zLhrHp86FCxcIAN2+fVtjm7S0NHJwcGjfwbUDdfHFx8dTbGysTv2Yav6IWpbD2NhYGjp0qNY2pprDxscFU94HO8yZj5qaGly8eBFRUVEqz0dFReHMmTNqtzl79myT9sOHD8fPP/+M2tpavY21PZSWlgIAXnrppWbbBgcHw83NDcOGDcPJkyf1PbQ2yc/Ph7u7O3x8fBAXF4dbt25pbGvO+aupqcHOnTsxc+bMZhdQNKf8PU8ul+PBgwcqORKLxYiIiNC4TwKa86ptG1NRWloKgUCAzp07a2339OlTeHl5oXv37hg9ejSkUqlhBtgKWVlZcHFxga+vL2bPno3i4mKt7c05fw8fPkRGRgZmzZrVbFtTzGHj44Ip74Mdpvh49OgR6uvr0bVrV5Xnu3btigcPHqjd5sGDB2rb19XV4dGjR3oba1sREd5//30MGjQIAQEBGtu5ublh27Zt2Lt3L/bt2wc/Pz8MGzYMp06dMuBoW+61117DV199hSNHjmD79u148OABBgwYgJKSErXtzTV/AHDgwAE8efIE06dP19jG3PLXWMN+p8s+2bCdrtuYgqqqKixbtgyTJ0/WuliXv78/0tPT8cMPP2D37t2wtrbGwIEDkZ+fb8DRtkx0dDR27dqFEydOYO3atcjJycHQoUNRXV2tcRtzzR8AfPnll7C3t8e4ceO0tjPFHKo7LpjyPmhyq9q2VeO/IolI61+W6tqre96UzJ8/H5cvX8aPP/6otZ2fnx/8/PyUP4eHh6OwsBCffPIJhgwZou9h6iw6Olr578DAQISHh6Nnz5748ssv8f7776vdxhzzBwCpqamIjo6Gu7u7xjbmlj9NdN0nW7uNMdXW1iIuLg4KhQJbt27V2jYsLEzlos2BAweib9++2Lx5MzZt2qTvoepk4sSJyn8HBASgX79+8PLyQkZGhtYDtLnlr8EXX3yBKVOmNHvthinmUNtxwRT3wQ5z5sPJyQmWlpZNKrPi4uImFVwDV1dXte2trKzg6Oiot7G2xYIFC/DDDz/g5MmT6N69u87bh4WFmeRfWOrY2dkhMDBQ43jNMX8AcPv2bRw/fhxvv/22ztuaU/4aZirpsk82bKfrNsZUW1uLCRMmQC6X49ixYzovUW5hYYH+/fubRV7d3Nzg5eWldazmlr8Gp0+fRl5eXqv2S2PnUNNxwZT3wQ5TfIhEIoSEhChnEDQ4duwYBgwYoHab8PDwJu2PHj2Kfv36QSgU6m2srUFEmD9/Pvbt24cTJ07Ax8enVf1IpVK4ubm18+j0o7q6GtevX9c4XnPK3/PS0tLg4uKCUaNG6bytOeXPx8cHrq6uKjmqqalBdna2xn0S0JxXbdsYS0PhkZ+fj+PHj7eq6CUi/PLLL2aR15KSEhQWFmodqznl73mpqakICQlBUFCQztsaK4fNHRdMeh9st0tXTcA333xDQqGQUlNT6dq1a7Rw4UKys7OjgoICIiJatmwZTZ06Vdn+1q1bZGtrS4sWLaJr165RamoqCYVC+u6774wVgkbvvvsuOTg4UFZWFhUVFSkfFRUVyjaN41u/fj3t37+fbty4QVeuXKFly5YRANq7d68xQmjW4sWLKSsri27dukXnzp2j0aNHk729fYfIX4P6+nry9PSkpUuXNnnNHPNXXl5OUqmUpFIpAaB169aRVCpVzvZITk4mBwcH2rdvH+Xm5tKkSZPIzc2NysrKlH1MnTpVZUbaTz/9RJaWlpScnEzXr1+n5ORksrKyonPnzplUfLW1tTRmzBjq3r07/fLLLyr7ZXV1tcb4VqxYQYcPHyaZTEZSqZRmzJhBVlZWdP78eZOKr7y8nBYvXkxnzpwhuVxOJ0+epPDwcOrWrZvZ5I+o+c8oEVFpaSnZ2trSZ599prYPU81hS44LproPdqjig4jo008/JS8vLxKJRNS3b1+Vqajx8fEUERGh0j4rK4uCg4NJJBKRt7e3xg+fsQFQ+0hLS1O2aRzfP/7xD+rZsydZW1tTly5daNCgQZSRkWH4wbfQxIkTyc3NjYRCIbm7u9O4cePo6tWrytfNOX8Njhw5QgAoLy+vyWvmmL+G6cCNH/Hx8UT0v6l+SUlJ5OrqSmKxmIYMGUK5ubkqfURERCjbN/j222/Jz8+PhEIh+fv7G63g0hafXC7XuF+ePHlS2Ufj+BYuXEienp4kEonI2dmZoqKi6MyZM4YPjrTHV1FRQVFRUeTs7ExCoZA8PT0pPj6e7ty5o9KHKeePqPnPKBHR559/TjY2NvTkyRO1fZhqDltyXDDVfVDw/wNgjDHGGDOIDnPNB2OMMcbMAxcfjDHGGDMoLj4YY4wxZlBcfDDGGGPMoLj4YIwxxphBcfHBGGOMMYPi4oMxxhhjBsXFB2OMMcYMiosPxhhjjBkUFx+MMcYYMyguPhhjjDFmUFx8MMYYY8yg/h+CDvRYYMrlRAAAAABJRU5ErkJggg==",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "demonstration(hyperdict)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "id": "3e468adb",
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "('image', 'average')\n",
      "\n",
      "Validation Set:\n",
      "Overall Dataset:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.6135    0.9901    0.7575     29306\n",
      "         1.0     0.8562    0.0863    0.1568     20010\n",
      "\n",
      "    accuracy                         0.6234     49316\n",
      "   macro avg     0.7348    0.5382    0.4572     49316\n",
      "weighted avg     0.7120    0.6234    0.5138     49316\n",
      "\n",
      "Training Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.6134    0.9921    0.7581     20463\n",
      "         1.0     0.8870    0.0899    0.1633     14058\n",
      "\n",
      "    accuracy                         0.6247     34521\n",
      "   macro avg     0.7502    0.5410    0.4607     34521\n",
      "weighted avg     0.7248    0.6247    0.5159     34521\n",
      "\n",
      "Validation Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.6110    0.9852    0.7542      4400\n",
      "         1.0     0.7848    0.0791    0.1437      2997\n",
      "\n",
      "    accuracy                         0.6181      7397\n",
      "   macro avg     0.6979    0.5322    0.4490      7397\n",
      "weighted avg     0.6814    0.6181    0.5069      7397\n",
      "\n",
      "Validation Set with both Images and Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8738    0.9194    0.8960       806\n",
      "         1.0     0.7848    0.6890    0.7337       344\n",
      "\n",
      "    accuracy                         0.8504      1150\n",
      "   macro avg     0.8293    0.8042    0.8149      1150\n",
      "weighted avg     0.8472    0.8504    0.8475      1150\n",
      "\n",
      "Validation Set with only Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.5753    1.0000    0.7304      3594\n",
      "         1.0     0.0000    0.0000    0.0000      2653\n",
      "\n",
      "    accuracy                         0.5753      6247\n",
      "   macro avg     0.2877    0.5000    0.3652      6247\n",
      "weighted avg     0.3310    0.5753    0.4202      6247\n",
      "\n",
      "Validation Set with unseen Images:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7867    0.8426    0.8137       324\n",
      "         1.0     0.6250    0.5346    0.5763       159\n",
      "\n",
      "    accuracy                         0.7412       483\n",
      "   macro avg     0.7059    0.6886    0.6950       483\n",
      "weighted avg     0.7335    0.7412    0.7355       483\n",
      "\n",
      "\n",
      "    Overall Accuracy: 62.34\n",
      "    Training Set Accuracy: 62.47\n",
      "    Validation Set Accuracy: 61.81\n",
      "    Validation Set with both Images and Texts: 85.04\n",
      "    Validation Set with only Texts: 57.53\n",
      "    Validation Set with unseen Images: 74.12\n",
      "    \n",
      "Test Set:\n",
      "Overall Dataset:\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\ProgramData\\Anaconda3\\Lib\\site-packages\\sklearn\\metrics\\_classification.py:1469: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "C:\\ProgramData\\Anaconda3\\Lib\\site-packages\\sklearn\\metrics\\_classification.py:1469: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "C:\\ProgramData\\Anaconda3\\Lib\\site-packages\\sklearn\\metrics\\_classification.py:1469: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "C:\\ProgramData\\Anaconda3\\Lib\\site-packages\\sklearn\\metrics\\_classification.py:1469: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "C:\\ProgramData\\Anaconda3\\Lib\\site-packages\\sklearn\\metrics\\_classification.py:1469: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "C:\\ProgramData\\Anaconda3\\Lib\\site-packages\\sklearn\\metrics\\_classification.py:1469: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.6135    0.9901    0.7575     29306\n",
      "         1.0     0.8562    0.0863    0.1568     20010\n",
      "\n",
      "    accuracy                         0.6234     49316\n",
      "   macro avg     0.7348    0.5382    0.4572     49316\n",
      "weighted avg     0.7120    0.6234    0.5138     49316\n",
      "\n",
      "Training Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.6134    0.9921    0.7581     20463\n",
      "         1.0     0.8870    0.0899    0.1633     14058\n",
      "\n",
      "    accuracy                         0.6247     34521\n",
      "   macro avg     0.7502    0.5410    0.4607     34521\n",
      "weighted avg     0.7248    0.6247    0.5159     34521\n",
      "\n",
      "Validation Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.6161    0.9856    0.7582      4443\n",
      "         1.0     0.7793    0.0765    0.1393      2955\n",
      "\n",
      "    accuracy                         0.6225      7398\n",
      "   macro avg     0.6977    0.5310    0.4487      7398\n",
      "weighted avg     0.6813    0.6225    0.5110      7398\n",
      "\n",
      "Validation Set with both Images and Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8424    0.9106    0.8752       716\n",
      "         1.0     0.7793    0.6494    0.7085       348\n",
      "\n",
      "    accuracy                         0.8252      1064\n",
      "   macro avg     0.8108    0.7800    0.7918      1064\n",
      "weighted avg     0.8218    0.8252    0.8206      1064\n",
      "\n",
      "Validation Set with only Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.5884    1.0000    0.7409      3727\n",
      "         1.0     0.0000    0.0000    0.0000      2607\n",
      "\n",
      "    accuracy                         0.5884      6334\n",
      "   macro avg     0.2942    0.5000    0.3704      6334\n",
      "weighted avg     0.3462    0.5884    0.4359      6334\n",
      "\n",
      "Validation Set with unseen Images:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7373    0.8351    0.7832       279\n",
      "         1.0     0.6102    0.4645    0.5275       155\n",
      "\n",
      "    accuracy                         0.7028       434\n",
      "   macro avg     0.6738    0.6498    0.6553       434\n",
      "weighted avg     0.6919    0.7028    0.6919       434\n",
      "\n",
      "\n",
      "    Overall Accuracy: 62.34\n",
      "    Training Set Accuracy: 62.47\n",
      "    Validation Set Accuracy: 62.25\n",
      "    Validation Set with both Images and Texts: 82.52\n",
      "    Validation Set with only Texts: 58.84\n",
      "    Validation Set with unseen Images: 70.28\n",
      "    \n",
      "Prediction for Golden Sentences:\n",
      "[[0.57319456 0.42680544]\n",
      " [0.57319456 0.42680544]]\n",
      "\n",
      "Full Dataset:\n",
      "count    49316.000000\n",
      "mean         0.613195\n",
      "std          0.110979\n",
      "min          0.500085\n",
      "25%          0.573195\n",
      "50%          0.573195\n",
      "75%          0.573195\n",
      "max          0.999913\n",
      "dtype: float64\n",
      "Training Set:\n",
      "count    34521.000000\n",
      "mean         0.613493\n",
      "std          0.111506\n",
      "min          0.500172\n",
      "25%          0.573195\n",
      "50%          0.573195\n",
      "75%          0.573195\n",
      "max          0.999913\n",
      "dtype: float64\n",
      "Validation Set:\n",
      "count    7397.000000\n",
      "mean        0.614568\n",
      "std         0.112293\n",
      "min         0.500085\n",
      "25%         0.573195\n",
      "50%         0.573195\n",
      "75%         0.573195\n",
      "max         0.999488\n",
      "dtype: float64\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\ProgramData\\Anaconda3\\Lib\\site-packages\\sklearn\\metrics\\_classification.py:1469: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "C:\\ProgramData\\Anaconda3\\Lib\\site-packages\\sklearn\\metrics\\_classification.py:1469: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "C:\\ProgramData\\Anaconda3\\Lib\\site-packages\\sklearn\\metrics\\_classification.py:1469: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "C:\\ProgramData\\Anaconda3\\Lib\\site-packages\\sklearn\\metrics\\_classification.py:1469: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "C:\\ProgramData\\Anaconda3\\Lib\\site-packages\\sklearn\\metrics\\_classification.py:1469: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "C:\\ProgramData\\Anaconda3\\Lib\\site-packages\\sklearn\\metrics\\_classification.py:1469: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\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": [
      "('multi', 'concat')\n",
      "\n",
      "Validation Set:\n",
      "Overall Dataset:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7998    0.7364    0.7668     29306\n",
      "         1.0     0.6541    0.7301    0.6900     20010\n",
      "\n",
      "    accuracy                         0.7338     49316\n",
      "   macro avg     0.7270    0.7332    0.7284     49316\n",
      "weighted avg     0.7407    0.7338    0.7356     49316\n",
      "\n",
      "Training Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8023    0.7410    0.7704     20463\n",
      "         1.0     0.6607    0.7342    0.6955     14058\n",
      "\n",
      "    accuracy                         0.7382     34521\n",
      "   macro avg     0.7315    0.7376    0.7330     34521\n",
      "weighted avg     0.7446    0.7382    0.7399     34521\n",
      "\n",
      "Validation Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7982    0.7327    0.7641      4400\n",
      "         1.0     0.6498    0.7281    0.6867      2997\n",
      "\n",
      "    accuracy                         0.7308      7397\n",
      "   macro avg     0.7240    0.7304    0.7254      7397\n",
      "weighted avg     0.7381    0.7308    0.7327      7397\n",
      "\n",
      "Validation Set with both Images and Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.9301    0.8747    0.9015       806\n",
      "         1.0     0.7423    0.8459    0.7908       344\n",
      "\n",
      "    accuracy                         0.8661      1150\n",
      "   macro avg     0.8362    0.8603    0.8461      1150\n",
      "weighted avg     0.8739    0.8661    0.8684      1150\n",
      "\n",
      "Validation Set with only Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7678    0.7009    0.7328      3594\n",
      "         1.0     0.6376    0.7128    0.6731      2653\n",
      "\n",
      "    accuracy                         0.7059      6247\n",
      "   macro avg     0.7027    0.7068    0.7029      6247\n",
      "weighted avg     0.7125    0.7059    0.7074      6247\n",
      "\n",
      "Validation Set with unseen Images:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8741    0.7932    0.8317       324\n",
      "         1.0     0.6455    0.7673    0.7011       159\n",
      "\n",
      "    accuracy                         0.7847       483\n",
      "   macro avg     0.7598    0.7803    0.7664       483\n",
      "weighted avg     0.7989    0.7847    0.7887       483\n",
      "\n",
      "\n",
      "    Overall Accuracy: 73.38\n",
      "    Training Set Accuracy: 73.82\n",
      "    Validation Set Accuracy: 73.08\n",
      "    Validation Set with both Images and Texts: 86.61\n",
      "    Validation Set with only Texts: 70.59\n",
      "    Validation Set with unseen Images: 78.47\n",
      "    \n",
      "Test Set:\n",
      "Overall Dataset:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7998    0.7364    0.7668     29306\n",
      "         1.0     0.6541    0.7301    0.6900     20010\n",
      "\n",
      "    accuracy                         0.7338     49316\n",
      "   macro avg     0.7270    0.7332    0.7284     49316\n",
      "weighted avg     0.7407    0.7338    0.7356     49316\n",
      "\n",
      "Training Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8023    0.7410    0.7704     20463\n",
      "         1.0     0.6607    0.7342    0.6955     14058\n",
      "\n",
      "    accuracy                         0.7382     34521\n",
      "   macro avg     0.7315    0.7376    0.7330     34521\n",
      "weighted avg     0.7446    0.7382    0.7399     34521\n",
      "\n",
      "Validation Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7900    0.7189    0.7528      4443\n",
      "         1.0     0.6277    0.7127    0.6675      2955\n",
      "\n",
      "    accuracy                         0.7164      7398\n",
      "   macro avg     0.7089    0.7158    0.7101      7398\n",
      "weighted avg     0.7252    0.7164    0.7187      7398\n",
      "\n",
      "Validation Set with both Images and Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8881    0.8869    0.8875       716\n",
      "         1.0     0.7679    0.7701    0.7690       348\n",
      "\n",
      "    accuracy                         0.8487      1064\n",
      "   macro avg     0.8280    0.8285    0.8283      1064\n",
      "weighted avg     0.8488    0.8487    0.8487      1064\n",
      "\n",
      "Validation Set with only Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7689    0.6866    0.7254      3727\n",
      "         1.0     0.6114    0.7050    0.6549      2607\n",
      "\n",
      "    accuracy                         0.6942      6334\n",
      "   macro avg     0.6902    0.6958    0.6902      6334\n",
      "weighted avg     0.7041    0.6942    0.6964      6334\n",
      "\n",
      "Validation Set with unseen Images:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8127    0.8244    0.8185       279\n",
      "         1.0     0.6755    0.6581    0.6667       155\n",
      "\n",
      "    accuracy                         0.7650       434\n",
      "   macro avg     0.7441    0.7412    0.7426       434\n",
      "weighted avg     0.7637    0.7650    0.7643       434\n",
      "\n",
      "\n",
      "    Overall Accuracy: 73.38\n",
      "    Training Set Accuracy: 73.82\n",
      "    Validation Set Accuracy: 71.64\n",
      "    Validation Set with both Images and Texts: 84.87\n",
      "    Validation Set with only Texts: 69.42\n",
      "    Validation Set with unseen Images: 76.50\n",
      "    \n",
      "Prediction for Golden Sentences:\n",
      "[[0.508907   0.49109295]\n",
      " [0.36659035 0.6334097 ]]\n",
      "\n",
      "Full Dataset:\n",
      "count    49316.000000\n",
      "mean         0.714303\n",
