3_2_final_modelling.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 50
    },
    "colab_type": "code",
    "id": "-hIM5G4oM5jh",
    "outputId": "18434e33-1f21-4f7a-ba13-c33360dcb2fe"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.6/dist-packages/statsmodels/tools/_testing.py:19: FutureWarning: pandas.util.testing is deprecated. Use the functions in the public API at pandas.testing instead.\n",
      "  import pandas.util.testing as tm\n"
     ]
    }
   ],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "from tqdm.notebook import tqdm\n",
    "import re\n",
    "import warnings\n",
    "import html\n",
    "warnings.filterwarnings('ignore')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 98
    },
    "colab_type": "code",
    "id": "Rg4Z4VFdO9AW",
    "outputId": "7656133a-e593-4398-b6db-2c89d80b3c75"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Archive:  /content/drive/My Drive/AAIC Course/Personal case study 2  - Google Quest/google-quest-challenge.zip\n",
      "  inflating: sample_submission.csv   \n",
      "  inflating: test.csv                \n",
      "  inflating: train.csv               \n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "((6079, 41), (476, 11))"
      ]
     },
     "execution_count": 2,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "!unzip '/content/drive/My Drive/AAIC Course/Personal case study 2  - Google Quest/google-quest-challenge.zip'\n",
    "df_train = pd.read_csv('train.csv')\n",
    "df_test = pd.read_csv('test.csv')\n",
    "df_train.shape, df_test.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "TLEozz70PP46"
   },
   "source": [
    "# 3. Data Preprocessing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 50
    },
    "colab_type": "code",
    "id": "BqbK7HdKPWhw",
    "outputId": "1c5daa67-ac52-4e77-f133-e2d4f7fe0124"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[nltk_data] Downloading package stopwords to /root/nltk_data...\n",
      "[nltk_data]   Unzipping corpora/stopwords.zip.\n"
     ]
    }
   ],
   "source": [
    "# defining a function to remove stop_words\n",
    "import nltk\n",
    "from nltk.corpus import stopwords\n",
    "nltk.download('stopwords')\n",
    "stop_words = set(stopwords.words('english'))\n",
    "stop_words.remove('no'); stop_words.remove('not'); stop_words.remove('nor')\n",
    " \n",
    "def stopwrd_removal(sent):\n",
    "  lst = []\n",
    "  for wrd in sent.split():\n",
    "    if wrd not in stop_words:\n",
    "      lst.append(wrd)\n",
    "  return \" \".join(lst)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "hoVwES0MPefG"
   },
   "outputs": [],
   "source": [
    "def text_preprocessor(column, remove_stopwords = False, remove_specialchar = False):\n",
    "  \"\"\"pass any column with Text in it from df_train | Note: returns nothing makes inplace changes in df_train\"\"\"\n",
    "  # 1. remove html tags, html urls, replace html comparison operators\n",
    "  # text = df_train[column].values\n",
    "  df_train[column] = [re.sub('<.*?>', ' ', i) for i in df_train[column].values]\n",
    "  df_train[column] = df_train[column].str.replace('&lt;', '<')\\\n",
    "                                          .str.replace('&gt;', '>')\\\n",
    "                                          .str.replace('&le;', '<=' )\\\n",
    "                                          .str.replace('&ge;', '>=')\n",
    " \n",
    "  # 2. remove latex i,e., if there is any formulas or latex we have to remove it\n",
    "  df_train[column] = [re.sub('\\$.*?\\$', ' ', i) for i in df_train[column].values]\n",
    " \n",
    "  # 3. all lowercase \n",
    "  df_train[column] = df_train[column].str.lower()\n",
    " \n",
    "  # 4. decontractions\n",
    "  df_train[column] = df_train[column].str.replace(\"won't\", \"will not\").str.replace(\"can\\'t\", \"can not\").str.replace(\"n\\'t\", \" not\").str.replace(\"\\'re\", \" are\").str.\\\n",
    "                                                replace(\"\\'s\", \" is\").str.replace(\"\\'d\", \" would\").str.replace(\"\\'ll\", \" will\").str.\\\n",
    "                                                replace(\"\\'t\", \" not\").str.replace(\"\\'ve\", \" have\").str.replace(\"\\'m\", \" am\")\n",
    "  \n",
    "  # 5. removing non-english or hebrew characters\n",
    "  df_train[column] = [i.encode(\"ascii\", \"ignore\").decode() for i in df_train[column].values]\n",
    " \n",
    "  # 6. remove all special-characters other than alpha-numericals\n",
    "  if remove_specialchar == True:\n",
    "    df_train[column] = [re.sub('[^A-Za-z0-9]+', ' ', i) for i in df_train[column].values]\n",
    " \n",
    "  # # 7. separating special chars from alphanumerics\n",
    "  # all_sc = [re.findall('[^ A-Za-z0-9]', i) for i in df_train[column].values]\n",
    "  # special_char = np.unique([j for i in all_sc for j in i])\n",
    "  # replace_char = [' '+i+' ' for i in special_char]\n",
    "  # for i,j in zip(special_char, replace_char):\n",
    "  #  df_train[column] = df_train[column].str.replace(i, j)\n",
    " \n",
    "  # 8. Stop_word removal\n",
    "  if remove_stopwords == True:\n",
    "    df_train[column] = [stopwrd_removal(i) for i in df_train[column].values]\n",
    " \n",
    "  # 9. remove all white-space i.e., \\n, \\t, and extra_spaces\n",
    "  df_train[column] = df_train[column].str.replace(\"\\n\", \" \").str.replace(\"\\t\", \" \").str.rstrip()\n",
    "  df_train[column] = [re.sub('  +', ' ', i) for i in df_train[column].values]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "IQ85mE8zjfSK"
   },
   "outputs": [],
   "source": [
    "# 1. text preprocessing\n",
    "df_train['clean_title'] = df_train['question_title']\n",
    "df_train['clean_body'] = df_train['question_body']\n",
    "df_train['clean_answer'] = df_train['answer']\n",
    "text_preprocessor('clean_title',  remove_stopwords = False, remove_specialchar = False)\n",
    "text_preprocessor('clean_body',  remove_stopwords = False, remove_specialchar = False)\n",
    "text_preprocessor('clean_answer',  remove_stopwords = False, remove_specialchar = False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "onEsk2ueQk_s"
   },
   "source": [
    "# 4. Train_test_split - Random split"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 33
    },
    "colab_type": "code",
    "id": "Etu9lo1iSwj0",
    "outputId": "f928ad91-67ba-4f6c-ecee-3612fb679a23"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "30"
      ]
     },
     "execution_count": 6,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 1. setting up target features\n",
    "question_tar = ['question_asker_intent_understanding',\n",
    "       'question_body_critical', 'question_conversational',\n",
    "       'question_expect_short_answer', 'question_fact_seeking',\n",
    "       'question_has_commonly_accepted_answer',\n",
    "       'question_interestingness_others', 'question_interestingness_self',\n",
    "       'question_multi_intent', 'question_not_really_a_question',\n",
    "       'question_opinion_seeking', 'question_type_choice',\n",
    "       'question_type_compare', 'question_type_consequence',\n",
    "       'question_type_definition', 'question_type_entity',\n",
    "       'question_type_instructions', 'question_type_procedure',\n",
    "       'question_type_reason_explanation', 'question_type_spelling',\n",
    "       'question_well_written']\n",
    "       \n",
    "answer_tar = ['answer_helpful',\n",
    "       'answer_level_of_information', 'answer_plausible', 'answer_relevance',\n",
    "       'answer_satisfaction', 'answer_type_instructions',\n",
    "       'answer_type_procedure', 'answer_type_reason_explanation',\n",
    "       'answer_well_written']\n",
    " \n",
    "tar_features = question_tar + answer_tar\n",
    "len(tar_features)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 33
    },
    "colab_type": "code",
    "id": "PpP3IoR0Qta-",
    "outputId": "0951fe8f-2a37-4c6c-ea77-5df1a1e33213"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((5349, 3), (730, 3), (5349, 30), (730, 30))"
      ]
     },
     "execution_count": 7,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 2. splitting dataset train_test_split\n",
    "from sklearn.model_selection import train_test_split\n",
    " \n",
    "X_train, X_cv, y_train, y_cv = train_test_split(df_train[['clean_title', 'clean_body', 'clean_answer']], df_train[tar_features], test_size = 0.12, random_state = 42)\n",
    "X_train.shape, X_cv.shape, y_train.shape, y_cv.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 33
    },
    "colab_type": "code",
    "id": "C43FZkxzXKWu",
    "outputId": "0711eba8-f21d-4b4d-aa68-6c92ba6558ab"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(5349, 5349, 730, 730)"
      ]
     },
     "execution_count": 8,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 3. creating training features : title + body = title_body | answer_train | title + body + answer = title_body_answer\n",
    "title_train = X_train['clean_title'].values\n",
    "body_train = X_train['clean_body'].values\n",
    "answer_train = X_train['clean_answer'].values\n",
    " \n",
    "title_cv = X_cv['clean_title'].values\n",
    "body_cv = X_cv['clean_body'].values\n",
    "answer_cv = X_cv['clean_answer'].values\n",
    " \n",
    "# train data\n",
    "title_body_train = [i+' '+j for i,j in zip(title_train, body_train)]\n",
    "y_train_ques = y_train[question_tar].values\n",
    "y_train_ans = y_train[answer_tar].values\n",
    " \n",
    "# cv data\n",
    "title_body_cv = [i+' '+j for i,j in zip(title_cv, body_cv)]\n",
    "y_cv_ques = y_cv[question_tar].values\n",
    "y_cv_ans = y_cv[answer_tar].values\n",
    " \n",
    "len(title_body_train), len(answer_train), len(title_body_cv), len(answer_cv)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "1nRC3ble_z3B"
   },
   "source": [