      "std          0.158927\n",
      "min          0.500006\n",
      "25%          0.574528\n",
      "50%          0.673523\n",
      "75%          0.855731\n",
      "max          0.999976\n",
      "dtype: float64\n",
      "Training Set:\n",
      "count    34521.000000\n",
      "mean         0.714782\n",
      "std          0.158895\n",
      "min          0.500006\n",
      "25%          0.575068\n",
      "50%          0.674363\n",
      "75%          0.856538\n",
      "max          0.999946\n",
      "dtype: float64\n",
      "Validation Set:\n",
      "count    7397.000000\n",
      "mean        0.713434\n",
      "std         0.158956\n",
      "min         0.500009\n",
      "25%         0.574206\n",
      "50%         0.671529\n",
      "75%         0.852156\n",
      "max         0.999942\n",
      "dtype: float64\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": [
      "('multi', 'average')\n",
      "\n",
      "Validation Set:\n",
      "Overall Dataset:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7716    0.7637    0.7677     29306\n",
      "         1.0     0.6591    0.6690    0.6640     20010\n",
      "\n",
      "    accuracy                         0.7253     49316\n",
      "   macro avg     0.7154    0.7163    0.7158     49316\n",
      "weighted avg     0.7260    0.7253    0.7256     49316\n",
      "\n",
      "Training Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7737    0.7678    0.7708     20463\n",
      "         1.0     0.6658    0.6731    0.6694     14058\n",
      "\n",
      "    accuracy                         0.7293     34521\n",
      "   macro avg     0.7197    0.7205    0.7201     34521\n",
      "weighted avg     0.7298    0.7293    0.7295     34521\n",
      "\n",
      "Validation Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7688    0.7570    0.7629      4400\n",
      "         1.0     0.6511    0.6657    0.6583      2997\n",
      "\n",
      "    accuracy                         0.7200      7397\n",
      "   macro avg     0.7099    0.7114    0.7106      7397\n",
      "weighted avg     0.7211    0.7200    0.7205      7397\n",
      "\n",
      "Validation Set with both Images and Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8912    0.8846    0.8879       806\n",
      "         1.0     0.7343    0.7471    0.7406       344\n",
      "\n",
      "    accuracy                         0.8435      1150\n",
      "   macro avg     0.8128    0.8159    0.8143      1150\n",
      "weighted avg     0.8443    0.8435    0.8439      1150\n",
      "\n",
      "Validation Set with only Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7410    0.7284    0.7347      3594\n",
      "         1.0     0.6404    0.6551    0.6477      2653\n",
      "\n",
      "    accuracy                         0.6973      6247\n",
      "   macro avg     0.6907    0.6918    0.6912      6247\n",
      "weighted avg     0.6983    0.6973    0.6977      6247\n",
      "\n",
      "Validation Set with unseen Images:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8381    0.8148    0.8263       324\n",
      "         1.0     0.6429    0.6792    0.6606       159\n",
      "\n",
      "    accuracy                         0.7702       483\n",
      "   macro avg     0.7405    0.7470    0.7434       483\n",
      "weighted avg     0.7738    0.7702    0.7717       483\n",
      "\n",
      "\n",
      "    Overall Accuracy: 72.53\n",
      "    Training Set Accuracy: 72.93\n",
      "    Validation Set Accuracy: 72.00\n",
      "    Validation Set with both Images and Texts: 84.35\n",
      "    Validation Set with only Texts: 69.73\n",
      "    Validation Set with unseen Images: 77.02\n",
      "    \n",
      "Test Set:\n",
      "Overall Dataset:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7716    0.7637    0.7677     29306\n",
      "         1.0     0.6591    0.6690    0.6640     20010\n",
      "\n",
      "    accuracy                         0.7253     49316\n",
      "   macro avg     0.7154    0.7163    0.7158     49316\n",
      "weighted avg     0.7260    0.7253    0.7256     49316\n",
      "\n",
      "Training Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7737    0.7678    0.7708     20463\n",
      "         1.0     0.6658    0.6731    0.6694     14058\n",
      "\n",
      "    accuracy                         0.7293     34521\n",
      "   macro avg     0.7197    0.7205    0.7201     34521\n",
      "weighted avg     0.7298    0.7293    0.7295     34521\n",
      "\n",
      "Validation Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7648    0.7515    0.7581      4443\n",
      "         1.0     0.6359    0.6525    0.6441      2955\n",
      "\n",
      "    accuracy                         0.7119      7398\n",
      "   macro avg     0.7003    0.7020    0.7011      7398\n",
      "weighted avg     0.7133    0.7119    0.7125      7398\n",
      "\n",
      "Validation Set with both Images and Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8648    0.8841    0.8743       716\n",
      "         1.0     0.7500    0.7155    0.7324       348\n",
      "\n",
      "    accuracy                         0.8289      1064\n",
      "   macro avg     0.8074    0.7998    0.8033      1064\n",
      "weighted avg     0.8272    0.8289    0.8279      1064\n",
      "\n",
      "Validation Set with only Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7446    0.7261    0.7352      3727\n",
      "         1.0     0.6219    0.6440    0.6327      2607\n",
      "\n",
      "    accuracy                         0.6923      6334\n",
      "   macro avg     0.6832    0.6850    0.6840      6334\n",
      "weighted avg     0.6941    0.6923    0.6930      6334\n",
      "\n",
      "Validation Set with unseen Images:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7891    0.8315    0.8098       279\n",
      "         1.0     0.6643    0.6000    0.6305       155\n",
      "\n",
      "    accuracy                         0.7488       434\n",
      "   macro avg     0.7267    0.7158    0.7201       434\n",
      "weighted avg     0.7445    0.7488    0.7458       434\n",
      "\n",
      "\n",
      "    Overall Accuracy: 72.53\n",
      "    Training Set Accuracy: 72.93\n",
      "    Validation Set Accuracy: 71.19\n",
      "    Validation Set with both Images and Texts: 82.89\n",
      "    Validation Set with only Texts: 69.23\n",
      "    Validation Set with unseen Images: 74.88\n",
      "    \n",
      "Prediction for Golden Sentences:\n",
      "[[0.497276   0.50272405]\n",
      " [0.3810212  0.61897874]]\n",
      "\n",
      "Full Dataset:\n",
      "count    49316.000000\n",
      "mean         0.707522\n",
      "std          0.157603\n",
      "min          0.500001\n",
      "25%          0.570560\n",
      "50%          0.663498\n",
      "75%          0.844147\n",
      "max          0.999985\n",
      "dtype: float64\n",
      "Training Set:\n",
      "count    34521.000000\n",
      "mean         0.707900\n",
      "std          0.157381\n",
      "min          0.500001\n",
      "25%          0.571040\n",
      "50%          0.664591\n",
      "75%          0.844388\n",
      "max          0.999981\n",
      "dtype: float64\n",
      "Validation Set:\n",
      "count    7397.000000\n",
      "mean        0.706777\n",
      "std         0.158184\n",
      "min         0.500037\n",
      "25%         0.569688\n",
      "50%         0.660311\n",
      "75%         0.841485\n",
      "max         0.999985\n",
      "dtype: float64\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAjEAAAGdCAYAAADjWSL8AAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjcuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/bCgiHAAAACXBIWXMAAA9hAAAPYQGoP6dpAAAmFElEQVR4nO3df1DU953H8dcKywIebESHX5WLpEONDfbqQUUwV51RITehNJNpTc+UJncmMWOqocR4el5aTK7Q0KtyhWii1ZgLWjuTlLvMxVDIXEPiYdRwcld/VO8aY/QCkqQIeDjLBj73R4Zvb0WNi+yuH3w+ZpxxP7z3u+/vmwVe81m+rMsYYwQAAGCZCZFuAAAAYDQIMQAAwEqEGAAAYCVCDAAAsBIhBgAAWIkQAwAArESIAQAAViLEAAAAK0VHuoFQGRoa0gcffKCEhAS5XK5ItwMAAK6CMUZ9fX1KT0/XhAlX3msZtyHmgw8+UEZGRqTbAAAAo3D69GlNnTr1ijXjNsQkJCRI+nQIiYmJY3psv9+vpqYmFRYWyu12j+mx8QfMOTyYc3gw5/Bh1uERqjn39vYqIyPD+Tl+JeM2xAy/hJSYmBiSEBMfH6/ExES+QEKIOYcHcw4P5hw+zDo8Qj3nq/lVEH6xFwAAWIkQAwAArESIAQAAViLEAAAAKxFiAACAlQgxAADASoQYAABgJUIMAACwEiEGAABYiRADAACsRIgBAABWIsQAAAArEWIAAICVCDEAAMBK0ZFuwGbZFb+Sb/Cz3yr8evHej+6MdAsAAIwZdmIAAICVCDEAAMBKhBgAAGAlQgwAALASIQYAAFiJEAMAAKxEiAEAAFYixAAAACsRYgAAgJUIMQAAwEqEGAAAYCVCDAAAsBIhBgAAWIkQAwAArESIAQAAViLEAAAAKxFiAACAlQgxAADASoQYAABgJUIMAACwEiEGAABYiRADAACsRIgBAABWIsQAAAArEWIAAICVCDEAAMBKhBgAAGAlQgwAALASIQYAAFiJEAMAAKxEiAEAAFYixAAAACsRYgAAgJWCCjGffPKJ/vZv/1aZmZmKi4vTLbfcoieffFJDQ0NOjTFGFRUVSk9PV1xcnObPn68jR44EHMfn82nFihWaMmWKJk6cqJKSEp05cyagpru7W6WlpfJ6vfJ6vSotLdW5c+dGf6YAAGBcCSrEPP3003r22WdVV1enY8eOqbq6Wj/+8Y9VW1vr1FRXV2vDhg2qq6vTwYMHlZqaqkWLFqmvr8+pKSsrU0NDg3bv3q29e/fq/PnzKi4u1uDgoFOzZMkStbe3q7GxUY2NjWpvb1dpaekYnDIAABgPooMp3rdvn77+9a/rzjvvlCRNmzZNP//5z/XOO+9I+nQXpqamRuvWrdPdd98tSXrhhReUkpKiXbt2admyZerp6dG2bdv04osvauHChZKk+vp6ZWRk6PXXX1dRUZGOHTumxsZGvf3228rLy5Mkbd26Vfn5+Tp+/LimT58+ZgMAAAB2CirE3H777Xr22Wd14sQJfeELX9B//Md/aO/evaqpqZEknTx5Up2dnSosLHTu4/F4NG/ePLW2tmrZsmVqa2uT3+8PqElPT1d2drZaW1tVVFSkffv2yev1OgFGkubMmSOv16vW1tZLhhifzyefz+fc7u3tlST5/X75/f5gTvMzDR/PM8GM6XFDbaznEGrD/drWt22Yc3gw5/Bh1uERqjkHc7ygQsxf//Vfq6enR7feequioqI0ODioH/7wh/qLv/gLSVJnZ6ckKSUlJeB+KSkpOnXqlFMTExOjSZMmjagZvn9nZ6eSk5NHPH5ycrJTc7GqqiqtX79+xHpTU5Pi4+ODOc2r9lTu0GcXXUf27NkT6RZGpbm5OdIt3BCYc3gw5/Bh1uEx1nPu7++/6tqgQswvfvEL1dfXa9euXbrtttvU3t6usrIypaen67777nPqXC5XwP2MMSPWLnZxzaXqr3SctWvXqry83Lnd29urjIwMFRYWKjEx8arO72r5/X41NzfriXcmyDd05fO6nhyuKIp0C0EZnvOiRYvkdrsj3c64xZzDgzmHD7MOj1DNefiVlKsRVIh5/PHHtWbNGn3rW9+SJM2cOVOnTp1SVVWV7rvvPqWmpkr6dCclLS3NuV9XV5ezO5OamqqBgQF1d3cH7MZ0dXWpoKDAqTl79uyIx//www9H7PIM83g88ng8I9bdbnfInsS+IZd8g/aEGFu/mEP5OcQfMOfwYM7hw6zDY6znHMyxgro6qb+/XxMmBN4lKirKucQ6MzNTqampAVtLAwMDamlpcQJKTk6O3G53QE1HR4cOHz7s1OTn56unp0cHDhxwavbv36+enh6nBgAA3NiC2on52te+ph/+8If64z/+Y9122206dOiQNmzYoL/6q7+S9OlLQGVlZaqsrFRWVpaysrJUWVmp+Ph4LVmyRJLk9Xq1dOlSPfbYY5o8ebKSkpK0atUqzZw507laacaMGbrjjjv04IMP6rnnnpMkPfTQQyouLubKpGswbc2rkW4hKJ4oo+rZke4CAHC9CirE1NbW6oknntDy5cvV1dWl9PR0LVu2TN///vedmtWrV+vChQtavny5uru7lZeXp6amJiUkJDg1GzduVHR0tBYvXqwLFy5owYIF2rFjh6KiopyanTt3auXKlc5VTCUlJaqrq7vW8wUAAONEUCEmISFBNTU1ziXVl+JyuVRRUaGKiorL1sTGxqq2tjbgj+RdLCkpSfX19cG0BwAAbiC8dxIAALASIQYAAFiJEAMAAKxEiAEAAFYixAAAACsRYgAAgJUIMQAAwEqEGAAAYCVCDAAAsBIhBgAAWIkQAwAArESIAQAAViLEAAAAKxFiAACAlQgxAADASoQYAABgJUIMAACwEiEGAABYiRADAACsRIgBAABWIsQAAAArEWIAAICVCDEAAMBKhBgAAGAlQgwAALASIQYAAFiJEAMAAKxEiAEAAFYixAAAACsRYgAAgJUIMQAAwEqEGAAAYCVCDAAAsBIhBgAAWIkQAwAArESIAQAAViLEAAAAKxFiAACAlQgxAADASoQYAABgJUIMAACwEiEGAABYiRADAACsRIgBAABWIsQAAAArEWIAAICVCDEAAMBKhBgAAGAlQgwAALASIQYAAFiJEAMAAKxEiAEAAFYixAAAACsRYgAAgJUIMQAAwEqEGAAAYCVCDAAAsFJ0pBsAPkt2xa/kG3RFuo2r9t6P7ox0CwBwQ2AnBgAAWIkQAwAArESIAQAAViLEAAAAKxFiAACAlQgxAADASoQYAABgJUIMAACwEiEGAABYKegQ8z//8z/69re/rcmTJys+Pl5f/vKX1dbW5nzcGKOKigqlp6crLi5O8+fP15EjRwKO4fP5tGLFCk2ZMkUTJ05USUmJzpw5E1DT3d2t0tJSeb1eeb1elZaW6ty5c6M7SwAAMO4EFWK6u7s1d+5cud1uvfbaazp69Kh+8pOf6KabbnJqqqurtWHDBtXV1engwYNKTU3VokWL1NfX59SUlZWpoaFBu3fv1t69e3X+/HkVFxdrcHDQqVmyZIna29vV2NioxsZGtbe3q7S09NrPGAAAjAtBvXfS008/rYyMDD3//PPO2rRp05z/G2NUU1OjdevW6e6775YkvfDCC0pJSdGuXbu0bNky9fT0aNu2bXrxxRe1cOFCSVJ9fb0yMjL0+uuvq6ioSMeOHVNjY6Pefvtt5eXlSZK2bt2q/Px8HT9+XNOnT7/W8wYAAJYLaifmlVdeUW5urr75zW8qOTlZs2bN0tatW52Pnzx5Up2dnSosLHTWPB6P5s2bp9bWVklSW1ub/H5/QE16erqys7Odmn379snr9ToBRpLmzJkjr9fr1AAAgBtbUDsx7777rjZv3qzy8nL9zd/8jQ4cOKCVK1fK4/HoO9/5jjo7OyVJKSkpAfdLSUnRqVOnJEmdnZ2KiYnRpEmTRtQM37+zs1PJyckjHj85OdmpuZjP55PP53Nu9/b2SpL8fr/8fn8wp/mZho/nmWDG9LgINDxf2+Y81s+3UBvu17a+bcOcw4dZh0eo5hzM8YIKMUNDQ8rNzVVlZaUkadasWTpy5Ig2b96s73znO06dy+UKuJ8xZsTaxS6uuVT9lY5TVVWl9evXj1hvampSfHz8FR97tJ7KHQrJcRHItjnv2bMn0i2MSnNzc6RbuCEw5/Bh1uEx1nPu7++/6tqgQkxaWpq++MUvBqzNmDFDL7/8siQpNTVV0qc7KWlpaU5NV1eXszuTmpqqgYEBdXd3B+zGdHV1qaCgwKk5e/bsiMf/8MMPR+zyDFu7dq3Ky8ud2729vcrIyFBhYaESExODOc3P5Pf71dzcrCfemSDf0JXDGUbPM8Hoqdwh6+Z8uKIo0i0EZfj5vGjRIrnd7ki3M24x5/Bh1uERqjkPv5JyNYIKMXPnztXx48cD1k6cOKGbb75ZkpSZmanU1FQ1Nzdr1qxZkqSBgQG1tLTo6aefliTl5OTI7XarublZixcvliR1dHTo8OHDqq6uliTl5+erp6dHBw4c0OzZsyVJ+/fvV09PjxN0LubxeOTxeEasu93ukD2JfUMu+Qbt+eFqK9vmbOs3zVB+reAPmHP4MOvwGOs5B3OsoELM9773PRUUFKiyslKLFy/WgQMHtGXLFm3ZskXSpy8BlZWVqbKyUllZWcrKylJlZaXi4+O1ZMkSSZLX69XSpUv12GOPafLkyUpKStKqVas0c+ZM52qlGTNm6I477tCDDz6o5557TpL00EMPqbi4mCuTAADj0rQ1r0a6haB4ooyqZ0e2h6BCzFe+8hU1NDRo7dq1evLJJ5WZmamamhrde++9Ts3q1at14cIFLV++XN3d3crLy1NTU5MSEhKcmo0bNyo6OlqLFy/WhQsXtGDBAu3YsUNRUVFOzc6dO7Vy5UrnKqaSkhLV1dVd6/kCAIBxIqgQI0nFxcUqLi6+7MddLpcqKipUUVFx2ZrY2FjV1taqtrb2sjVJSUmqr68Ptj0AAHCD4L2TAACAlQgxAADASoQYAABgJUIMAACwEiEGAABYiRADAACsRIgBAABWIsQAAAArEWIAAICVCDEAAMBKhBgAAGAlQgwAALASIQYAAFiJEAMAAKxEiAEAAFYixAAAACsRYgAAgJUIMQAAwEqEGAAAYCVCDAAAsBIhBgAAWIkQAwAArESIAQAAViLEAAAAKxFiAACAlQgxAADASoQYAABgJUIMAACwEiEGAABYiRADAACsRIgBAABWIsQAAAArEWIAAICVCDEAAMBKhBgAAGAlQgwAALASIQYAAFiJEAMAAKxEiAEAAFYixAAAACsRYgAAgJUIMQAAwEqEGAAAYCVCDAAAsBIhBgAAWIkQAwAArBQd6QaA8Wbamlcj3UJQPFFG1bMj3QUABI+dGAAAYCVCDAAAsBIhBgAAWIkQAwAArESIAQAAViLEAAAAKxFiAACAlQgxAADASoQYAABgJUIMAACwEiEGAABYiRADAACsRIgBAABWIsQAAAArEWIAAICVCDEAAMBKhBgAAGAlQgwAALASIQYAAFiJEAMAAKxEiAEAAFa6phBTVVUll8ulsrIyZ80Yo4qKCqWnpysuLk7z58/XkSNHAu7n8/m0YsUKTZkyRRMnTlRJSYnOnDkTUNPd3a3S0lJ5vV55vV6Vlpbq3Llz19IuAAAYR0YdYg4ePKgtW7boS1/6UsB6dXW1NmzYoLq6Oh08eFCpqalatGiR+vr6nJqysjI1NDRo9+7d2rt3r86fP6/i4mINDg46NUuWLFF7e7saGxvV2Nio9vZ2lZaWjrZdAAAwzowqxJw/f1733nuvtm7dqkmTJjnrxhjV1NRo3bp1uvvuu5Wdna0XXnhB/f392rVrlySpp6dH27Zt009+8hMtXLhQs2bNUn19vX7zm9/o9ddflyQdO3ZMjY2N+tnPfqb8/Hzl5+dr69at+pd/+RcdP358DE4bAADYLno0d3rkkUd05513auHChfq7v/s7Z/3kyZPq7OxUYWGhs+bxeDRv3jy1trZq2bJlamtrk9/vD6hJT09Xdna2WltbVVRUpH379snr9SovL8+pmTNnjrxer1pbWzV9+vQRPfl8Pvl8Pud2b2+vJMnv98vv94/mNC9r+HieCWZMj4tAw/NlzqE1PN+x/jpBoOH5MufQs3XWnii7vteF6ntHMMcLOsTs3r1b//7v/66DBw+O+FhnZ6ckKSUlJWA9JSVFp06dcmpiYmICdnCGa4bv39nZqeTk5BHHT05OdmouVlVVpfXr149Yb2pqUnx8/FWcWfCeyh0KyXERiDmHR3Nzc6RbuCEw5/CxbdbVsyPdweiM9Zz7+/uvujaoEHP69Gk9+uijampqUmxs7GXrXC5XwG1jzIi1i11cc6n6Kx1n7dq1Ki8vd2739vYqIyNDhYWFSkxMvOJjB8vv96u5uVlPvDNBvqErnxdGzzPB6KncIeYcYsNzXrRokdxud6TbGbeGv28w59CzddbZFb+KdAtBCdX3juFXUq5GUCGmra1NXV1dysnJcdYGBwf15ptvqq6uzvl9lc7OTqWlpTk1XV1dzu5MamqqBgYG1N3dHbAb09XVpYKCAqfm7NmzIx7/ww8/HLHLM8zj8cjj8YxYd7vdIXsS+4Zc8g3ywzXUmHN4hPJrBX/AnMPHtlnb+n1urOcczLGC+sXeBQsW6De/+Y3a29udf7m5ubr33nvV3t6uW265RampqQFbSwMDA2ppaXECSk5Ojtxud0BNR0eHDh8+7NTk5+erp6dHBw4ccGr279+vnp4epwYAANzYgtqJSUhIUHZ2dsDaxIkTNXnyZGe9rKxMlZWVysrKUlZWliorKxUfH68lS5ZIkrxer5YuXarHHntMkydPVlJSklatWqWZM2dq4cKFkqQZM2bojjvu0IMPPqjnnntOkvTQQw+puLj4kr/UCwAAbjyjujrpSlavXq0LFy5o+fLl6u7uVl5enpqampSQkODUbNy4UdHR0Vq8eLEuXLigBQsWaMeOHYqKinJqdu7cqZUrVzpXMZWUlKiurm6s2wUAAJa65hDzxhtvBNx2uVyqqKhQRUXFZe8TGxur2tpa1dbWXrYmKSlJ9fX119oeAAAYp3jvJAAAYCVCDAAAsBIhBgAAWIkQAwAArESIAQAAViLEAAAAKxFiAACAlcb8j90BsFN2xa+seu+W9350Z6RbABBh7MQAAAArEWIAAICVCDEAAMBKhBgAAGAlQgwAALASIQYAAFiJEAMAAKxEiAEAAFYixAAAACsRYgAAgJUIMQAAwEq8dxIAYFzJrviVqmfb935gCB47MQAAwEqEGAAAYCVCDAAAsBIhBgAAWIkQAwAArESIAQAAViLEAAAAKxFiAACAlQgxAADASoQYAABgJUIMAACwEu+dBMBK09a8GukWguKJMqqeHekugPGFnRgAAGAlQgwAALASIQYAAFiJEAMAAKxEiAEAAFYixAAAACsRYgAAgJUIMQAAwEqEGAAAYCX+Yi8A4LJs+8vIkuSJinQHCBd2YgAAgJUIMQAAwEq8nAQAYZRd8Sv5Bl2RbgMYF9iJAQAAViLEAAAAKxFiAACAlQgxAADASoQYAABgJUIMAACwEiEGAABYiRADAACsRIgBAABWIsQAAAArEWIAAICVCDEAAMBKhBgAAGAlQgwAALASIQYAAFiJEAMAAKxEiAEAAFYixAAAACsRYgAAgJUIMQAAwEqEGAAAYCVCDAAAsBIhBgAAWCmoEFNVVaWvfOUrSkhIUHJysu666y4dP348oMYYo4qKCqWnpysuLk7z58/XkSNHAmp8Pp9WrFihKVOmaOLEiSopKdGZM2cCarq7u1VaWiqv1yuv16vS0lKdO3dudGcJAADGnaBCTEtLix555BG9/fbbam5u1ieffKLCwkL97//+r1NTXV2tDRs2qK6uTgcPHlRqaqoWLVqkvr4+p6asrEwNDQ3avXu39u7dq/Pnz6u4uFiDg4NOzZIlS9Te3q7GxkY1Njaqvb1dpaWlY3DKAABgPIgOprixsTHg9vPPP6/k5GS1tbXpq1/9qowxqqmp0bp163T33XdLkl544QWlpKRo165dWrZsmXp6erRt2za9+OKLWrhwoSSpvr5eGRkZev3111VUVKRjx46psbFRb7/9tvLy8iRJW7duVX5+vo4fP67p06ePxbkDAACLBRViLtbT0yNJSkpKkiSdPHlSnZ2dKiwsdGo8Ho/mzZun1tZWLVu2TG1tbfL7/QE16enpys7OVmtrq4qKirRv3z55vV4nwEjSnDlz5PV61draeskQ4/P55PP5nNu9vb2SJL/fL7/ffy2nOcLw8TwTzJgeF4GG58ucQ4s5hwdzDh9mHR7D8w3Vz9irMeoQY4xReXm5br/9dmVnZ0uSOjs7JUkpKSkBtSkpKTp16pRTExMTo0mTJo2oGb5/Z2enkpOTRzxmcnKyU3OxqqoqrV+/fsR6U1OT4uPjgzy7q/NU7lBIjotAzDk8mHN4MOfwYdbh0dzcPKbH6+/vv+raUYeY7373u/rP//xP7d27d8THXC5XwG1jzIi1i11cc6n6Kx1n7dq1Ki8vd2739vYqIyNDhYWFSkxMvOJjB8vv96u5uVlPvDNBvqErnxdGzzPB6KncIeYcYsw5PJhz+DDr8Bie86JFi+R2u8fsuMOvpFyNUYWYFStW6JVXXtGbb76pqVOnOuupqamSPt1JSUtLc9a7urqc3ZnU1FQNDAyou7s7YDemq6tLBQUFTs3Zs2dHPO6HH344YpdnmMfjkcfjGbHudrvHdLj/n2/IJd8gXyChxpzDgzmHB3MOH2YdHmP9czaYYwV1dZIxRt/97nf1y1/+Uv/6r/+qzMzMgI9nZmYqNTU1YGtpYGBALS0tTkDJycmR2+0OqOno6NDhw4edmvz8fPX09OjAgQNOzf79+9XT0+PUAACAG1tQOzGPPPKIdu3apX/+539WQkKC8/spXq9XcXFxcrlcKisrU2VlpbKyspSVlaXKykrFx8dryZIlTu3SpUv12GOPafLkyUpKStKqVas0c+ZM52qlGTNm6I477tCDDz6o5557TpL00EMPqbi4mCuTAACApCBDzObNmyVJ8+fPD1h//vnndf/990uSVq9erQsXLmj58uXq7u5WXl6empqalJCQ4NRv3LhR0dHRWrx4sS5cuKAFCxZox44dioqKcmp27typlStXOlcxlZSUqK6ubjTnCAAAxqGgQowxn325msvlUkVFhSoqKi5bExsbq9raWtXW1l62JikpSfX19cG0BwAAbiC8dxIAALASIQYAAFiJEAMAAKxEiAEAAFYixAAAACsRYgAAgJUIMQAAwEqEGAAAYCVCDAAAsBIhBgAAWIkQAwAArESIAQAAViLEAAAAKxFiAACAlQgxAADASoQYAABgJUIMAACwEiEGAABYiRADAACsRIgBAABWIsQAAAArEWIAAICVCDEAAMBKhBgAAGAlQgwAALASIQYAAFiJEAMAAKxEiAEAAFYixAAAACsRYgAAgJUIMQAAwEqEGAAAYCVCDAAAsBIhBgAAWIkQAwAArESIAQAAViLEAAAAKxFiAACAlQgxAADASoQYAABgJUIMAACwEiEGAABYiRADAACsRIgBAABWIsQAAAArEWIAAICVCDEAAMBKhBgAAGAlQgwAALASIQYAAFiJEAMAAKxEiAEAAFYixAAAACsRYgAAgJUIMQAAwEqEGAAAYCVCDAAAsBIhBgAAWIkQAwAArESIAQAAViLEAAAAKxFiAACAlQgxAADASoQYAABgJUIMAACwEiEGAABYiRADAACsdN2HmE2bNikzM1OxsbHKycnRW2+9FemWAADAdeC6DjG/+MUvVFZWpnXr1unQoUP6sz/7M/35n/+53n///Ui3BgAAIuy6DjEbNmzQ0qVL9cADD2jGjBmqqalRRkaGNm/eHOnWAABAhEVHuoHLGRgYUFtbm9asWROwXlhYqNbW1hH1Pp9PPp/Pud3T0yNJ+v3vfy+/3z+mvfn9fvX39yvaP0GDQ64xPTb+IHrIqL9/iDmHGHMOD+YcPsw6PIbn/PHHH8vtdo/Zcfv6+iRJxpjP7mHMHnWMffTRRxocHFRKSkrAekpKijo7O0fUV1VVaf369SPWMzMzQ9YjQm9JpBu4QTDn8GDO4cOswyOUc+7r65PX671izXUbYoa5XIEp2hgzYk2S1q5dq/Lycuf20NCQfv/732vy5MmXrL8Wvb29ysjI0OnTp5WYmDimx8YfMOfwYM7hwZzDh1mHR6jmbIxRX1+f0tPTP7P2ug0xU6ZMUVRU1Ihdl66urhG7M5Lk8Xjk8XgC1m666aZQtqjExES+QMKAOYcHcw4P5hw+zDo8QjHnz9qBGXbd/mJvTEyMcnJy1NzcHLDe3NysgoKCCHUFAACuF9ftTowklZeXq7S0VLm5ucrPz9eWLVv0/vvv6+GHH450awAAIMKu6xBzzz336OOPP9aTTz6pjo4OZWdna8+ePbr55psj2pfH49EPfvCDES9fYWwx5/BgzuHBnMOHWYfH9TBnl7maa5gAAACuM9ft78QAAABcCSEGAABYiRADAACsRIgBAABWIsRcxqZNm5SZmanY2Fjl5OTorbfeumztG2+8IZfLNeLfb3/72zB2bKdg5ix9+h5Z69at08033yyPx6PPf/7z2r59e5i6tVcwc77//vsv+Xy+7bbbwtixnYJ9Pu/cuVN/8id/ovj4eKWlpekv//Iv9fHHH4epW3sFO+dnnnlGM2bMUFxcnKZPn65//Md/DFOn9nrzzTf1ta99Tenp6XK5XPqnf/qnz7xPS0uLcnJyFBsbq1tuuUXPPvts6Bs1GGH37t3G7XabrVu3mqNHj5pHH33UTJw40Zw6deqS9b/+9a+NJHP8+HHT0dHh/Pvkk0/C3Lldgp2zMcaUlJSYvLw809zcbE6ePGn2799v/u3f/i2MXdsn2DmfO3cu4Hl8+vRpk5SUZH7wgx+Et3HLBDvnt956y0yYMMH8wz/8g3n33XfNW2+9ZW677TZz1113hblzuwQ7502bNpmEhASze/du87vf/c78/Oc/N3/0R39kXnnllTB3bpc9e/aYdevWmZdfftlIMg0NDVesf/fdd018fLx59NFHzdGjR83WrVuN2+02L730Ukj7JMRcwuzZs83DDz8csHbrrbeaNWvWXLJ+OMR0d3eHobvxI9g5v/baa8br9ZqPP/44HO2NG8HO+WINDQ3G5XKZ9957LxTtjRvBzvnHP/6xueWWWwLWfvrTn5qpU6eGrMfxINg55+fnm1WrVgWsPfroo2bu3Lkh63G8uZoQs3r1anPrrbcGrC1btszMmTMnhJ0Zw8tJFxkYGFBbW5sKCwsD1gsLC9Xa2nrF+86aNUtpaWlasGCBfv3rX4eyTeuNZs6vvPKKcnNzVV1drc997nP6whe+oFWrVunChQvhaNlK1/J8HrZt2zYtXLgw4n9k8no2mjkXFBTozJkz2rNnj4wxOnv2rF566SXdeeed4WjZSqOZs8/nU2xsbMBaXFycDhw4IL/fH7JebzT79u0b8XkpKirSO++8E9I5E2Iu8tFHH2lwcHDEm0ympKSMeDPKYWlpadqyZYtefvll/fKXv9T06dO1YMECvfnmm+Fo2UqjmfO7776rvXv36vDhw2poaFBNTY1eeuklPfLII+Fo2UqjmfP/19HRoddee00PPPBAqFocF0Yz54KCAu3cuVP33HOPYmJilJqaqptuukm1tbXhaNlKo5lzUVGRfvazn6mtrU3GGL3zzjvavn27/H6/Pvroo3C0fUPo7Oy85Oflk08+Cemcr+u3HYgkl8sVcNsYM2Jt2PTp0zV9+nTndn5+vk6fPq2///u/11e/+tWQ9mm7YOY8NDQkl8ulnTt3Ou9wumHDBn3jG9/QM888o7i4uJD3a6tg5vz/7dixQzfddJPuuuuuEHU2vgQz56NHj2rlypX6/ve/r6KiInV0dOjxxx/Xww8/rG3btoWjXWsFM+cnnnhCnZ2dmjNnjowxSklJ0f3336/q6mpFRUWFo90bxqU+L5daH0vsxFxkypQpioqKGpHqu7q6RqTMK5kzZ47+67/+a6zbGzdGM+e0tDR97nOfC3iL9hkzZsgYozNnzoS0X1tdy/PZGKPt27ertLRUMTExoWzTeqOZc1VVlebOnavHH39cX/rSl1RUVKRNmzZp+/bt6ujoCEfb1hnNnOPi4rR9+3b19/frvffe0/vvv69p06YpISFBU6ZMCUfbN4TU1NRLfl6io6M1efLkkD0uIeYiMTExysnJUXNzc8B6c3OzCgoKrvo4hw4dUlpa2li3N26MZs5z587VBx98oPPnzztrJ06c0IQJEzR16tSQ9mura3k+t7S06L//+7+1dOnSULY4Loxmzv39/ZowIfBb8PDOgOEt7S7pWp7PbrdbU6dOVVRUlHbv3q3i4uIR88fo5efnj/i8NDU1KTc3V263O3QPHNJfG7bU8CV827ZtM0ePHjVlZWVm4sSJztUZa9asMaWlpU79xo0bTUNDgzlx4oQ5fPiwWbNmjZFkXn755UidghWCnXNfX5+ZOnWq+cY3vmGOHDliWlpaTFZWlnnggQcidQpWCHbOw7797W+bvLy8cLdrrWDn/Pzzz5vo6GizadMm87vf/c7s3bvX5ObmmtmzZ0fqFKwQ7JyPHz9uXnzxRXPixAmzf/9+c88995ikpCRz8uTJCJ2BHfr6+syhQ4fMoUOHjCSzYcMGc+jQIedS9ovnPHyJ9fe+9z1z9OhRs23bNi6xjqRnnnnG3HzzzSYmJsb86Z/+qWlpaXE+dt9995l58+Y5t59++mnz+c9/3sTGxppJkyaZ22+/3bz66qsR6No+wczZGGOOHTtmFi5caOLi4szUqVNNeXm56e/vD3PX9gl2zufOnTNxcXFmy5YtYe7UbsHO+ac//an54he/aOLi4kxaWpq59957zZkzZ8LctX2CmfPRo0fNl7/8ZRMXF2cSExPN17/+dfPb3/42Al3bZfhPh1z877777jPGXPr5/MYbb5hZs2aZmJgYM23aNLN58+aQ9+kyhn1LAABgH14QBAAAViLEAAAAKxFiAACAlQgxAADASoQYAABgJUIMAACwEiEGAABYiRADAACsRIgBAABWIsQAAAArEWIAAICVCDEAAMBK/wdOHI9AWcZKtQAAAABJRU5ErkJggg==",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "('multi', 'complex')\n",
      "\n",
      "Validation Set:\n",
      "Overall Dataset:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8454    0.7854    0.8143     29306\n",
      "         1.0     0.7153    0.7896    0.7506     20010\n",
      "\n",
      "    accuracy                         0.7871     49316\n",
      "   macro avg     0.7803    0.7875    0.7824     49316\n",
      "weighted avg     0.7926    0.7871    0.7884     49316\n",
      "\n",
      "Training Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8655    0.8038    0.8335     20463\n",
      "         1.0     0.7413    0.8182    0.7778     14058\n",
      "\n",
      "    accuracy                         0.8097     34521\n",
      "   macro avg     0.8034    0.8110    0.8057     34521\n",
      "weighted avg     0.8149    0.8097    0.8108     34521\n",
      "\n",
      "Validation Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8041    0.7511    0.7767      4400\n",
      "         1.0     0.6669    0.7314    0.6976      2997\n",
      "\n",
      "    accuracy                         0.7431      7397\n",
      "   macro avg     0.7355    0.7413    0.7372      7397\n",
      "weighted avg     0.7485    0.7431    0.7447      7397\n",
      "\n",
      "Validation Set with both Images and Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.9233    0.9256    0.9244       806\n",
      "         1.0     0.8246    0.8198    0.8222       344\n",
      "\n",
      "    accuracy                         0.8939      1150\n",
      "   macro avg     0.8739    0.8727    0.8733      1150\n",
      "weighted avg     0.8937    0.8939    0.8938      1150\n",
      "\n",
      "Validation Set with only Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7750    0.7120    0.7422      3594\n",
      "         1.0     0.6486    0.7199    0.6824      2653\n",
      "\n",
      "    accuracy                         0.7154      6247\n",
      "   macro avg     0.7118    0.7160    0.7123      6247\n",
      "weighted avg     0.7213    0.7154    0.7168      6247\n",
      "\n",
      "Validation Set with unseen Images:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8323    0.8272    0.8297       324\n",
      "         1.0     0.6522    0.6604    0.6563       159\n",
      "\n",
      "    accuracy                         0.7723       483\n",
      "   macro avg     0.7422    0.7438    0.7430       483\n",
      "weighted avg     0.7730    0.7723    0.7726       483\n",
      "\n",
      "\n",
      "    Overall Accuracy: 78.71\n",
      "    Training Set Accuracy: 80.97\n",
      "    Validation Set Accuracy: 74.31\n",
      "    Validation Set with both Images and Texts: 89.39\n",
      "    Validation Set with only Texts: 71.54\n",
      "    Validation Set with unseen Images: 77.23\n",
      "    \n",
      "Test Set:\n",
      "Overall Dataset:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8454    0.7854    0.8143     29306\n",
      "         1.0     0.7153    0.7896    0.7506     20010\n",
      "\n",
      "    accuracy                         0.7871     49316\n",
      "   macro avg     0.7803    0.7875    0.7824     49316\n",
      "weighted avg     0.7926    0.7871    0.7884     49316\n",
      "\n",
      "Training Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8655    0.8038    0.8335     20463\n",
      "         1.0     0.7413    0.8182    0.7778     14058\n",
      "\n",
      "    accuracy                         0.8097     34521\n",
      "   macro avg     0.8034    0.8110    0.8057     34521\n",
      "weighted avg     0.8149    0.8097    0.8108     34521\n",
      "\n",
      "Validation Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7936    0.7346    0.7630      4443\n",
      "         1.0     0.6411    0.7127    0.6750      2955\n",
      "\n",
      "    accuracy                         0.7259      7398\n",
      "   macro avg     0.7173    0.7237    0.7190      7398\n",
      "weighted avg     0.7327    0.7259    0.7278      7398\n",
      "\n",
      "Validation Set with both Images and Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8951    0.9176    0.9062       716\n",
      "         1.0     0.8212    0.7787    0.7994       348\n",
      "\n",
      "    accuracy                         0.8722      1064\n",
      "   macro avg     0.8582    0.8482    0.8528      1064\n",
      "weighted avg     0.8709    0.8722    0.8713      1064\n",
      "\n",
      "Validation Set with only Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7715    0.6995    0.7337      3727\n",
      "         1.0     0.6210    0.7039    0.6598      2607\n",
      "\n",
      "    accuracy                         0.7013      6334\n",
      "   macro avg     0.6963    0.7017    0.6968      6334\n",
      "weighted avg     0.7096    0.7013    0.7033      6334\n",
      "\n",
      "Validation Set with unseen Images:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7823    0.8244    0.8028       279\n",
      "         1.0     0.6500    0.5871    0.6169       155\n",
      "\n",
      "    accuracy                         0.7396       434\n",
      "   macro avg     0.7162    0.7057    0.7099       434\n",
      "weighted avg     0.7351    0.7396    0.7364       434\n",
      "\n",
      "\n",
      "    Overall Accuracy: 78.71\n",
      "    Training Set Accuracy: 80.97\n",
      "    Validation Set Accuracy: 72.59\n",
      "    Validation Set with both Images and Texts: 87.22\n",
      "    Validation Set with only Texts: 70.13\n",
      "    Validation Set with unseen Images: 73.96\n",
      "    \n",
      "Prediction for Golden Sentences:\n",
      "[[0.70748734 0.29251266]\n",
      " [0.5725778  0.42742226]]\n",
      "\n",
      "Full Dataset:\n",
      "count    49316.000000\n",
      "mean         0.775794\n",
      "std          0.170720\n",
      "min          0.500005\n",
      "25%          0.616368\n",
      "50%          0.770212\n",
      "75%          0.958147\n",
      "max          1.000000\n",
      "dtype: float64\n",
      "Training Set:\n",
      "count    34521.000000\n",
      "mean         0.778932\n",
      "std          0.170915\n",
      "min          0.500010\n",
      "25%          0.618910\n",
      "50%          0.775913\n",
      "75%          0.962189\n",
      "max          1.000000\n",
      "dtype: float64\n",
      "Validation Set:\n",
      "count    7397.000000\n",
      "mean        0.769461\n",
      "std         0.170158\n",
      "min         0.500080\n",
      "25%         0.611897\n",
      "50%         0.758349\n",
      "75%         0.948882\n",
      "max         1.000000\n",
      "dtype: float64\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": [
      "('text', 'average')\n",
      "\n",
      "Validation Set:\n",
      "Overall Dataset:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7881    0.7450    0.7659     29306\n",
      "         1.0     0.6542    0.7066    0.6794     20010\n",
      "\n",
      "    accuracy                         0.7294     49316\n",
      "   macro avg     0.7211    0.7258    0.7226     49316\n",
      "weighted avg     0.7338    0.7294    0.7308     49316\n",
      "\n",
      "Training Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7906    0.7495    0.7695     20463\n",
      "         1.0     0.6610    0.7111    0.6852     14058\n",
      "\n",
      "    accuracy                         0.7339     34521\n",
      "   macro avg     0.7258    0.7303    0.7273     34521\n",
      "weighted avg     0.7379    0.7339    0.7352     34521\n",
      "\n",
      "Validation Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7852    0.7384    0.7611      4400\n",
      "         1.0     0.6468    0.7034    0.6739      2997\n",
      "\n",
      "    accuracy                         0.7242      7397\n",
      "   macro avg     0.7160    0.7209    0.7175      7397\n",
      "weighted avg     0.7291    0.7242    0.7258      7397\n",
      "\n",
      "Validation Set with both Images and Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8685    0.8933    0.8807       806\n",
      "         1.0     0.7321    0.6831    0.7068       344\n",
      "\n",
      "    accuracy                         0.8304      1150\n",
      "   macro avg     0.8003    0.7882    0.7938      1150\n",
      "weighted avg     0.8277    0.8304    0.8287      1150\n",
      "\n",
      "Validation Set with only Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7643    0.7037    0.7327      3594\n",
      "         1.0     0.6375    0.7060    0.6700      2653\n",
      "\n",
      "    accuracy                         0.7047      6247\n",
      "   macro avg     0.7009    0.7048    0.7014      6247\n",
      "weighted avg     0.7104    0.7047    0.7061      6247\n",
      "\n",
      "Validation Set with unseen Images:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8239    0.8519    0.8376       324\n",
      "         1.0     0.6757    0.6289    0.6515       159\n",
      "\n",
      "    accuracy                         0.7785       483\n",
      "   macro avg     0.7498    0.7404    0.7445       483\n",
      "weighted avg     0.7751    0.7785    0.7763       483\n",
      "\n",
      "\n",
      "    Overall Accuracy: 72.94\n",
      "    Training Set Accuracy: 73.39\n",
      "    Validation Set Accuracy: 72.42\n",
      "    Validation Set with both Images and Texts: 83.04\n",
      "    Validation Set with only Texts: 70.47\n",
      "    Validation Set with unseen Images: 77.85\n",
      "    \n",
      "Test Set:\n",
      "Overall Dataset:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7881    0.7450    0.7659     29306\n",
      "         1.0     0.6542    0.7066    0.6794     20010\n",
      "\n",
      "    accuracy                         0.7294     49316\n",
      "   macro avg     0.7211    0.7258    0.7226     49316\n",
      "weighted avg     0.7338    0.7294    0.7308     49316\n",
      "\n",
      "Training Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7906    0.7495    0.7695     20463\n",
      "         1.0     0.6610    0.7111    0.6852     14058\n",
      "\n",
      "    accuracy                         0.7339     34521\n",
      "   macro avg     0.7258    0.7303    0.7273     34521\n",
      "weighted avg     0.7379    0.7339    0.7352     34521\n",
      "\n",
      "Validation Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7790    0.7306    0.7540      4443\n",
      "         1.0     0.6295    0.6883    0.6576      2955\n",
      "\n",
      "    accuracy                         0.7137      7398\n",
      "   macro avg     0.7043    0.7095    0.7058      7398\n",
      "weighted avg     0.7193    0.7137    0.7155      7398\n",
      "\n",
      "Validation Set with both Images and Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8570    0.8953    0.8757       716\n",
      "         1.0     0.7627    0.6925    0.7259       348\n",
      "\n",
      "    accuracy                         0.8289      1064\n",
      "   macro avg     0.8098    0.7939    0.8008      1064\n",
      "weighted avg     0.8261    0.8289    0.8267      1064\n",
      "\n",
      "Validation Set with only Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7619    0.6990    0.7291      3727\n",
      "         1.0     0.6151    0.6878    0.6494      2607\n",
      "\n",
      "    accuracy                         0.6943      6334\n",
      "   macro avg     0.6885    0.6934    0.6892      6334\n",
      "weighted avg     0.7015    0.6943    0.6963      6334\n",
      "\n",
      "Validation Set with unseen Images:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8103    0.8423    0.8260       279\n",
      "         1.0     0.6944    0.6452    0.6689       155\n",
      "\n",
      "    accuracy                         0.7719       434\n",
      "   macro avg     0.7524    0.7437    0.7475       434\n",
      "weighted avg     0.7690    0.7719    0.7699       434\n",
      "\n",
      "\n",
      "    Overall Accuracy: 72.94\n",
      "    Training Set Accuracy: 73.39\n",
      "    Validation Set Accuracy: 71.37\n",
      "    Validation Set with both Images and Texts: 82.89\n",
      "    Validation Set with only Texts: 69.43\n",
      "    Validation Set with unseen Images: 77.19\n",
      "    \n",
      "Prediction for Golden Sentences:\n",
      "[[0.4889501  0.51104987]\n",
      " [0.40325156 0.5967484 ]]\n",
      "\n",
      "Full Dataset:\n",
      "count    49316.000000\n",
      "mean         0.718178\n",
      "std          0.154859\n",
      "min          0.500006\n",
      "25%          0.583101\n",
      "50%          0.684278\n",
      "75%          0.852116\n",
      "max          0.999378\n",
      "dtype: float64\n",
      "Training Set:\n",
      "count    34521.000000\n",
      "mean         0.718607\n",
      "std          0.154648\n",
      "min          0.500006\n",
      "25%          0.583785\n",
      "50%          0.685241\n",
      "75%          0.852700\n",
      "max          0.999378\n",
      "dtype: float64\n",