    "# 5.1. Sentence Embedding : BERT"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 570
    },
    "colab_type": "code",
    "id": "Vjqn1vExAbT8",
    "outputId": "c40a6d53-e33d-4a33-b1cd-d0f2b172ee9e"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Collecting transformers\n",
      "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/27/3c/91ed8f5c4e7ef3227b4119200fc0ed4b4fd965b1f0172021c25701087825/transformers-3.0.2-py3-none-any.whl (769kB)\n",
      "\r",
      "\u001b[K     |▍                               | 10kB 22.0MB/s eta 0:00:01\r",
      "\u001b[K     |▉                               | 20kB 3.0MB/s eta 0:00:01\r",
      "\u001b[K     |█▎                              | 30kB 4.0MB/s eta 0:00:01\r",
      "\u001b[K     |█▊                              | 40kB 4.3MB/s eta 0:00:01\r",
      "\u001b[K     |██▏                             | 51kB 3.5MB/s eta 0:00:01\r",
      "\u001b[K     |██▋                             | 61kB 3.9MB/s eta 0:00:01\r",
      "\u001b[K     |███                             | 71kB 4.2MB/s eta 0:00:01\r",
      "\u001b[K     |███▍                            | 81kB 4.6MB/s eta 0:00:01\r",
      "\u001b[K     |███▉                            | 92kB 4.9MB/s eta 0:00:01\r",
      "\u001b[K     |████▎                           | 102kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |████▊                           | 112kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |█████▏                          | 122kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |█████▌                          | 133kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |██████                          | 143kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |██████▍                         | 153kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |██████▉                         | 163kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |███████▎                        | 174kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |███████▊                        | 184kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |████████                        | 194kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |████████▌                       | 204kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |█████████                       | 215kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |█████████▍                      | 225kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |█████████▉                      | 235kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |██████████▎                     | 245kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |██████████▋                     | 256kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |███████████                     | 266kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |███████████▌                    | 276kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |████████████                    | 286kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |████████████▍                   | 296kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |████████████▉                   | 307kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |█████████████▏                  | 317kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |█████████████▋                  | 327kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |██████████████                  | 337kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |██████████████▌                 | 348kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |███████████████                 | 358kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |███████████████▍                | 368kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |███████████████▊                | 378kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |████████████████▏               | 389kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |████████████████▋               | 399kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |█████████████████               | 409kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |█████████████████▌              | 419kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |██████████████████              | 430kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |██████████████████▎             | 440kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |██████████████████▊             | 450kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |███████████████████▏            | 460kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |███████████████████▋            | 471kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |████████████████████            | 481kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |████████████████████▌           | 491kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |████████████████████▉           | 501kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |█████████████████████▎          | 512kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |█████████████████████▊          | 522kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |██████████████████████▏         | 532kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |██████████████████████▋         | 542kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |███████████████████████         | 552kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |███████████████████████▍        | 563kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |███████████████████████▉        | 573kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |████████████████████████▎       | 583kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |████████████████████████▊       | 593kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |█████████████████████████▏      | 604kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |█████████████████████████▋      | 614kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |██████████████████████████      | 624kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |██████████████████████████▍     | 634kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |██████████████████████████▉     | 645kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |███████████████████████████▎    | 655kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |███████████████████████████▊    | 665kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |████████████████████████████▏   | 675kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |████████████████████████████▌   | 686kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |█████████████████████████████   | 696kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |█████████████████████████████▍  | 706kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |█████████████████████████████▉  | 716kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |██████████████████████████████▎ | 727kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |██████████████████████████████▊ | 737kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |███████████████████████████████ | 747kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |███████████████████████████████▌| 757kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |████████████████████████████████| 768kB 4.7MB/s eta 0:00:01\r",
      "\u001b[K     |████████████████████████████████| 778kB 4.7MB/s \n",
      "\u001b[?25hCollecting sacremoses\n",
      "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/7d/34/09d19aff26edcc8eb2a01bed8e98f13a1537005d31e95233fd48216eed10/sacremoses-0.0.43.tar.gz (883kB)\n",
      "\u001b[K     |████████████████████████████████| 890kB 19.7MB/s \n",
      "\u001b[?25hCollecting sentencepiece!=0.1.92\n",
      "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/d4/a4/d0a884c4300004a78cca907a6ff9a5e9fe4f090f5d95ab341c53d28cbc58/sentencepiece-0.1.91-cp36-cp36m-manylinux1_x86_64.whl (1.1MB)\n",
      "\u001b[K     |████████████████████████████████| 1.1MB 34.2MB/s \n",
      "\u001b[?25hRequirement already satisfied: requests in /usr/local/lib/python3.6/dist-packages (from transformers) (2.23.0)\n",
      "Requirement already satisfied: filelock in /usr/local/lib/python3.6/dist-packages (from transformers) (3.0.12)\n",
      "Requirement already satisfied: packaging in /usr/local/lib/python3.6/dist-packages (from transformers) (20.4)\n",
      "Requirement already satisfied: dataclasses; python_version < \"3.7\" in /usr/local/lib/python3.6/dist-packages (from transformers) (0.7)\n",
      "Requirement already satisfied: regex!=2019.12.17 in /usr/local/lib/python3.6/dist-packages (from transformers) (2019.12.20)\n",
      "Collecting tokenizers==0.8.1.rc1\n",
      "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/40/d0/30d5f8d221a0ed981a186c8eb986ce1c94e3a6e87f994eae9f4aa5250217/tokenizers-0.8.1rc1-cp36-cp36m-manylinux1_x86_64.whl (3.0MB)\n",
      "\u001b[K     |████████████████████████████████| 3.0MB 20.5MB/s \n",
      "\u001b[?25hRequirement already satisfied: numpy in /usr/local/lib/python3.6/dist-packages (from transformers) (1.18.5)\n",
      "Requirement already satisfied: tqdm>=4.27 in /usr/local/lib/python3.6/dist-packages (from transformers) (4.41.1)\n",
      "Requirement already satisfied: six in /usr/local/lib/python3.6/dist-packages (from sacremoses->transformers) (1.15.0)\n",
      "Requirement already satisfied: click in /usr/local/lib/python3.6/dist-packages (from sacremoses->transformers) (7.1.2)\n",
      "Requirement already satisfied: joblib in /usr/local/lib/python3.6/dist-packages (from sacremoses->transformers) (0.16.0)\n",
      "Requirement already satisfied: urllib3!=1.25.0,!=1.25.1,<1.26,>=1.21.1 in /usr/local/lib/python3.6/dist-packages (from requests->transformers) (1.24.3)\n",
      "Requirement already satisfied: certifi>=2017.4.17 in /usr/local/lib/python3.6/dist-packages (from requests->transformers) (2020.6.20)\n",
      "Requirement already satisfied: chardet<4,>=3.0.2 in /usr/local/lib/python3.6/dist-packages (from requests->transformers) (3.0.4)\n",
      "Requirement already satisfied: idna<3,>=2.5 in /usr/local/lib/python3.6/dist-packages (from requests->transformers) (2.10)\n",
      "Requirement already satisfied: pyparsing>=2.0.2 in /usr/local/lib/python3.6/dist-packages (from packaging->transformers) (2.4.7)\n",
      "Building wheels for collected packages: sacremoses\n",
      "  Building wheel for sacremoses (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
      "  Created wheel for sacremoses: filename=sacremoses-0.0.43-cp36-none-any.whl size=893260 sha256=c0f9aded4ebb4e4636bbf0fa634ce8eb2f96d0040bcc3ce968a8d113942eaab4\n",
      "  Stored in directory: /root/.cache/pip/wheels/29/3c/fd/7ce5c3f0666dab31a50123635e6fb5e19ceb42ce38d4e58f45\n",
      "Successfully built sacremoses\n",
      "Installing collected packages: sacremoses, sentencepiece, tokenizers, transformers\n",
      "Successfully installed sacremoses-0.0.43 sentencepiece-0.1.91 tokenizers-0.8.1rc1 transformers-3.0.2\n"
     ]
    }
   ],
   "source": [
    "!pip install transformers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 262,
     "referenced_widgets": [
      "b941cbd32450435fb453f1c88ade22d6",
      "8dc39021daaa447393d7bdd41736088a",
      "3a87b11b1846450588eeb287c14a9cd0",
      "d7664b5f00614731b660cb1f093490c3",
      "e98d667af83c4abb9e2caa7b13648770",
      "2da60bb609d04930bcc574d975625c22",
      "1555f6aac6074ba09dcf569cf9d612f8",
      "88938176fe364f9ab389f5764dd791c1",
      "28696c687449462e9566ad837a105bfc",
      "91fd6282e75e4e71b388a1ff67d450d6",
      "a32279f5d1204557ae0e51c239d3f4c8",
      "ca4adeb16771401d8bbd25352e4fb126",
      "3757d8ddd2e04e11a44fdb802a4af458",
      "d995f5dcd7d14093bacca3750e50e818",
      "2dfb108a53f54071afe40a4f66b80833",
      "d9f6b0bf75eb49588614011e4c453a14",
      "3184ca332a4c410ab0a8abf172d6a247",
      "3b5d20bf24834e0bb9b8856a152e3b1e",
      "0bb5a03c228449f497f57d9fc02f4bbc",
      "999357f02a2446e2815e71a3e6dc979a",
      "48f5e96933a249198d4a7cea366447e6",
      "177475e15cc14018b2352132d426b4e4",
      "ccca2cd356de49f1954702b20136af3a",
      "070ba289643e4bdb8873cc225aad2326"
     ]
    },
    "colab_type": "code",
    "id": "-vamY42PAmkf",
    "outputId": "45202ffa-5604-4421-f2cc-a1a84281216e"
   },
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "b941cbd32450435fb453f1c88ade22d6",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, description='Downloading', max=433.0, style=ProgressStyle(description_…"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "28696c687449462e9566ad837a105bfc",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, description='Downloading', max=231508.0, style=ProgressStyle(descripti…"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "3184ca332a4c410ab0a8abf172d6a247",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, description='Downloading', max=536063208.0, style=ProgressStyle(descri…"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Some weights of the model checkpoint at bert-base-uncased were not used when initializing TFBertModel: ['mlm___cls', 'nsp___cls']\n",
      "- This IS expected if you are initializing TFBertModel from the checkpoint of a model trained on another task or with another architecture (e.g. initializing a BertForSequenceClassification model from a BertForPretraining model).\n",
      "- This IS NOT expected if you are initializing TFBertModel from the checkpoint of a model that you expect to be exactly identical (initializing a BertForSequenceClassification model from a BertForSequenceClassification model).\n",
      "All the weights of TFBertModel were initialized from the model checkpoint at bert-base-uncased.\n",
      "If your task is similar to the task the model of the ckeckpoint was trained on, you can already use TFBertModel for predictions without further training.\n"
     ]
    }
   ],
   "source": [
    "import tensorflow as tf\n",
    "from transformers import BertTokenizer, TFBertModel, BertConfig\n",
    " \n",
    "# Load pretrained model/tokenizer\n",
    "config = BertConfig.from_pretrained('bert-base-uncased', output_hidden_states=True)\n",
    "bert_tokenizer = BertTokenizer.from_pretrained('bert-base-uncased')\n",
    "bert_model = TFBertModel.from_pretrained('bert-base-uncased', config = config)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 33
    },
    "colab_type": "code",
    "id": "93WIPzIFZFNe",
    "outputId": "f933fa30-f5c5-4dc6-f9fe-14197ed81f47"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((5349, 512), (5349, 512), (5349, 512), (730, 512), (730, 512), (730, 512))"
      ]
     },
     "execution_count": 11,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Tokenizing : title_body\n",
    "title_body_train_tokens = bert_tokenizer.batch_encode_plus(title_body_train, max_length = 512, truncation = True, pad_to_max_length = True, return_tensors=\"tf\")\n",