      "Validation Set:\n",
      "count    7397.000000\n",
      "mean        0.716596\n",
      "std         0.155026\n",
      "min         0.500034\n",
      "25%         0.581351\n",
      "50%         0.682202\n",
      "75%         0.847366\n",
      "max         0.999214\n",
      "dtype: float64\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "perform_results['O1'] = statistics_F1(hyperdict, stored_data, df, labs, subfolder='001/')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "id": "8b67f85f",
   "metadata": {},
   "outputs": [],
   "source": [
    "with open(args.save_dir+'002/hyperdict.p', 'rb') as fp:\n",
    "    hyperdict = pickle.load(fp)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "id": "e3409605",
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train_acc\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": [
      "val_acc\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": [
      "val_img_txt\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": [
      "val_txt_only\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": [
      "val_unseen_img\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "demonstration(hyperdict)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "id": "6cc6c9d1",
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "('image', 'average')\n",
      "\n",
      "Validation Set:\n",
      "Overall Dataset:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.6139    0.9882    0.7573     29306\n",
      "         1.0     0.8379    0.0896    0.1619     20010\n",
      "\n",
      "    accuracy                         0.6236     49316\n",
      "   macro avg     0.7259    0.5389    0.4596     49316\n",
      "weighted avg     0.7047    0.6236    0.5157     49316\n",
      "\n",
      "Training Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.6138    0.9902    0.7579     20463\n",
      "         1.0     0.8676    0.0933    0.1684     14058\n",
      "\n",
      "    accuracy                         0.6250     34521\n",
      "   macro avg     0.7407    0.5417    0.4631     34521\n",
      "weighted avg     0.7172    0.6250    0.5178     34521\n",
      "\n",
      "Validation Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.6114    0.9836    0.7541      4400\n",
      "         1.0     0.7736    0.0821    0.1484      2997\n",
      "\n",
      "    accuracy                         0.6184      7397\n",
      "   macro avg     0.6925    0.5329    0.4512      7397\n",
      "weighted avg     0.6771    0.6184    0.5087      7397\n",
      "\n",
      "Validation Set with both Images and Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8822    0.9107    0.8962       806\n",
      "         1.0     0.7736    0.7151    0.7432       344\n",
      "\n",
      "    accuracy                         0.8522      1150\n",
      "   macro avg     0.8279    0.8129    0.8197      1150\n",
      "weighted avg     0.8497    0.8522    0.8504      1150\n",
      "\n",
      "Validation Set with only Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.5753    1.0000    0.7304      3594\n",
      "         1.0     0.0000    0.0000    0.0000      2653\n",
      "\n",
      "    accuracy                         0.5753      6247\n",
      "   macro avg     0.2877    0.5000    0.3652      6247\n",
      "weighted avg     0.3310    0.5753    0.4202      6247\n",
      "\n",
      "Validation Set with unseen Images:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8000    0.8272    0.8134       324\n",
      "         1.0     0.6216    0.5786    0.5993       159\n",
      "\n",
      "    accuracy                         0.7453       483\n",
      "   macro avg     0.7108    0.7029    0.7064       483\n",
      "weighted avg     0.7413    0.7453    0.7429       483\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\ProgramData\\Anaconda3\\Lib\\site-packages\\sklearn\\metrics\\_classification.py:1469: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "C:\\ProgramData\\Anaconda3\\Lib\\site-packages\\sklearn\\metrics\\_classification.py:1469: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "C:\\ProgramData\\Anaconda3\\Lib\\site-packages\\sklearn\\metrics\\_classification.py:1469: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "C:\\ProgramData\\Anaconda3\\Lib\\site-packages\\sklearn\\metrics\\_classification.py:1469: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "C:\\ProgramData\\Anaconda3\\Lib\\site-packages\\sklearn\\metrics\\_classification.py:1469: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "C:\\ProgramData\\Anaconda3\\Lib\\site-packages\\sklearn\\metrics\\_classification.py:1469: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "    Overall Accuracy: 62.36\n",
      "    Training Set Accuracy: 62.50\n",
      "    Validation Set Accuracy: 61.84\n",
      "    Validation Set with both Images and Texts: 85.22\n",
      "    Validation Set with only Texts: 57.53\n",
      "    Validation Set with unseen Images: 74.53\n",
      "    \n",
      "Test Set:\n",
      "Overall Dataset:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.6139    0.9882    0.7573     29306\n",
      "         1.0     0.8379    0.0896    0.1619     20010\n",
      "\n",
      "    accuracy                         0.6236     49316\n",
      "   macro avg     0.7259    0.5389    0.4596     49316\n",
      "weighted avg     0.7047    0.6236    0.5157     49316\n",
      "\n",
      "Training Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.6138    0.9902    0.7579     20463\n",
      "         1.0     0.8676    0.0933    0.1684     14058\n",
      "\n",
      "    accuracy                         0.6250     34521\n",
      "   macro avg     0.7407    0.5417    0.4631     34521\n",
      "weighted avg     0.7172    0.6250    0.5178     34521\n",
      "\n",
      "Validation Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.6163    0.9831    0.7577      4443\n",
      "         1.0     0.7588    0.0799    0.1445      2955\n",
      "\n",
      "    accuracy                         0.6223      7398\n",
      "   macro avg     0.6876    0.5315    0.4511      7398\n",
      "weighted avg     0.6733    0.6223    0.5128      7398\n",
      "\n",
      "Validation Set with both Images and Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8513    0.8953    0.8727       716\n",
      "         1.0     0.7588    0.6782    0.7162       348\n",
      "\n",
      "    accuracy                         0.8242      1064\n",
      "   macro avg     0.8051    0.7867    0.7945      1064\n",
      "weighted avg     0.8210    0.8242    0.8215      1064\n",
      "\n",
      "Validation Set with only Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.5884    1.0000    0.7409      3727\n",
      "         1.0     0.0000    0.0000    0.0000      2607\n",
      "\n",
      "    accuracy                         0.5884      6334\n",
      "   macro avg     0.2942    0.5000    0.3704      6334\n",
      "weighted avg     0.3462    0.5884    0.4359      6334\n",
      "\n",
      "Validation Set with unseen Images:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7433    0.7993    0.7703       279\n",
      "         1.0     0.5821    0.5032    0.5398       155\n",
      "\n",
      "    accuracy                         0.6935       434\n",
      "   macro avg     0.6627    0.6513    0.6550       434\n",
      "weighted avg     0.6857    0.6935    0.6880       434\n",
      "\n",
      "\n",
      "    Overall Accuracy: 62.36\n",
      "    Training Set Accuracy: 62.50\n",
      "    Validation Set Accuracy: 62.23\n",
      "    Validation Set with both Images and Texts: 82.42\n",
      "    Validation Set with only Texts: 58.84\n",
      "    Validation Set with unseen Images: 69.35\n",
      "    \n",
      "Prediction for Golden Sentences:\n",
      "[[0.574058 0.425942]\n",
      " [0.574058 0.425942]]\n",
      "\n",
      "Full Dataset:\n",
      "count    49316.000000\n",
      "mean         0.612865\n",
      "std          0.108499\n",
      "min          0.500252\n",
      "25%          0.574058\n",
      "50%          0.574058\n",
      "75%          0.574058\n",
      "max          0.999867\n",
      "dtype: float64\n",
      "Training Set:\n",
      "count    34521.000000\n",
      "mean         0.613178\n",
      "std          0.109085\n",
      "min          0.500713\n",
      "25%          0.574058\n",
      "50%          0.574058\n",
      "75%          0.574058\n",
      "max          0.999867\n",
      "dtype: float64\n",
      "Validation Set:\n",
      "count    7397.000000\n",
      "mean        0.614017\n",
      "std         0.109520\n",
      "min         0.500252\n",
      "25%         0.574058\n",
      "50%         0.574058\n",
      "75%         0.574058\n",
      "max         0.999391\n",
      "dtype: float64\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\ProgramData\\Anaconda3\\Lib\\site-packages\\sklearn\\metrics\\_classification.py:1469: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "C:\\ProgramData\\Anaconda3\\Lib\\site-packages\\sklearn\\metrics\\_classification.py:1469: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "C:\\ProgramData\\Anaconda3\\Lib\\site-packages\\sklearn\\metrics\\_classification.py:1469: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "C:\\ProgramData\\Anaconda3\\Lib\\site-packages\\sklearn\\metrics\\_classification.py:1469: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "C:\\ProgramData\\Anaconda3\\Lib\\site-packages\\sklearn\\metrics\\_classification.py:1469: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "C:\\ProgramData\\Anaconda3\\Lib\\site-packages\\sklearn\\metrics\\_classification.py:1469: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\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": [
      "('multi', 'concat')\n",
      "\n",
      "Validation Set:\n",
      "Overall Dataset:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8051    0.7296    0.7655     29306\n",
      "         1.0     0.6518    0.7413    0.6936     20010\n",
      "\n",
      "    accuracy                         0.7343     49316\n",
      "   macro avg     0.7284    0.7354    0.7296     49316\n",
      "weighted avg     0.7429    0.7343    0.7363     49316\n",
      "\n",
      "Training Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8076    0.7336    0.7688     20463\n",
      "         1.0     0.6578    0.7456    0.6990     14058\n",
      "\n",
      "    accuracy                         0.7385     34521\n",
      "   macro avg     0.7327    0.7396    0.7339     34521\n",
      "weighted avg     0.7466    0.7385    0.7404     34521\n",
      "\n",
      "Validation Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8032    0.7255    0.7624      4400\n",
      "         1.0     0.6471    0.7391    0.6900      2997\n",
      "\n",
      "    accuracy                         0.7310      7397\n",
      "   macro avg     0.7252    0.7323    0.7262      7397\n",
      "weighted avg     0.7400    0.7310    0.7331      7397\n",
      "\n",
      "Validation Set with both Images and Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.9314    0.8759    0.9028       806\n",
      "         1.0     0.7449    0.8488    0.7935       344\n",
      "\n",
      "    accuracy                         0.8678      1150\n",
      "   macro avg     0.8381    0.8624    0.8481      1150\n",
      "weighted avg     0.8756    0.8678    0.8701      1150\n",
      "\n",
      "Validation Set with only Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7730    0.6917    0.7301      3594\n",
      "         1.0     0.6344    0.7248    0.6766      2653\n",
      "\n",
      "    accuracy                         0.7058      6247\n",
      "   macro avg     0.7037    0.7083    0.7034      6247\n",
      "weighted avg     0.7142    0.7058    0.7074      6247\n",
      "\n",
      "Validation Set with unseen Images:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8801    0.7932    0.8344       324\n",
      "         1.0     0.6492    0.7799    0.7086       159\n",
      "\n",
      "    accuracy                         0.7888       483\n",
      "   macro avg     0.7647    0.7865    0.7715       483\n",
      "weighted avg     0.8041    0.7888    0.7930       483\n",
      "\n",
      "\n",
      "    Overall Accuracy: 73.43\n",
      "    Training Set Accuracy: 73.85\n",
      "    Validation Set Accuracy: 73.10\n",
      "    Validation Set with both Images and Texts: 86.78\n",
      "    Validation Set with only Texts: 70.58\n",
      "    Validation Set with unseen Images: 78.88\n",
      "    \n",
      "Test Set:\n",
      "Overall Dataset:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8051    0.7296    0.7655     29306\n",
      "         1.0     0.6518    0.7413    0.6936     20010\n",
      "\n",
      "    accuracy                         0.7343     49316\n",
      "   macro avg     0.7284    0.7354    0.7296     49316\n",
      "weighted avg     0.7429    0.7343    0.7363     49316\n",
      "\n",
      "Training Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8076    0.7336    0.7688     20463\n",
      "         1.0     0.6578    0.7456    0.6990     14058\n",
      "\n",
      "    accuracy                         0.7385     34521\n",
      "   macro avg     0.7327    0.7396    0.7339     34521\n",
      "weighted avg     0.7466    0.7385    0.7404     34521\n",
      "\n",
      "Validation Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7951    0.7153    0.7531      4443\n",
      "         1.0     0.6281    0.7228    0.6721      2955\n",
      "\n",
      "    accuracy                         0.7183      7398\n",
      "   macro avg     0.7116    0.7191    0.7126      7398\n",
      "weighted avg     0.7284    0.7183    0.7207      7398\n",
      "\n",
      "Validation Set with both Images and Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8908    0.8883    0.8895       716\n",
      "         1.0     0.7714    0.7759    0.7736       348\n",
      "\n",
      "    accuracy                         0.8515      1064\n",
      "   macro avg     0.8311    0.8321    0.8316      1064\n",
      "weighted avg     0.8517    0.8515    0.8516      1064\n",
      "\n",
      "Validation Set with only Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7743    0.6820    0.7252      3727\n",
      "         1.0     0.6116    0.7158    0.6596      2607\n",
      "\n",
      "    accuracy                         0.6959      6334\n",
      "   macro avg     0.6929    0.6989    0.6924      6334\n",
      "weighted avg     0.7073    0.6959    0.6982      6334\n",
      "\n",
      "Validation Set with unseen Images:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8149    0.8208    0.8179       279\n",
      "         1.0     0.6732    0.6645    0.6688       155\n",
      "\n",
      "    accuracy                         0.7650       434\n",
      "   macro avg     0.7441    0.7427    0.7433       434\n",
      "weighted avg     0.7643    0.7650    0.7646       434\n",
      "\n",
      "\n",
      "    Overall Accuracy: 73.43\n",
      "    Training Set Accuracy: 73.85\n",
      "    Validation Set Accuracy: 71.83\n",
      "    Validation Set with both Images and Texts: 85.15\n",
      "    Validation Set with only Texts: 69.59\n",
      "    Validation Set with unseen Images: 76.50\n",
      "    \n",
      "Prediction for Golden Sentences:\n",
      "[[0.5881782  0.41182184]\n",
      " [0.29911017 0.7008899 ]]\n",
      "\n",
      "Full Dataset:\n",
      "count    49316.000000\n",
      "mean         0.714921\n",
      "std          0.157838\n",
      "min          0.500003\n",
      "25%          0.577101\n",
      "50%          0.674591\n",
      "75%          0.854305\n",
      "max          0.999972\n",
      "dtype: float64\n",
      "Training Set:\n",
      "count    34521.000000\n",
      "mean         0.715390\n",
      "std          0.157815\n",
      "min          0.500008\n",
      "25%          0.577379\n",
      "50%          0.675346\n",
      "75%          0.855049\n",
      "max          0.999948\n",
      "dtype: float64\n",
      "Validation Set:\n",
      "count    7397.000000\n",
      "mean        0.714114\n",
      "std         0.157805\n",
      "min         0.500003\n",
      "25%         0.575797\n",
      "50%         0.673884\n",
      "75%         0.850201\n",
      "max         0.999929\n",
      "dtype: float64\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": [
      "('multi', 'average')\n",
      "\n",
      "Validation Set:\n",
      "Overall Dataset:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7728    0.7595    0.7661     29306\n",
      "         1.0     0.6565    0.6731    0.6647     20010\n",
      "\n",
      "    accuracy                         0.7244     49316\n",
      "   macro avg     0.7147    0.7163    0.7154     49316\n",