    "title_body_cv_tokens = bert_tokenizer.batch_encode_plus(title_body_cv, max_length = 512, truncation = True, pad_to_max_length = True, return_tensors=\"tf\")\n",
    "# tokens = bert_tokenizer.batch_encode_plus(final_train, max_length = 50, truncation = True, pad_to_max_length = True)\n",
    "\n",
    "tb_train_input_ids  = np.array(title_body_train_tokens['input_ids'])\n",
    "tb_train_attn_mask = np.array(title_body_train_tokens['attention_mask'])\n",
    "tb_train_token_typ_ids = np.array(title_body_train_tokens['token_type_ids'])\n",
    " \n",
    "tb_cv_input_ids  = np.array(title_body_cv_tokens['input_ids'])\n",
    "tb_cv_attn_mask = np.array(title_body_cv_tokens['attention_mask'])\n",
    "tb_cv_token_typ_ids = np.array(title_body_cv_tokens['token_type_ids'])\n",
    "\n",
    "tb_train_input_ids.shape, tb_train_attn_mask.shape, tb_train_token_typ_ids.shape, tb_cv_input_ids.shape, tb_cv_attn_mask.shape, tb_cv_token_typ_ids.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 33
    },
    "colab_type": "code",
    "id": "kPt6fUmWAtad",
    "outputId": "a36f0324-f6bf-4435-b9a3-a6aac0b8127e"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((5349, 512), (5349, 512), (5349, 512), (730, 512), (730, 512), (730, 512))"
      ]
     },
     "execution_count": 12,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Tokenizing : answer\n",
    "ans_train_tokens = bert_tokenizer.batch_encode_plus(answer_train, max_length = 512, truncation = True, pad_to_max_length = True, return_tensors=\"tf\")\n",
    "ans_cv_tokens = bert_tokenizer.batch_encode_plus(answer_cv, max_length = 512, truncation = True, pad_to_max_length = True, return_tensors=\"tf\")\n",
    "# tokens = bert_tokenizer.batch_encode_plus(final_train, max_length = 50, truncation = True, pad_to_max_length = True)\n",
    " \n",
    "ans_train_input_ids  = np.array(ans_train_tokens['input_ids'])\n",
    "ans_train_attn_mask = np.array(ans_train_tokens['attention_mask'])\n",
    "ans_train_token_typ_ids = np.array(ans_train_tokens['token_type_ids'])\n",
    " \n",
    "ans_cv_input_ids  = np.array(ans_cv_tokens['input_ids'])\n",
    "ans_cv_attn_mask = np.array(ans_cv_tokens['attention_mask'])\n",
    "ans_cv_token_typ_ids = np.array(ans_cv_tokens['token_type_ids'])\n",
    "seq_len = ans_train_input_ids.shape[1]\n",
    " \n",
    "ans_train_input_ids.shape, ans_train_attn_mask.shape, ans_train_token_typ_ids.shape, ans_cv_input_ids.shape, ans_cv_attn_mask.shape, ans_cv_token_typ_ids.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 146,
     "referenced_widgets": [
      "2d594e25d5e84813a13228cfde6fa74c",
      "e4e3afa6697b44c1b46ce4723d57b2e0",
      "d61b2a1d66d541c38f3c47fba3f3e00e",
      "a2c7c9107f244cc5a0bb04ef18324e8f",
      "c347122704754d009fd6aee346165275",
      "181bc27df5024994a34545b243d21b58",
      "065a222b4ac14adb8a332e8b6f2cf2a8",
      "22cd761781ac42388dae310574759cd7",
      "a852197af681419d85139b39a428c5a2",
      "82519ce7196643a2b49840e00b183358",
      "01c5462b34a741deb508c4a6aa9c1308",
      "14453fe91077495fa7b767c2885f3d5a",
      "15b4a5793e464151844a3ac0c6cacad5",
      "fc3ee3f19d244a48bd1d145efd9d9264",
      "9f8c8e97c53c4238a65dc69672dbad81",
      "aed12c255e93476f900aa1e3cbdad25e"
     ]
    },
    "colab_type": "code",
    "id": "KPsBh01em1pF",
    "outputId": "06ed85b5-375c-4f48-e7ff-55ec37c45440"
   },
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "2d594e25d5e84813a13228cfde6fa74c",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=168.0), HTML(value='')))"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "(5349, 3072)\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "a852197af681419d85139b39a428c5a2",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=23.0), HTML(value='')))"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "(730, 3072)\n"
     ]
    }
   ],
   "source": [
    "# train : title_body BERT embedings\n",
    "batch_size = 32\n",
    "l = tb_train_input_ids.shape[0]\n",
    "\n",
    "lst1 = []\n",
    "for i in tqdm(range((l//batch_size)+1)):\n",
    "  x = bert_model([tb_train_input_ids[i*batch_size:(i+1)*batch_size], tb_train_attn_mask[i*batch_size:(i+1)*batch_size], tb_train_token_typ_ids[i*batch_size:(i+1)*batch_size]])[2]\n",
    "\n",
    "  # getting hidden states of last 4 layers (avg of each timrstep)\n",
    "  lst = [tf.reduce_mean(i, axis = 1) for i in x[-4:]]\n",
    "\n",
    "  # concat all 4 averaged hidden_states\n",
    "  conc = tf.concat(lst, axis = 1)\n",
    "  lst1.append(conc)\n",
    "tb_BERT_train = tf.concat(lst1, axis = 0)\n",
    "print(tb_BERT_train.shape)\n",
    "\n",
    "# cv : title_body BERT embedings\n",
    "batch_size = 32\n",
    "l = ans_cv_input_ids.shape[0]\n",
    "\n",
    "lst1 = []\n",
    "for i in tqdm(range((l//batch_size)+1)):\n",