      "weighted avg     0.7256    0.7244    0.7250     49316\n",
      "\n",
      "Training Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7753    0.7638    0.7695     20463\n",
      "         1.0     0.6634    0.6777    0.6705     14058\n",
      "\n",
      "    accuracy                         0.7287     34521\n",
      "   macro avg     0.7193    0.7208    0.7200     34521\n",
      "weighted avg     0.7297    0.7287    0.7292     34521\n",
      "\n",
      "Validation Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7694    0.7532    0.7612      4400\n",
      "         1.0     0.6485    0.6687    0.6585      2997\n",
      "\n",
      "    accuracy                         0.7189      7397\n",
      "   macro avg     0.7090    0.7109    0.7098      7397\n",
      "weighted avg     0.7205    0.7189    0.7196      7397\n",
      "\n",
      "Validation Set with both Images and Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8911    0.8834    0.8872       806\n",
      "         1.0     0.7322    0.7471    0.7396       344\n",
      "\n",
      "    accuracy                         0.8426      1150\n",
      "   macro avg     0.8117    0.8152    0.8134      1150\n",
      "weighted avg     0.8436    0.8426    0.8431      1150\n",
      "\n",
      "Validation Set with only Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7417    0.7240    0.7328      3594\n",
      "         1.0     0.6378    0.6585    0.6480      2653\n",
      "\n",
      "    accuracy                         0.6962      6247\n",
      "   macro avg     0.6898    0.6912    0.6904      6247\n",
      "weighted avg     0.6976    0.6962    0.6968      6247\n",
      "\n",
      "Validation Set with unseen Images:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8429    0.8117    0.8270       324\n",
      "         1.0     0.6433    0.6918    0.6667       159\n",
      "\n",
      "    accuracy                         0.7723       483\n",
      "   macro avg     0.7431    0.7518    0.7469       483\n",
      "weighted avg     0.7772    0.7723    0.7742       483\n",
      "\n",
      "\n",
      "    Overall Accuracy: 72.44\n",
      "    Training Set Accuracy: 72.87\n",
      "    Validation Set Accuracy: 71.89\n",
      "    Validation Set with both Images and Texts: 84.26\n",
      "    Validation Set with only Texts: 69.62\n",
      "    Validation Set with unseen Images: 77.23\n",
      "    \n",
      "Test Set:\n",
      "Overall Dataset:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7728    0.7595    0.7661     29306\n",
      "         1.0     0.6565    0.6731    0.6647     20010\n",
      "\n",
      "    accuracy                         0.7244     49316\n",
      "   macro avg     0.7147    0.7163    0.7154     49316\n",
      "weighted avg     0.7256    0.7244    0.7250     49316\n",
      "\n",
      "Training Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7753    0.7638    0.7695     20463\n",
      "         1.0     0.6634    0.6777    0.6705     14058\n",
      "\n",
      "    accuracy                         0.7287     34521\n",
      "   macro avg     0.7193    0.7208    0.7200     34521\n",
      "weighted avg     0.7297    0.7287    0.7292     34521\n",
      "\n",
      "Validation Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7650    0.7459    0.7553      4443\n",
      "         1.0     0.6318    0.6555    0.6434      2955\n",
      "\n",
      "    accuracy                         0.7098      7398\n",
      "   macro avg     0.6984    0.7007    0.6994      7398\n",
      "weighted avg     0.7118    0.7098    0.7106      7398\n",
      "\n",
      "Validation Set with both Images and Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8632    0.8813    0.8721       716\n",
      "         1.0     0.7447    0.7126    0.7283       348\n",
      "\n",
      "    accuracy                         0.8261      1064\n",
      "   macro avg     0.8040    0.7970    0.8002      1064\n",
      "weighted avg     0.8245    0.8261    0.8251      1064\n",
      "\n",
      "Validation Set with only Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7451    0.7199    0.7323      3727\n",
      "         1.0     0.6180    0.6479    0.6326      2607\n",
      "\n",
      "    accuracy                         0.6902      6334\n",
      "   macro avg     0.6815    0.6839    0.6824      6334\n",
      "weighted avg     0.6928    0.6902    0.6912      6334\n",
      "\n",
      "Validation Set with unseen Images:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7831    0.8280    0.8049       279\n",
      "         1.0     0.6547    0.5871    0.6190       155\n",
      "\n",
      "    accuracy                         0.7419       434\n",
      "   macro avg     0.7189    0.7075    0.7120       434\n",
      "weighted avg     0.7372    0.7419    0.7385       434\n",
      "\n",
      "\n",
      "    Overall Accuracy: 72.44\n",
      "    Training Set Accuracy: 72.87\n",
      "    Validation Set Accuracy: 70.98\n",
      "    Validation Set with both Images and Texts: 82.61\n",
      "    Validation Set with only Texts: 69.02\n",
      "    Validation Set with unseen Images: 74.19\n",
      "    \n",
      "Prediction for Golden Sentences:\n",
      "[[0.5735477  0.4264523 ]\n",
      " [0.3379908  0.66200924]]\n",
      "\n",
      "Full Dataset:\n",
      "count    49316.000000\n",
      "mean         0.706688\n",
      "std          0.157617\n",
      "min          0.500000\n",
      "25%          0.569488\n",
      "50%          0.662532\n",
      "75%          0.843639\n",
      "max          0.999985\n",
      "dtype: float64\n",
      "Training Set:\n",
      "count    34521.000000\n",
      "mean         0.707055\n",
      "std          0.157420\n",
      "min          0.500003\n",
      "25%          0.570361\n",
      "50%          0.663670\n",
      "75%          0.843835\n",
      "max          0.999981\n",
      "dtype: float64\n",
      "Validation Set:\n",
      "count    7397.000000\n",
      "mean        0.706048\n",
      "std         0.158160\n",
      "min         0.500000\n",
      "25%         0.568214\n",
      "50%         0.659093\n",
      "75%         0.841878\n",
      "max         0.999985\n",
      "dtype: float64\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": [
      "('multi', 'complex')\n",
      "\n",
      "Validation Set:\n",
      "Overall Dataset:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8446    0.8008    0.8221     29306\n",
      "         1.0     0.7288    0.7842    0.7555     20010\n",
      "\n",
      "    accuracy                         0.7940     49316\n",
      "   macro avg     0.7867    0.7925    0.7888     49316\n",
      "weighted avg     0.7976    0.7940    0.7951     49316\n",
      "\n",
      "Training Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8668    0.8226    0.8441     20463\n",
      "         1.0     0.7596    0.8160    0.7868     14058\n",
      "\n",
      "    accuracy                         0.8199     34521\n",
      "   macro avg     0.8132    0.8193    0.8155     34521\n",
      "weighted avg     0.8232    0.8199    0.8208     34521\n",
      "\n",
      "Validation Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7989    0.7548    0.7762      4400\n",
      "         1.0     0.6670    0.7211    0.6930      2997\n",
      "\n",
      "    accuracy                         0.7411      7397\n",
      "   macro avg     0.7329    0.7379    0.7346      7397\n",
      "weighted avg     0.7454    0.7411    0.7425      7397\n",
      "\n",
      "Validation Set with both Images and Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.9303    0.9268    0.9285       806\n",
      "         1.0     0.8300    0.8372    0.8336       344\n",
      "\n",
      "    accuracy                         0.9000      1150\n",
      "   macro avg     0.8801    0.8820    0.8811      1150\n",
      "weighted avg     0.9003    0.9000    0.9001      1150\n",
      "\n",
      "Validation Set with only Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7674    0.7162    0.7409      3594\n",
      "         1.0     0.6474    0.7060    0.6754      2653\n",
      "\n",
      "    accuracy                         0.7119      6247\n",
      "   macro avg     0.7074    0.7111    0.7082      6247\n",
      "weighted avg     0.7165    0.7119    0.7131      6247\n",
      "\n",
      "Validation Set with unseen Images:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8454    0.8272    0.8362       324\n",
      "         1.0     0.6627    0.6918    0.6769       159\n",
      "\n",
      "    accuracy                         0.7826       483\n",
      "   macro avg     0.7540    0.7595    0.7566       483\n",
      "weighted avg     0.7853    0.7826    0.7838       483\n",
      "\n",
      "\n",
      "    Overall Accuracy: 79.40\n",
      "    Training Set Accuracy: 81.99\n",
      "    Validation Set Accuracy: 74.11\n",
      "    Validation Set with both Images and Texts: 90.00\n",
      "    Validation Set with only Texts: 71.19\n",
      "    Validation Set with unseen Images: 78.26\n",
      "    \n",
      "Test Set:\n",
      "Overall Dataset:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8446    0.8008    0.8221     29306\n",
      "         1.0     0.7288    0.7842    0.7555     20010\n",
      "\n",
      "    accuracy                         0.7940     49316\n",
      "   macro avg     0.7867    0.7925    0.7888     49316\n",
      "weighted avg     0.7976    0.7940    0.7951     49316\n",
      "\n",
      "Training Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8668    0.8226    0.8441     20463\n",
      "         1.0     0.7596    0.8160    0.7868     14058\n",
      "\n",
      "    accuracy                         0.8199     34521\n",
      "   macro avg     0.8132    0.8193    0.8155     34521\n",
      "weighted avg     0.8232    0.8199    0.8208     34521\n",
      "\n",
      "Validation Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7870    0.7459    0.7659      4443\n",
      "         1.0     0.6457    0.6964    0.6701      2955\n",
      "\n",
      "    accuracy                         0.7261      7398\n",
      "   macro avg     0.7164    0.7212    0.7180      7398\n",
      "weighted avg     0.7306    0.7261    0.7276      7398\n",
      "\n",
      "Validation Set with both Images and Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.9087    0.9176    0.9131       716\n",
      "         1.0     0.8270    0.8103    0.8186       348\n",
      "\n",
      "    accuracy                         0.8825      1064\n",
      "   macro avg     0.8678    0.8640    0.8659      1064\n",
      "weighted avg     0.8820    0.8825    0.8822      1064\n",
      "\n",
      "Validation Set with only Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7618    0.7129    0.7365      3727\n",
      "         1.0     0.6240    0.6812    0.6514      2607\n",
      "\n",
      "    accuracy                         0.6999      6334\n",
      "   macro avg     0.6929    0.6971    0.6940      6334\n",
      "weighted avg     0.7051    0.6999    0.7015      6334\n",
      "\n",
      "Validation Set with unseen Images:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7918    0.8315    0.8112       279\n",
      "         1.0     0.6667    0.6065    0.6351       155\n",
      "\n",
      "    accuracy                         0.7512       434\n",
      "   macro avg     0.7292    0.7190    0.7232       434\n",
      "weighted avg     0.7471    0.7512    0.7483       434\n",
      "\n",
      "\n",
      "    Overall Accuracy: 79.40\n",
      "    Training Set Accuracy: 81.99\n",
      "    Validation Set Accuracy: 72.61\n",
      "    Validation Set with both Images and Texts: 88.25\n",
      "    Validation Set with only Texts: 69.99\n",
      "    Validation Set with unseen Images: 75.12\n",
      "    \n",
      "Prediction for Golden Sentences:\n",
      "[[0.9462746  0.05372537]\n",
      " [0.12124777 0.8787522 ]]\n",
      "\n",
      "Full Dataset:\n",
      "count    49316.000000\n",
      "mean         0.788646\n",
      "std          0.171703\n",
      "min          0.500004\n",
      "25%          0.626610\n",
      "50%          0.794965\n",
      "75%          0.972980\n",
      "max          1.000000\n",
      "dtype: float64\n",
      "Training Set:\n",
      "count    34521.000000\n",
      "mean         0.791797\n",
      "std          0.171815\n",
      "min          0.500004\n",
      "25%          0.630058\n",
      "50%          0.801859\n",
      "75%          0.976593\n",
      "max          1.000000\n",
      "dtype: float64\n",
      "Validation Set:\n",
      "count    7397.000000\n",
      "mean        0.782243\n",
      "std         0.171122\n",
      "min         0.500013\n",
      "25%         0.620971\n",
      "50%         0.781853\n",
      "75%         0.965381\n",
      "max         1.000000\n",
      "dtype: float64\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": [
      "('text', 'average')\n",
      "\n",
      "Validation Set:\n",
      "Overall Dataset:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7860    0.7696    0.7777     29306\n",
      "         1.0     0.6726    0.6931    0.6827     20010\n",
      "\n",
      "    accuracy                         0.7386     49316\n",
      "   macro avg     0.7293    0.7314    0.7302     49316\n",
      "weighted avg     0.7400    0.7386    0.7392     49316\n",
      "\n",
      "Training Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7918    0.7766    0.7841     20463\n",
      "         1.0     0.6836    0.7028    0.6931     14058\n",
      "\n",
      "    accuracy                         0.7465     34521\n",
      "   macro avg     0.7377    0.7397    0.7386     34521\n",
      "weighted avg     0.7478    0.7465    0.7471     34521\n",
      "\n",
      "Validation Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7751    0.7575    0.7662      4400\n",
      "         1.0     0.6555    0.6773    0.6662      2997\n",
      "\n",
      "    accuracy                         0.7250      7397\n",
      "   macro avg     0.7153    0.7174    0.7162      7397\n",
      "weighted avg     0.7266    0.7250    0.7257      7397\n",
      "\n",
      "Validation Set with both Images and Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8705    0.9007    0.8854       806\n",
      "         1.0     0.7468    0.6860    0.7152       344\n",
      "\n",
      "    accuracy                         0.8365      1150\n",
      "   macro avg     0.8087    0.7934    0.8003      1150\n",
      "weighted avg     0.8335    0.8365    0.8344      1150\n",
      "\n",
      "Validation Set with only Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7522    0.7254    0.7385      3594\n",
      "         1.0     0.6451    0.6762    0.6603      2653\n",
      "\n",
      "    accuracy                         0.7045      6247\n",
      "   macro avg     0.6986    0.7008    0.6994      6247\n",
      "weighted avg     0.7067    0.7045    0.7053      6247\n",
      "\n",
      "Validation Set with unseen Images:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8269    0.8549    0.8407       324\n",
      "         1.0     0.6824    0.6352    0.6580       159\n",
      "\n",
      "    accuracy                         0.7826       483\n",
      "   macro avg     0.7546    0.7451    0.7493       483\n",
      "weighted avg     0.7793    0.7826    0.7805       483\n",
      "\n",
      "\n",
      "    Overall Accuracy: 73.86\n",
      "    Training Set Accuracy: 74.65\n",
      "    Validation Set Accuracy: 72.50\n",
      "    Validation Set with both Images and Texts: 83.65\n",
      "    Validation Set with only Texts: 70.45\n",
      "    Validation Set with unseen Images: 78.26\n",
      "    \n",
      "Test Set:\n",
      "Overall Dataset:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7860    0.7696    0.7777     29306\n",
      "         1.0     0.6726    0.6931    0.6827     20010\n",
      "\n",
      "    accuracy                         0.7386     49316\n",
      "   macro avg     0.7293    0.7314    0.7302     49316\n",
      "weighted avg     0.7400    0.7386    0.7392     49316\n",
      "\n",
      "Training Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7918    0.7766    0.7841     20463\n",
      "         1.0     0.6836    0.7028    0.6931     14058\n",
      "\n",
      "    accuracy                         0.7465     34521\n",
      "   macro avg     0.7377    0.7397    0.7386     34521\n",
      "weighted avg     0.7478    0.7465    0.7471     34521\n",
      "\n",
      "Validation Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7698    0.7495    0.7595      4443\n",
      "         1.0     0.6377    0.6629    0.6501      2955\n",
      "\n",
      "    accuracy                         0.7149      7398\n",
      "   macro avg     0.7037    0.7062    0.7048      7398\n",