    "  x = bert_model([tb_cv_input_ids[i*batch_size:(i+1)*batch_size], tb_cv_attn_mask[i*batch_size:(i+1)*batch_size], tb_cv_token_typ_ids[i*batch_size:(i+1)*batch_size]])[2]\n",
    "  # getting hidden states of last 4 layers (avg of each timrstep)\n",
    "  lst = [tf.reduce_mean(i, axis = 1) for i in x[-4:]]\n",
    "  # lst = [i[:, 0, :] for i in x[-4:]]\n",
    "  conc = tf.concat(lst, axis = 1)\n",
    "  lst1.append(conc)\n",
    "tb_BERT_cv = tf.concat(lst1, axis = 0)\n",
    "print(tb_BERT_cv.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 146,
     "referenced_widgets": [
      "4b28705d9d7c4369b98013ae1363dea3",
      "d732539b2ed94c09b2e38ba0727c222d",
      "81abdaa83ea747928118b9b8f47d0693",
      "26d2be5c8a654680a86a7272cf3b023b",
      "7a93bc995be5417998fe1a05a792cd9c",
      "58a0ed8938df49608ed05e277596eaad",
      "c101ca918f094a3cb940e384f89c558e",
      "bf49bac4c7494e39b5d742a6098446c4",
      "9f80ed5cf07e4c07a1a79e1e1af476ac",
      "69bdefc56b30498f943a24e96e99175d",
      "9474a20ee953478890830ce6d1d7f797",
      "fa6af85380124e24b5a4cf8421f168a3",
      "b0d45f5420774ba59f0fea05ea9788f3",
      "2cb7dcc84d30404789680ea54e8265d9",
      "c0c6772a00d942e08aa6ae47de8981ce",
      "07e5725f24544b1580a3ae1699973779"
     ]
    },
    "colab_type": "code",
    "id": "QrVUlZt6dPEh",
    "outputId": "102e5dc4-e325-4289-97dd-d17927b08ddc"
   },
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "4b28705d9d7c4369b98013ae1363dea3",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=168.0), HTML(value='')))"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "(5349, 3072)\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "9f80ed5cf07e4c07a1a79e1e1af476ac",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=23.0), HTML(value='')))"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "(730, 3072)\n"
     ]
    }
   ],
   "source": [
    "# train : ans BERT embedings\n",
    "batch_size = 32\n",
    "l = ans_train_input_ids.shape[0]\n",
    "\n",
    "lst1 = []\n",
    "for i in tqdm(range((l//batch_size)+1)):\n",
    "  x = bert_model([ans_train_input_ids[i*batch_size:(i+1)*batch_size], ans_train_attn_mask[i*batch_size:(i+1)*batch_size], ans_train_token_typ_ids[i*batch_size:(i+1)*batch_size]])[2]\n",
    "  # getting hidden states of last 4 layers (avg of each timrstep)\n",
    "  lst = [i[:, 0, :] for i in x[-4:]]\n",
    "  conc = tf.concat(lst, axis = 1)\n",
    "  lst1.append(conc)\n",
    "ans_BERT_train = tf.concat(lst1, axis = 0)\n",
    "print(ans_BERT_train.shape)\n",
    "\n",
    "# cv : ans BERT embedings\n",
    "batch_size = 32\n",
    "l = ans_cv_input_ids.shape[0]\n",
    "lst1 = []\n",
    "for i in tqdm(range((l//batch_size)+1)):\n",
    "  x = bert_model([ans_cv_input_ids[i*batch_size:(i+1)*batch_size], ans_cv_attn_mask[i*batch_size:(i+1)*batch_size], ans_cv_token_typ_ids[i*batch_size:(i+1)*batch_size]])[2]\n",
    "  # getting hidden states of last 4 layers (avg of each timrstep)\n",
    "  lst = [i[:, 0, :] for i in x[-4:]]\n",
    "  conc = tf.concat(lst, axis = 1)\n",
    "  lst1.append(conc)\n",
    "ans_BERT_cv = tf.concat(lst1, axis = 0)\n",
    "print(ans_BERT_cv.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "zgSO3OsO8J0Q"
   },
   "source": [
    "# 5.2. USE Embeddings"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "yhfrVy3JBjH1"
   },
   "outputs": [],
   "source": [
    "import tensorflow as tf\n",
    "\n",
    "import tensorflow_hub as hub\n",
    "from tensorflow.keras.layers import Input, Softmax, GRU, LSTM, RNN, Embedding, Dense, RepeatVector, TimeDistributed, Bidirectional, Concatenate\n",
    "from tensorflow.keras.models import Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 82
    },
    "colab_type": "code",
    "id": "n-aQpp3S8MAZ",
    "outputId": "92f0991c-3ea3-4247-f933-508c08b574c8"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:absl:Using /tmp/tfhub_modules to cache modules.\n",
      "INFO:absl:Downloading TF-Hub Module 'https://tfhub.dev/google/universal-sentence-encoder-large/5'.\n",
      "INFO:absl:Downloaded https://tfhub.dev/google/universal-sentence-encoder-large/5, Total size: 577.10MB\n",
      "INFO:absl:Downloaded TF-Hub Module 'https://tfhub.dev/google/universal-sentence-encoder-large/5'.\n"
     ]
    }
   ],
   "source": [
    "# loading USE model\n",
    "use_model = hub.load(\"https://tfhub.dev/google/universal-sentence-encoder-large/5\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 146,
     "referenced_widgets": [
      "fd8fc0fbf3bf4c749d2754026dec3024",
      "96a674ad9acc45f58713274d3ffe9cef",
      "dc2f041f43144db6a2c90ed08dca051b",
      "e3e8f8a326df488e97d4c7d5fd398890",
      "ffdf5fb1173b4c18ae7327756ec28887",
      "2b8a84b54afa476db34de197efe98b86",
      "8bee015c4f52479b8d7fba6898b643c0",
      "86074fd920aa4ebab4a9c338192b4166",
      "6e5bf6c7f6aa458ab32b47159153e4fa",
      "c0769515e0da4f53b88a3e472a7045e5",
      "5e886f6e73de42b5ab762216070e9993",
      "9ff4b21640d548a2b4e54af0793a0a92",