      "weighted avg     0.7170    0.7149    0.7158      7398\n",
      "\n",
      "Validation Set with both Images and Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8571    0.8966    0.8765       716\n",
      "         1.0     0.7651    0.6925    0.7270       348\n",
      "\n",
      "    accuracy                         0.8299      1064\n",
      "   macro avg     0.8111    0.7946    0.8017      1064\n",
      "weighted avg     0.8270    0.8299    0.8276      1064\n",
      "\n",
      "Validation Set with only Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7515    0.7212    0.7360      3727\n",
      "         1.0     0.6231    0.6590    0.6406      2607\n",
      "\n",
      "    accuracy                         0.6956      6334\n",
      "   macro avg     0.6873    0.6901    0.6883      6334\n",
      "weighted avg     0.6986    0.6956    0.6967      6334\n",
      "\n",
      "Validation Set with unseen Images:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8069    0.8387    0.8225       279\n",
      "         1.0     0.6875    0.6387    0.6622       155\n",
      "\n",
      "    accuracy                         0.7673       434\n",
      "   macro avg     0.7472    0.7387    0.7424       434\n",
      "weighted avg     0.7643    0.7673    0.7652       434\n",
      "\n",
      "\n",
      "    Overall Accuracy: 73.86\n",
      "    Training Set Accuracy: 74.65\n",
      "    Validation Set Accuracy: 71.49\n",
      "    Validation Set with both Images and Texts: 82.99\n",
      "    Validation Set with only Texts: 69.56\n",
      "    Validation Set with unseen Images: 76.73\n",
      "    \n",
      "Prediction for Golden Sentences:\n",
      "[[0.697579   0.30242094]\n",
      " [0.32964227 0.67035776]]\n",
      "\n",
      "Full Dataset:\n",
      "count    49316.000000\n",
      "mean         0.720344\n",
      "std          0.156829\n",
      "min          0.500005\n",
      "25%          0.581471\n",
      "50%          0.688032\n",
      "75%          0.858864\n",
      "max          0.999555\n",
      "dtype: float64\n",
      "Training Set:\n",
      "count    34521.000000\n",
      "mean         0.720928\n",
      "std          0.156615\n",
      "min          0.500009\n",
      "25%          0.582197\n",
      "50%          0.689335\n",
      "75%          0.859285\n",
      "max          0.999508\n",
      "dtype: float64\n",
      "Validation Set:\n",
      "count    7397.000000\n",
      "mean        0.718287\n",
      "std         0.157240\n",
      "min         0.500005\n",
      "25%         0.578939\n",
      "50%         0.685660\n",
      "75%         0.856218\n",
      "max         0.999525\n",
      "dtype: float64\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "perform_results['O2'] = statistics_F1(hyperdict, stored_data, df, labs, subfolder='002/')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b9967be3",
   "metadata": {},
   "source": [
    "## Zero-shot Classification with Embeddings"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "id": "1a2ed3cc",
   "metadata": {},
   "outputs": [],
   "source": [
    "embeddings = {}\n",
    "for v in ['O1', 'O2']:\n",
    "    embeddings[v] = {}\n",
    "    for l in ['L2', 'L1', 'L0']:\n",
    "        embeddings[v][l] = {}\n",
    "        for t in ['val', 'test']:\n",
    "            embeddings[v][l][t] = {}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "id": "b022ae3d",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'O1': {'L2': {'val': {}, 'test': {}},\n",
       "  'L1': {'val': {}, 'test': {}},\n",
       "  'L0': {'val': {}, 'test': {}}},\n",
       " 'O2': {'L2': {'val': {}, 'test': {}},\n",
       "  'L1': {'val': {}, 'test': {}},\n",
       "  'L0': {'val': {}, 'test': {}}}}"
      ]
     },
     "execution_count": 52,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "embeddings"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7e44c55c",
   "metadata": {},
   "source": [
    "### Option 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "id": "dbb347b4",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Namespace(model_state_file='MLP_classifier.pth', save_dir='model_storage/CLIP/', embedding_size=512, early_stopping_criteria=20, num_epochs=20, seed=42, batch_size=32, embedding_type='text', fusion_type='average', hidden_dim=128, learning_rate=5e-05, l2=1e-05, dropout_p=0.1, catch_keyboard_interrupt=True, device='cuda', cuda=True, verbose=True)"
      ]
     },
     "execution_count": 53,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "args"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "id": "2afe1375",
   "metadata": {},
   "outputs": [],
   "source": [
    "args.embedding_type = 'multi'\n",
    "args.fusion_type = 'complex'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "id": "72541c20",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "MLPClassifier(\n",
       "  (dropout): Dropout(p=0.1, inplace=False)\n",
       "  (fc1): Linear(in_features=512, out_features=128, bias=True)\n",
       "  (fc2): Linear(in_features=128, out_features=2, bias=True)\n",
       "  (fc_concat): Linear(in_features=1024, out_features=128, bias=True)\n",
       "  (fc_complex): Linear(in_features=2560, out_features=1024, bias=True)\n",
       ")"
      ]
     },
     "execution_count": 55,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "classifier = MLPClassifier(embedding_size=args.embedding_size, \n",
    "                        hidden_dim=args.hidden_dim, \n",
    "                        embedding_type=args.embedding_type, \n",
    "                        fusion_type=args.fusion_type,\n",
    "                        num_classes=len(labels), \n",
    "                        dropout_p=args.dropout_p).to(args.device)\n",
    "classifier.load_state_dict(torch.load(args.save_dir + '001/' + args.embedding_type + \"_\" + args.fusion_type + \"_\" + args.model_state_file, map_location=torch.device('cpu')))\n",
    "classifier.eval()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "id": "3961b71e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "torch.Size([49316, 128])"
      ]
     },
     "execution_count": 56,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "new_emb = F.relu(classifier.dropout(classifier.fc_concat(F.relu(classifier.dropout(classifier.fc_complex(prepare_embedding(stored_data, args.embedding_type, args.fusion_type)))))))\n",
    "new_emb.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "id": "2fc1bfcd",
   "metadata": {},
   "outputs": [],
   "source": [
    "labs = {}\n",
    "labs['txt_emb'] = stored_data['lab_emb']\n",
    "empty_img_emb = stored_data['empty_img_emb']\n",
    "labs['img_emb'] = torch.vstack([empty_img_emb] * labs['txt_emb'].shape[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "id": "6bf457bb",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "torch.Size([2, 512])"
      ]
     },
     "execution_count": 58,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "labs['img_emb'].shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "id": "e01ffae8",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "torch.Size([2, 128])"
      ]
     },
     "execution_count": 59,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "new_lab_emb = F.relu(classifier.dropout(classifier.fc_concat(F.relu(classifier.dropout(classifier.fc_complex(prepare_embedding(labs, args.embedding_type, args.fusion_type)))))))\n",
    "new_lab_emb.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "id": "3583acd6",
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Zero-shot using new multi-modal embedding:\n",
      "Overall Dataset:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7579    0.7920    0.7745     29306\n",
      "         1.0     0.6738    0.6294    0.6509     20010\n",
      "\n",
      "    accuracy                         0.7260     49316\n",
      "   macro avg     0.7158    0.7107    0.7127     49316\n",
      "weighted avg     0.7238    0.7260    0.7244     49316\n",
      "\n",
      "Training Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7682    0.7986    0.7831     20463\n",
      "         1.0     0.6889    0.6493    0.6685     14058\n",
      "\n",
      "    accuracy                         0.7378     34521\n",
      "   macro avg     0.7286    0.7239    0.7258     34521\n",
      "weighted avg     0.7359    0.7378    0.7364     34521\n",
      "\n",
      "Validation Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7418    0.7827    0.7617      4400\n",
      "         1.0     0.6529    0.5999    0.6253      2997\n",
      "\n",
      "    accuracy                         0.7087      7397\n",
      "   macro avg     0.6973    0.6913    0.6935      7397\n",
      "weighted avg     0.7057    0.7087    0.7064      7397\n",
      "\n",
      "Validation Set with both Images and Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8185    0.9789    0.8915       806\n",
      "         1.0     0.9086    0.4913    0.6377       344\n",
      "\n",
      "    accuracy                         0.8330      1150\n",
      "   macro avg     0.8635    0.7351    0.7646      1150\n",
      "weighted avg     0.8454    0.8330    0.8156      1150\n",
      "\n",
      "Validation Set with only Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7217    0.7387    0.7301      3594\n",
      "         1.0     0.6343    0.6140    0.6240      2653\n",
      "\n",
      "    accuracy                         0.6858      6247\n",
      "   macro avg     0.6780    0.6764    0.6771      6247\n",
      "weighted avg     0.6846    0.6858    0.6850      6247\n",
      "\n",
      "Validation Set with unseen Images:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7506    0.9475    0.8377       324\n",
      "         1.0     0.7703    0.3585    0.4893       159\n",
      "\n",
      "    accuracy                         0.7536       483\n",
      "   macro avg     0.7604    0.6530    0.6635       483\n",
      "weighted avg     0.7571    0.7536    0.7230       483\n",
      "\n",
      "\n",
      "    Overall Accuracy: 72.60\n",
      "    Training Set Accuracy: 73.78\n",
      "    Validation Set Accuracy: 70.87\n",
      "    Validation Set with both Images and Texts: 83.30\n",
      "    Validation Set with only Texts: 68.58\n",
      "    Validation Set with unseen Images: 75.36\n",
      "    \n",
      "Zero-shot using new multi-modal embedding (test):\n",
      "Overall Dataset:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7579    0.7920    0.7745     29306\n",
      "         1.0     0.6738    0.6294    0.6509     20010\n",
      "\n",
      "    accuracy                         0.7260     49316\n",
      "   macro avg     0.7158    0.7107    0.7127     49316\n",
      "weighted avg     0.7238    0.7260    0.7244     49316\n",
      "\n",
      "Training Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7682    0.7986    0.7831     20463\n",
      "         1.0     0.6889    0.6493    0.6685     14058\n",
      "\n",
      "    accuracy                         0.7378     34521\n",
      "   macro avg     0.7286    0.7239    0.7258     34521\n",
      "weighted avg     0.7359    0.7378    0.7364     34521\n",
      "\n",
      "Validation Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7270    0.7707    0.7482      4443\n",
      "         1.0     0.6209    0.5648    0.5915      2955\n",
      "\n",
      "    accuracy                         0.6884      7398\n",
      "   macro avg     0.6739    0.6677    0.6698      7398\n",
      "weighted avg     0.6846    0.6884    0.6856      7398\n",
      "\n",
      "Validation Set with both Images and Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7819    0.9763    0.8683       716\n",
      "         1.0     0.9000    0.4397    0.5907       348\n",
      "\n",
      "    accuracy                         0.8008      1064\n",
      "   macro avg     0.8409    0.7080    0.7295      1064\n",
      "weighted avg     0.8205    0.8008    0.7775      1064\n",
      "\n",
      "Validation Set with only Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7141    0.7312    0.7225      3727\n",
      "         1.0     0.6021    0.5815    0.5916      2607\n",
      "\n",
      "    accuracy                         0.6696      6334\n",
      "   macro avg     0.6581    0.6563    0.6571      6334\n",
      "weighted avg     0.6680    0.6696    0.6686      6334\n",
      "\n",
      "Validation Set with unseen Images:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.6939    0.9427    0.7994       279\n",
      "         1.0     0.7091    0.2516    0.3714       155\n",
      "\n",
      "    accuracy                         0.6959       434\n",
      "   macro avg     0.7015    0.5971    0.5854       434\n",
      "weighted avg     0.6993    0.6959    0.6465       434\n",
      "\n",
      "\n",
      "    Overall Accuracy: 72.60\n",
      "    Training Set Accuracy: 73.78\n",
      "    Validation Set Accuracy: 68.84\n",
      "    Validation Set with both Images and Texts: 80.08\n",
      "    Validation Set with only Texts: 66.96\n",
      "    Validation Set with unseen Images: 69.59\n",
      "    \n"
     ]
    }
   ],
   "source": [
    "MUL_LAB = util.cos_sim(new_lab_emb, new_emb)\n",
    "pred_MUL = torch.argmax(MUL_LAB[:2,:], dim=0).float()\n",
    "lab = torch.tensor(np.array(df['skepticism'])).to('cuda').float()\n",
    "print('Zero-shot using new multi-modal embedding:')\n",
    "embeddings['O1']['L2']['val'] = evaluate_dataset_F1(df, pred_MUL)\n",
    "print('Zero-shot using new multi-modal embedding (test):')\n",
    "embeddings['O1']['L2']['test'] = evaluate_dataset_F1(df, pred_MUL, val_set='test')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "id": "88e2ed53",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "torch.Size([49316, 1024])"
      ]
     },
     "execution_count": 61,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "new_emb = F.relu(classifier.dropout(classifier.fc_complex(prepare_embedding(stored_data, args.embedding_type, args.fusion_type))))\n",
    "new_emb.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "id": "109bc18d",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "torch.Size([2, 1024])"
      ]
     },
     "execution_count": 62,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "new_lab_emb = F.relu(classifier.dropout(classifier.fc_complex(prepare_embedding(labs, args.embedding_type, args.fusion_type))))\n",
    "new_lab_emb.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "id": "b3d4a62d",
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Zero-shot using new multi-modal embedding:\n",
      "Overall Dataset:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.6283    0.8658    0.7282     29306\n",
      "         1.0     0.5598    0.2500    0.3457     20010\n",
      "\n",
      "    accuracy                         0.6159     49316\n",
      "   macro avg     0.5941    0.5579    0.5369     49316\n",
      "weighted avg     0.6005    0.6159    0.5730     49316\n",
      "\n",
      "Training Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.6253    0.8635    0.7253     20463\n",
      "         1.0     0.5539    0.2468    0.3414     14058\n",
      "\n",
      "    accuracy                         0.6123     34521\n",
      "   macro avg     0.5896    0.5551    0.5334     34521\n",
      "weighted avg     0.5962    0.6123    0.5690     34521\n",
      "\n",
      "Validation Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.6362    0.8757    0.7370      4400\n",
      "         1.0     0.5921    0.2649    0.3661      2997\n",
      "\n",
      "    accuracy                         0.6282      7397\n",
      "   macro avg     0.6142    0.5703    0.5515      7397\n",
      "weighted avg     0.6183    0.6282    0.5867      7397\n",
      "\n",
      "Validation Set with both Images and Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7157    0.9901    0.8308       806\n",
      "         1.0     0.7714    0.0785    0.1425       344\n",
      "\n",
      "    accuracy                         0.7174      1150\n",
      "   macro avg     0.7436    0.5343    0.4866      1150\n",
      "weighted avg     0.7324    0.7174    0.6249      1150\n",
      "\n",
      "Validation Set with only Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.6183    0.8500    0.7159      3594\n",
      "         1.0     0.5873    0.2891    0.3875      2653\n",
      "\n",
      "    accuracy                         0.6118      6247\n",
      "   macro avg     0.6028    0.5696    0.5517      6247\n",
      "weighted avg     0.6051    0.6118    0.5764      6247\n",
      "\n",
      "Validation Set with unseen Images:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.6885    0.9753    0.8072       324\n",
      "         1.0     0.6667    0.1006    0.1749       159\n",
      "\n",
      "    accuracy                         0.6874       483\n",
      "   macro avg     0.6776    0.5380    0.4910       483\n",
      "weighted avg     0.6813    0.6874    0.5990       483\n",
      "\n",
      "\n",
      "    Overall Accuracy: 61.59\n",
      "    Training Set Accuracy: 61.23\n",
      "    Validation Set Accuracy: 62.82\n",
      "    Validation Set with both Images and Texts: 71.74\n",
      "    Validation Set with only Texts: 61.18\n",
      "    Validation Set with unseen Images: 68.74\n",
      "    \n",
      "Zero-shot using new multi-modal embedding (test):\n",
      "Overall Dataset:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.6283    0.8658    0.7282     29306\n",
      "         1.0     0.5598    0.2500    0.3457     20010\n",
      "\n",
      "    accuracy                         0.6159     49316\n",
      "   macro avg     0.5941    0.5579    0.5369     49316\n",
      "weighted avg     0.6005    0.6159    0.5730     49316\n",
      "\n",
      "Training Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.6253    0.8635    0.7253     20463\n",
      "         1.0     0.5539    0.2468    0.3414     14058\n",
      "\n",
      "    accuracy                         0.6123     34521\n",
      "   macro avg     0.5896    0.5551    0.5334     34521\n",
      "weighted avg     0.5962    0.6123    0.5690     34521\n",
      "\n",
      "Validation Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.6348    0.8665    0.7328      4443\n",
      "         1.0     0.5551    0.2504    0.3451      2955\n",
      "\n",
      "    accuracy                         0.6204      7398\n",
      "   macro avg     0.5950    0.5585    0.5390      7398\n",
      "weighted avg     0.6030    0.6204    0.5779      7398\n",
      "\n",
      "Validation Set with both Images and Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.6864    0.9902    0.8107       716\n",
      "         1.0     0.7742    0.0690    0.1266       348\n",
      "\n",
      "    accuracy                         0.6889      1064\n",
      "   macro avg     0.7303    0.5296    0.4687      1064\n",
      "weighted avg     0.7151    0.6889    0.5870      1064\n",
      "\n",
      "Validation Set with only Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.6242    0.8428    0.7172      3727\n",
      "         1.0     0.5499    0.2746    0.3663      2607\n",
      "\n",
      "    accuracy                         0.6089      6334\n",
      "   macro avg     0.5871    0.5587    0.5418      6334\n",
      "weighted avg     0.5936    0.6089    0.5728      6334\n",
      "\n",
      "Validation Set with unseen Images:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.6643    0.9857    0.7937       279\n",
      "         1.0     0.8000    0.1032    0.1829       155\n",
      "\n",
      "    accuracy                         0.6705       434\n",
      "   macro avg     0.7321    0.5444    0.4883       434\n",
      "weighted avg     0.7127    0.6705    0.5755       434\n",
      "\n",
      "\n",
      "    Overall Accuracy: 61.59\n",
      "    Training Set Accuracy: 61.23\n",
      "    Validation Set Accuracy: 62.04\n",
      "    Validation Set with both Images and Texts: 68.89\n",
      "    Validation Set with only Texts: 60.89\n",
      "    Validation Set with unseen Images: 67.05\n",
      "    \n"
     ]
    }
   ],
   "source": [
    "MUL_LAB = util.cos_sim(new_lab_emb, new_emb)\n",
    "pred_MUL = torch.argmax(MUL_LAB[:2,:], dim=0).float()\n",
    "lab = torch.tensor(np.array(df['skepticism'])).to('cuda').float()\n",
    "print('Zero-shot using new multi-modal embedding:')\n",
    "embeddings['O1']['L1']['val'] = evaluate_dataset_F1(df, pred_MUL)\n",
    "print('Zero-shot using new multi-modal embedding (test):')\n",
    "embeddings['O1']['L1']['test'] = evaluate_dataset_F1(df, pred_MUL, val_set='test')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "id": "f491e654",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "torch.Size([49316, 2560])"
      ]
     },
     "execution_count": 64,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "new_emb = prepare_embedding(stored_data, args.embedding_type, args.fusion_type)\n",
    "new_emb.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "id": "a727b55c",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "torch.Size([2, 2560])"
      ]
     },
     "execution_count": 65,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "new_lab_emb = prepare_embedding(labs, args.embedding_type, args.fusion_type)\n",
    "new_lab_emb.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "id": "ddbe5e58",
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Zero-shot using new multi-modal embedding:\n",
      "Overall Dataset:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.6476    0.2473    0.3579     29306\n",
      "         1.0     0.4214    0.8029    0.5527     20010\n",
      "\n",
      "    accuracy                         0.4727     49316\n",
      "   macro avg     0.5345    0.5251    0.4553     49316\n",
      "weighted avg     0.5558    0.4727    0.4370     49316\n",
      "\n",
      "Training Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.6414    0.2436    0.3531     20463\n",
      "         1.0     0.4214    0.8017    0.5524     14058\n",
      "\n",
      "    accuracy                         0.4709     34521\n",
      "   macro avg     0.5314    0.5227    0.4528     34521\n",
      "weighted avg     0.5518    0.4709    0.4343     34521\n",
      "\n",
      "Validation Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.6618    0.2659    0.3794      4400\n",
      "         1.0     0.4262    0.8005    0.5562      2997\n",
      "\n",
      "    accuracy                         0.4825      7397\n",
      "   macro avg     0.5440    0.5332    0.4678      7397\n",
      "weighted avg     0.5663    0.4825    0.4510      7397\n",
      "\n",
      "Validation Set with both Images and Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7950    0.6687    0.7264       806\n",
      "         1.0     0.4343    0.5959    0.5025       344\n",
      "\n",
      "    accuracy                         0.6470      1150\n",
      "   macro avg     0.6147    0.6323    0.6144      1150\n",
      "weighted avg     0.6871    0.6470    0.6594      1150\n",
      "\n",
      "Validation Set with only Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.5789    0.1756    0.2694      3594\n",
      "         1.0     0.4254    0.8270    0.5618      2653\n",
      "\n",
      "    accuracy                         0.4522      6247\n",
      "   macro avg     0.5022    0.5013    0.4156      6247\n",
      "weighted avg     0.5137    0.4522    0.3936      6247\n",
      "\n",
      "Validation Set with unseen Images:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7538    0.6142    0.6769       324\n",
      "         1.0     0.4292    0.5912    0.4974       159\n",
      "\n",
      "    accuracy                         0.6066       483\n",
      "   macro avg     0.5915    0.6027    0.5871       483\n",
      "weighted avg     0.6469    0.6066    0.6178       483\n",
      "\n",
      "\n",
      "    Overall Accuracy: 47.27\n",
      "    Training Set Accuracy: 47.09\n",
      "    Validation Set Accuracy: 48.25\n",
      "    Validation Set with both Images and Texts: 64.70\n",
      "    Validation Set with only Texts: 45.22\n",
      "    Validation Set with unseen Images: 60.66\n",
      "    \n",
      "Zero-shot using new multi-modal embedding (test):\n",
      "Overall Dataset:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.6476    0.2473    0.3579     29306\n",
      "         1.0     0.4214    0.8029    0.5527     20010\n",
      "\n",
      "    accuracy                         0.4727     49316\n",
      "   macro avg     0.5345    0.5251    0.4553     49316\n",
      "weighted avg     0.5558    0.4727    0.4370     49316\n",
      "\n",
      "Training Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.6414    0.2436    0.3531     20463\n",
      "         1.0     0.4214    0.8017    0.5524     14058\n",
      "\n",
      "    accuracy                         0.4709     34521\n",
      "   macro avg     0.5314    0.5227    0.4528     34521\n",
      "weighted avg     0.5518    0.4709    0.4343     34521\n",
      "\n",
      "Validation Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.6616    0.2460    0.3587      4443\n",
      "         1.0     0.4170    0.8108    0.5507      2955\n",
      "\n",
      "    accuracy                         0.4716      7398\n",
      "   macro avg     0.5393    0.5284    0.4547      7398\n",
      "weighted avg     0.5639    0.4716    0.4354      7398\n",
      "\n",
      "Validation Set with both Images and Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7602    0.6774    0.7164       716\n",
      "         1.0     0.4577    0.5603    0.5039       348\n",
      "\n",
      "    accuracy                         0.6391      1064\n",
      "   macro avg     0.6090    0.6189    0.6101      1064\n",
      "weighted avg     0.6613    0.6391    0.6469      1064\n",
      "\n",
      "Validation Set with only Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.5996    0.1631    0.2565      3727\n",
      "         1.0     0.4137    0.8443    0.5553      2607\n",
      "\n",
      "    accuracy                         0.4435      6334\n",
      "   macro avg     0.5067    0.5037    0.4059      6334\n",
      "weighted avg     0.5231    0.4435    0.3795      6334\n",
      "\n",
      "Validation Set with unseen Images:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7121    0.6559    0.6828       279\n",
      "         1.0     0.4576    0.5226    0.4880       155\n",
      "\n",
      "    accuracy                         0.6083       434\n",
      "   macro avg     0.5848    0.5892    0.5854       434\n",
      "weighted avg     0.6212    0.6083    0.6132       434\n",
      "\n",
      "\n",
      "    Overall Accuracy: 47.27\n",
      "    Training Set Accuracy: 47.09\n",
      "    Validation Set Accuracy: 47.16\n",
      "    Validation Set with both Images and Texts: 63.91\n",
      "    Validation Set with only Texts: 44.35\n",
      "    Validation Set with unseen Images: 60.83\n",
      "    \n"
     ]
    }
   ],
   "source": [
    "MUL_LAB = util.cos_sim(new_lab_emb, new_emb)\n",
    "pred_MUL = torch.argmax(MUL_LAB[:2,:], dim=0).float()\n",
    "lab = torch.tensor(np.array(df['skepticism'])).to('cuda').float()\n",
    "print('Zero-shot using new multi-modal embedding:')\n",
    "embeddings['O1']['L0']['val'] = evaluate_dataset_F1(df, pred_MUL)\n",
    "print('Zero-shot using new multi-modal embedding (test):')\n",
    "embeddings['O1']['L0']['test'] = evaluate_dataset_F1(df, pred_MUL, val_set='test')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ba260af9",
   "metadata": {},
   "source": [
    "### Option 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "id": "87ab7f77",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "MLPClassifier(\n",
       "  (dropout): Dropout(p=0.1, inplace=False)\n",
       "  (fc1): Linear(in_features=512, out_features=128, bias=True)\n",
       "  (fc2): Linear(in_features=128, out_features=2, bias=True)\n",
       "  (fc_concat): Linear(in_features=1024, out_features=128, bias=True)\n",
       "  (fc_complex): Linear(in_features=2560, out_features=1024, bias=True)\n",
       ")"
      ]
     },
     "execution_count": 67,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "classifier = MLPClassifier(embedding_size=args.embedding_size, \n",
    "                        hidden_dim=args.hidden_dim, \n",
    "                        embedding_type=args.embedding_type, \n",
    "                        fusion_type=args.fusion_type,\n",
    "                        num_classes=len(labels), \n",
    "                        dropout_p=args.dropout_p).to(args.device)\n",
    "classifier.load_state_dict(torch.load(args.save_dir + '002/' + args.embedding_type + \"_\" + args.fusion_type + \"_\" + args.model_state_file,map_location=torch.device('cpu')))\n",
    "classifier.eval()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "id": "913f8c7a",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "torch.Size([49316, 128])"
      ]
     },
     "execution_count": 68,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "new_emb = F.relu(classifier.dropout(classifier.fc_concat(F.relu(classifier.dropout(classifier.fc_complex(prepare_embedding(stored_data, args.embedding_type, args.fusion_type)))))))\n",
    "new_emb.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "id": "bb70907b",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "torch.Size([2, 128])"
      ]
     },
     "execution_count": 69,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "new_lab_emb = F.relu(classifier.dropout(classifier.fc_concat(F.relu(classifier.dropout(classifier.fc_complex(prepare_embedding(labs, args.embedding_type, args.fusion_type)))))))\n",
    "new_lab_emb.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "id": "dfbc8057",
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Zero-shot using new multi-modal embedding:\n",
      "Overall Dataset:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8708    0.7593    0.8113     29306\n",
      "         1.0     0.7032    0.8350    0.7634     20010\n",
      "\n",
      "    accuracy                         0.7900     49316\n",
      "   macro avg     0.7870    0.7972    0.7873     49316\n",
      "weighted avg     0.8028    0.7900    0.7918     49316\n",
      "\n",
      "Training Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8930    0.7813    0.8334     20463\n",
      "         1.0     0.7307    0.8637    0.7917     14058\n",
      "\n",
      "    accuracy                         0.8149     34521\n",
      "   macro avg     0.8118    0.8225    0.8125     34521\n",
      "weighted avg     0.8269    0.8149    0.8164     34521\n",
      "\n",
      "Validation Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8247    0.7175    0.7674      4400\n",
      "         1.0     0.6517    0.7761    0.7085      2997\n",
      "\n",
      "    accuracy                         0.7412      7397\n",
      "   macro avg     0.7382    0.7468    0.7379      7397\n",
      "weighted avg     0.7546    0.7412    0.7435      7397\n",
      "\n",
      "Validation Set with both Images and Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.9385    0.9082    0.9231       806\n",
      "         1.0     0.8000    0.8605    0.8291       344\n",
      "\n",
      "    accuracy                         0.8939      1150\n",
      "   macro avg     0.8692    0.8843    0.8761      1150\n",
      "weighted avg     0.8970    0.8939    0.8950      1150\n",
      "\n",
      "Validation Set with only Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7956    0.6747    0.7302      3594\n",
      "         1.0     0.6346    0.7652    0.6938      2653\n",
      "\n",
      "    accuracy                         0.7131      6247\n",
      "   macro avg     0.7151    0.7200    0.7120      6247\n",
      "weighted avg     0.7272    0.7131    0.7147      6247\n",
      "\n",
      "Validation Set with unseen Images:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8571    0.7963    0.8256       324\n",
      "         1.0     0.6374    0.7296    0.6804       159\n",
      "\n",
      "    accuracy                         0.7743       483\n",
      "   macro avg     0.7473    0.7629    0.7530       483\n",
      "weighted avg     0.7848    0.7743    0.7778       483\n",
      "\n",
      "\n",
      "    Overall Accuracy: 79.00\n",
      "    Training Set Accuracy: 81.49\n",
      "    Validation Set Accuracy: 74.12\n",
      "    Validation Set with both Images and Texts: 89.39\n",
      "    Validation Set with only Texts: 71.31\n",
      "    Validation Set with unseen Images: 77.43\n",
      "    \n",
      "Zero-shot using new multi-modal embedding (test):\n",
      "Overall Dataset:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8708    0.7593    0.8113     29306\n",
      "         1.0     0.7032    0.8350    0.7634     20010\n",
      "\n",
      "    accuracy                         0.7900     49316\n",
      "   macro avg     0.7870    0.7972    0.7873     49316\n",
      "weighted avg     0.8028    0.7900    0.7918     49316\n",
      "\n",
      "Training Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8930    0.7813    0.8334     20463\n",
      "         1.0     0.7307    0.8637    0.7917     14058\n",
      "\n",
      "    accuracy                         0.8149     34521\n",
      "   macro avg     0.8118    0.8225    0.8125     34521\n",
      "weighted avg     0.8269    0.8149    0.8164     34521\n",
      "\n",
      "Validation Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8130    0.6995    0.7520      4443\n",
      "         1.0     0.6266    0.7580    0.6861      2955\n",
      "\n",
      "    accuracy                         0.7229      7398\n",
      "   macro avg     0.7198    0.7288    0.7190      7398\n",
      "weighted avg     0.7385    0.7229    0.7257      7398\n",
      "\n",
      "Validation Set with both Images and Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.9148    0.8994    0.9070       716\n",
      "         1.0     0.8000    0.8276    0.8136       348\n",
      "\n",
      "    accuracy                         0.8759      1064\n",
      "   macro avg     0.8574    0.8635    0.8603      1064\n",
      "weighted avg     0.8772    0.8759    0.8765      1064\n",
      "\n",
      "Validation Set with only Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7900    0.6611    0.7198      3727\n",
      "         1.0     0.6072    0.7488    0.6706      2607\n",
      "\n",
      "    accuracy                         0.6972      6334\n",
      "   macro avg     0.6986    0.7049    0.6952      6334\n",
      "weighted avg     0.7147    0.6972    0.6996      6334\n",
      "\n",
      "Validation Set with unseen Images:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7942    0.7885    0.7914       279\n",
      "         1.0     0.6242    0.6323    0.6282       155\n",
      "\n",
      "    accuracy                         0.7327       434\n",
      "   macro avg     0.7092    0.7104    0.7098       434\n",
      "weighted avg     0.7335    0.7327    0.7331       434\n",
      "\n",
      "\n",
      "    Overall Accuracy: 79.00\n",
      "    Training Set Accuracy: 81.49\n",
      "    Validation Set Accuracy: 72.29\n",
      "    Validation Set with both Images and Texts: 87.59\n",
      "    Validation Set with only Texts: 69.72\n",
      "    Validation Set with unseen Images: 73.27\n",
      "    \n"
     ]
    }
   ],
   "source": [
    "MUL_LAB = util.cos_sim(new_lab_emb, new_emb)\n",
    "pred_MUL = torch.argmax(MUL_LAB[:2,:], dim=0).float()\n",
    "lab = torch.tensor(np.array(df['skepticism'])).to('cuda').float()\n",
    "print('Zero-shot using new multi-modal embedding:')\n",
    "embeddings['O2']['L2']['val'] = evaluate_dataset_F1(df, pred_MUL)\n",
    "print('Zero-shot using new multi-modal embedding (test):')\n",
    "embeddings['O2']['L2']['test'] = evaluate_dataset_F1(df, pred_MUL, val_set='test')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "id": "d6aa36c6",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "torch.Size([49316, 1024])"
      ]
     },
     "execution_count": 71,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "new_emb = F.relu(classifier.dropout(classifier.fc_complex(prepare_embedding(stored_data, args.embedding_type, args.fusion_type))))\n",
    "new_emb.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "id": "ffcaaf9a",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "torch.Size([2, 1024])"
      ]
     },
     "execution_count": 72,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "new_lab_emb = F.relu(classifier.dropout(classifier.fc_complex(prepare_embedding(labs, args.embedding_type, args.fusion_type))))\n",
    "new_lab_emb.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "id": "aec96d26",
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Zero-shot using new multi-modal embedding:\n",
      "Overall Dataset:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7304    0.7667    0.7481     29306\n",
      "         1.0     0.6315    0.5855    0.6076     20010\n",
      "\n",
      "    accuracy                         0.6932     49316\n",
      "   macro avg     0.6809    0.6761    0.6779     49316\n",
      "weighted avg     0.6903    0.6932    0.6911     49316\n",
      "\n",
      "Training Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7343    0.7696    0.7515     20463\n",
      "         1.0     0.6394    0.5946    0.6162     14058\n",
      "\n",
      "    accuracy                         0.6984     34521\n",
      "   macro avg     0.6868    0.6821    0.6839     34521\n",
      "weighted avg     0.6957    0.6984    0.6964     34521\n",
      "\n",
      "Validation Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7235    0.7661    0.7442      4400\n",
      "         1.0     0.6242    0.5702    0.5960      2997\n",
      "\n",
      "    accuracy                         0.6868      7397\n",
      "   macro avg     0.6739    0.6682    0.6701      7397\n",
      "weighted avg     0.6833    0.6868    0.6842      7397\n",
      "\n",
      "Validation Set with both Images and Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.8078    0.9541    0.8749       806\n",
      "         1.0     0.8131    0.4680    0.5941       344\n",
      "\n",
      "    accuracy                         0.8087      1150\n",
      "   macro avg     0.8105    0.7111    0.7345      1150\n",
      "weighted avg     0.8094    0.8087    0.7909      1150\n",
      "\n",
      "Validation Set with only Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7019    0.7240    0.7128      3594\n",
      "         1.0     0.6094    0.5835    0.5962      2653\n",
      "\n",
      "    accuracy                         0.6643      6247\n",
      "   macro avg     0.6557    0.6537    0.6545      6247\n",
      "weighted avg     0.6626    0.6643    0.6633      6247\n",
      "\n",
      "Validation Set with unseen Images:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7708    0.9136    0.8362       324\n",
      "         1.0     0.7172    0.4465    0.5504       159\n",
      "\n",
      "    accuracy                         0.7598       483\n",
      "   macro avg     0.7440    0.6801    0.6933       483\n",
      "weighted avg     0.7532    0.7598    0.7421       483\n",
      "\n",
      "\n",
      "    Overall Accuracy: 69.32\n",
      "    Training Set Accuracy: 69.84\n",
      "    Validation Set Accuracy: 68.68\n",
      "    Validation Set with both Images and Texts: 80.87\n",
      "    Validation Set with only Texts: 66.43\n",
      "    Validation Set with unseen Images: 75.98\n",
      "    \n",
      "Zero-shot using new multi-modal embedding (test):\n",
      "Overall Dataset:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7304    0.7667    0.7481     29306\n",
      "         1.0     0.6315    0.5855    0.6076     20010\n",
      "\n",
      "    accuracy                         0.6932     49316\n",
      "   macro avg     0.6809    0.6761    0.6779     49316\n",
      "weighted avg     0.6903    0.6932    0.6911     49316\n",
      "\n",
      "Training Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7343    0.7696    0.7515     20463\n",
      "         1.0     0.6394    0.5946    0.6162     14058\n",
      "\n",
      "    accuracy                         0.6984     34521\n",
      "   macro avg     0.6868    0.6821    0.6839     34521\n",
      "weighted avg     0.6957    0.6984    0.6964     34521\n",
      "\n",
      "Validation Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7192    0.7540    0.7362      4443\n",
      "         1.0     0.6011    0.5574    0.5784      2955\n",
      "\n",
      "    accuracy                         0.6755      7398\n",
      "   macro avg     0.6601    0.6557    0.6573      7398\n",
      "weighted avg     0.6720    0.6755    0.6732      7398\n",
      "\n",
      "Validation Set with both Images and Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7832    0.9483    0.8579       716\n",
      "         1.0     0.8122    0.4598    0.5872       348\n",
      "\n",
      "    accuracy                         0.7885      1064\n",
      "   macro avg     0.7977    0.7040    0.7225      1064\n",
      "weighted avg     0.7927    0.7885    0.7693      1064\n",
      "\n",
      "Validation Set with only Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7046    0.7167    0.7106      3727\n",
      "         1.0     0.5847    0.5704    0.5775      2607\n",
      "\n",
      "    accuracy                         0.6565      6334\n",
      "   macro avg     0.6447    0.6435    0.6440      6334\n",
      "weighted avg     0.6552    0.6565    0.6558      6334\n",
      "\n",
      "Validation Set with unseen Images:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7217    0.8925    0.7981       279\n",
      "         1.0     0.6629    0.3806    0.4836       155\n",
      "\n",
      "    accuracy                         0.7097       434\n",
      "   macro avg     0.6923    0.6366    0.6408       434\n",
      "weighted avg     0.7007    0.7097    0.6858       434\n",
      "\n",
      "\n",
      "    Overall Accuracy: 69.32\n",
      "    Training Set Accuracy: 69.84\n",
      "    Validation Set Accuracy: 67.55\n",
      "    Validation Set with both Images and Texts: 78.85\n",
      "    Validation Set with only Texts: 65.65\n",
      "    Validation Set with unseen Images: 70.97\n",
      "    \n"
     ]
    }
   ],
   "source": [
    "MUL_LAB = util.cos_sim(new_lab_emb, new_emb)\n",
    "pred_MUL = torch.argmax(MUL_LAB[:2,:], dim=0).float()\n",
    "lab = torch.tensor(np.array(df['skepticism'])).to('cuda').float()\n",
    "print('Zero-shot using new multi-modal embedding:')\n",
    "embeddings['O2']['L1']['val'] = evaluate_dataset_F1(df, pred_MUL)\n",
    "print('Zero-shot using new multi-modal embedding (test):')\n",
    "embeddings['O2']['L1']['test'] = evaluate_dataset_F1(df, pred_MUL, val_set='test')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "id": "8bbf9085",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "torch.Size([49316, 2560])"
      ]
     },
     "execution_count": 74,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "new_emb = prepare_embedding(stored_data, args.embedding_type, args.fusion_type)\n",
    "new_emb.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "id": "fe5505c3",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "torch.Size([2, 2560])"
      ]
     },
     "execution_count": 75,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "new_lab_emb = prepare_embedding(labs, args.embedding_type, args.fusion_type)\n",
    "new_lab_emb.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "id": "68d0e2e6",
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Zero-shot using new multi-modal embedding:\n",
      "Overall Dataset:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.6476    0.2473    0.3579     29306\n",
      "         1.0     0.4214    0.8029    0.5527     20010\n",
      "\n",
      "    accuracy                         0.4727     49316\n",
      "   macro avg     0.5345    0.5251    0.4553     49316\n",
      "weighted avg     0.5558    0.4727    0.4370     49316\n",
      "\n",
      "Training Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.6414    0.2436    0.3531     20463\n",
      "         1.0     0.4214    0.8017    0.5524     14058\n",
      "\n",
      "    accuracy                         0.4709     34521\n",
      "   macro avg     0.5314    0.5227    0.4528     34521\n",
      "weighted avg     0.5518    0.4709    0.4343     34521\n",
      "\n",
      "Validation Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.6618    0.2659    0.3794      4400\n",
      "         1.0     0.4262    0.8005    0.5562      2997\n",
      "\n",
      "    accuracy                         0.4825      7397\n",
      "   macro avg     0.5440    0.5332    0.4678      7397\n",
      "weighted avg     0.5663    0.4825    0.4510      7397\n",
      "\n",
      "Validation Set with both Images and Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7950    0.6687    0.7264       806\n",
      "         1.0     0.4343    0.5959    0.5025       344\n",
      "\n",
      "    accuracy                         0.6470      1150\n",
      "   macro avg     0.6147    0.6323    0.6144      1150\n",
      "weighted avg     0.6871    0.6470    0.6594      1150\n",
      "\n",
      "Validation Set with only Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.5789    0.1756    0.2694      3594\n",
      "         1.0     0.4254    0.8270    0.5618      2653\n",
      "\n",
      "    accuracy                         0.4522      6247\n",
      "   macro avg     0.5022    0.5013    0.4156      6247\n",
      "weighted avg     0.5137    0.4522    0.3936      6247\n",
      "\n",
      "Validation Set with unseen Images:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7538    0.6142    0.6769       324\n",
      "         1.0     0.4292    0.5912    0.4974       159\n",
      "\n",
      "    accuracy                         0.6066       483\n",
      "   macro avg     0.5915    0.6027    0.5871       483\n",
      "weighted avg     0.6469    0.6066    0.6178       483\n",
      "\n",
      "\n",
      "    Overall Accuracy: 47.27\n",
      "    Training Set Accuracy: 47.09\n",
      "    Validation Set Accuracy: 48.25\n",
      "    Validation Set with both Images and Texts: 64.70\n",
      "    Validation Set with only Texts: 45.22\n",
      "    Validation Set with unseen Images: 60.66\n",
      "    \n",
      "Zero-shot using new multi-modal embedding (test):\n",
      "Overall Dataset:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.6476    0.2473    0.3579     29306\n",
      "         1.0     0.4214    0.8029    0.5527     20010\n",
      "\n",
      "    accuracy                         0.4727     49316\n",
      "   macro avg     0.5345    0.5251    0.4553     49316\n",
      "weighted avg     0.5558    0.4727    0.4370     49316\n",
      "\n",
      "Training Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.6414    0.2436    0.3531     20463\n",
      "         1.0     0.4214    0.8017    0.5524     14058\n",
      "\n",
      "    accuracy                         0.4709     34521\n",
      "   macro avg     0.5314    0.5227    0.4528     34521\n",
      "weighted avg     0.5518    0.4709    0.4343     34521\n",
      "\n",
      "Validation Set:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.6616    0.2460    0.3587      4443\n",
      "         1.0     0.4170    0.8108    0.5507      2955\n",
      "\n",
      "    accuracy                         0.4716      7398\n",
      "   macro avg     0.5393    0.5284    0.4547      7398\n",
      "weighted avg     0.5639    0.4716    0.4354      7398\n",
      "\n",
      "Validation Set with both Images and Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7602    0.6774    0.7164       716\n",
      "         1.0     0.4577    0.5603    0.5039       348\n",
      "\n",
      "    accuracy                         0.6391      1064\n",
      "   macro avg     0.6090    0.6189    0.6101      1064\n",
      "weighted avg     0.6613    0.6391    0.6469      1064\n",
      "\n",
      "Validation Set with only Texts:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.5996    0.1631    0.2565      3727\n",
      "         1.0     0.4137    0.8443    0.5553      2607\n",
      "\n",
      "    accuracy                         0.4435      6334\n",
      "   macro avg     0.5067    0.5037    0.4059      6334\n",
      "weighted avg     0.5231    0.4435    0.3795      6334\n",
      "\n",
      "Validation Set with unseen Images:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0     0.7121    0.6559    0.6828       279\n",
      "         1.0     0.4576    0.5226    0.4880       155\n",
      "\n",
      "    accuracy                         0.6083       434\n",
      "   macro avg     0.5848    0.5892    0.5854       434\n",
      "weighted avg     0.6212    0.6083    0.6132       434\n",
      "\n",
      "\n",
      "    Overall Accuracy: 47.27\n",
      "    Training Set Accuracy: 47.09\n",
      "    Validation Set Accuracy: 47.16\n",
      "    Validation Set with both Images and Texts: 63.91\n",
      "    Validation Set with only Texts: 44.35\n",
      "    Validation Set with unseen Images: 60.83\n",
      "    \n"
     ]
    }
   ],
   "source": [
    "MUL_LAB = util.cos_sim(new_lab_emb, new_emb)\n",
    "pred_MUL = torch.argmax(MUL_LAB[:2,:], dim=0).float()\n",
    "lab = torch.tensor(np.array(df['skepticism'])).to('cuda').float()\n",
    "print('Zero-shot using new multi-modal embedding:')\n",
    "embeddings['O2']['L0']['val'] = evaluate_dataset_F1(df, pred_MUL)\n",
    "print('Zero-shot using new multi-modal embedding (test):')\n",
    "embeddings['O2']['L0']['test'] = evaluate_dataset_F1(df, pred_MUL, val_set='test')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "id": "b46a0293",
   "metadata": {},
   "outputs": [],
   "source": [
    "results = {}\n",
    "results['fusion'] = perform_results\n",
    "results['new_embedding'] = embeddings"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "id": "474c0a22",
   "metadata": {},
   "outputs": [],
   "source": [
    "import pickle\n",
    "with open(args.result_dir + 'results_F1.pkl', \"wb\") as fOut:\n",
    "    pickle.dump(results, fOut, protocol=pickle.HIGHEST_PROTOCOL)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "5d4960e8",
   "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": "3.11.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}