      "86d084ebdaaa4d94a0e23a6752448647",
      "715c3419ef9e4197bfc89fff7aee0e09",
      "75212dbf2fbb4e04b7a728b438f339ce",
      "1412b01aadf24634a7bbb50ecfc1fac9"
     ]
    },
    "colab_type": "code",
    "id": "_6P_HI9CDUIV",
    "outputId": "da114d37-46e8-4db1-ca60-80415217bd78"
   },
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "fd8fc0fbf3bf4c749d2754026dec3024",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=168.0), HTML(value='')))"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "(5349, 512)\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "6e5bf6c7f6aa458ab32b47159153e4fa",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=12.0), HTML(value='')))"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "(730, 512)\n"
     ]
    }
   ],
   "source": [
    "# title_body\n",
    "tb_USE_train = np.zeros((len(title_body_train), 512))\n",
    "batch_size = 32\n",
    "for i in tqdm(range(int(len(title_body_train)/batch_size)+1)):\n",
    "  tb_USE_train[i*batch_size : (i+1)*batch_size] = use_model(title_body_train[i*batch_size : (i+1)*batch_size]).numpy()\n",
    "print(tb_USE_train.shape)\n",
    "\n",
    "tb_USE_cv = np.zeros((len(title_body_cv), 512))\n",
    "batch_size = 64\n",
    "for i in tqdm(range(int(len(title_body_cv)/batch_size)+1)):\n",
    "  tb_USE_cv[i*batch_size : (i+1)*batch_size] = use_model(title_body_cv[i*batch_size : (i+1)*batch_size]).numpy()\n",
    "print(tb_USE_cv.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 146,
     "referenced_widgets": [
      "9a3a8a8e335c4c32983f541dd97fa462",
      "cd8e3aae7e98400a8244c1cd8e67a6f5",
      "d6ab42d1436e49e68fc84273f8942ec7",
      "ac743d8ca32b4b1289df3969c99b4861",
      "fd7687612b704c0f82548318d32146ef",
      "655a1b4b626f4bdd93a993f7147fd858",
      "b13d72305b1c47789c2c2a05228d2ea1",
      "40d0cdf589f64c0b8cc73059b7be6bad",
      "8788d63631334938aac76a16a84e5f62",
      "5e4a1fb1b41240428062b9106cb8f6d4",
      "2521fd1894dd4da2a0a638ce7aed888a",
      "400f42347a444c6486c7380b00dc54e1",
      "d1b0f94f3e6f469684ff1010635af586",
      "d835b46e4dba47599effe36cc0150ce9",
      "dbd9635795b848f5b654109f4a34ec3a",
      "70586fb1124f418f99f76fda730d401d"
     ]
    },
    "colab_type": "code",
    "id": "2ssPuJrJFXmt",
    "outputId": "76e3f38a-5672-4182-a857-6331d9760952"
   },
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "9a3a8a8e335c4c32983f541dd97fa462",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=168.0), HTML(value='')))"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "(5349, 512)\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "8788d63631334938aac76a16a84e5f62",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=23.0), HTML(value='')))"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "(730, 512)\n"
     ]
    }
   ],
   "source": [
    "# answer\n",
    "ans_USE_train = np.zeros((len(answer_train), 512))\n",
    "batch_size = 32\n",
    "for i in tqdm(range(int(len(answer_train)/batch_size)+1)):\n",
    "  ans_USE_train[i*batch_size : (i+1)*batch_size] = use_model(answer_train[i*batch_size : (i+1)*batch_size]).numpy()\n",
    "print(ans_USE_train.shape)\n",
    "\n",
    "ans_USE_cv = np.zeros((len(answer_cv), 512))\n",
    "batch_size = 32\n",
    "for i in tqdm(range(int(len(answer_cv)/batch_size)+1)):\n",
    "  ans_USE_cv[i*batch_size : (i+1)*batch_size] = use_model(answer_cv[i*batch_size : (i+1)*batch_size]).numpy()\n",
    "print(ans_USE_cv.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "nC7qxGeu9BLT"
   },
   "source": [
    "# 5.3. Sentence Embedding : RoBERTa"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 311,
     "referenced_widgets": [
      "94bd21b6c89d47488ef43dd045ff282a",
      "4353b94a332043f088936ab5c19fa9a5",
      "55b4169d866b426aa34752d3707126d0",
      "3214e2dce9314362b7b71a88791f04b9",
      "f58a2d64686a4750b1a3190067b26f6b",
      "711e4a4223d744feadd33967ad2a77ac",
      "5dde40b97bbf497981d9f4962ed48873",
      "742e8976c14f4ab2b50c54545e359f0a",
      "5892924a220043ada70ecc5435672697",
      "34c916e95c114420bcf488d97dcb1c1b",
      "d469efbe30b1486a9c61c67d97167e49",
      "87ac7ee2f67d447c9c2549fcb3b5ed43",
      "055725b181bb4d25b7a083b6cd7b373c",
      "87419ed4f82e4f9589c5563cd2b8f7b9",
      "2fe1e199b53240318444af597fde0cdb",
      "2f56058ab7e84240889cc2b2ca8899ee",
      "09bba54f402943c38fae70019553586c",
      "23e91246f4e04ad0bdd125bcdd098c3f",
      "4df465b390af4c0a84fc7841d6e35c5d",
      "f87b02922496412183d3ff676fa0318c",
      "4ade78eb35de46be8e9d4c5f0dde3c98",
      "defe087e9c494f10b2c00c39bfd449d6",
      "04b1246c315743e885e28770d34fd83f",
      "c30dcfac72a140d5b9317e382d9c7edc",
      "2bb40cb966f144ba87c4595d508ede35",
      "7d5cc91153244093b8571aca86edae46",
      "e02611fc3370422eab8db7952872060f",
      "4b53887915d340438b79eb0d501539b2",
      "7f812edd5c5e4451b043e4e85f9da4c1",
      "1dadb5f7f4bb4e71b0864c472d7712e4",
      "78a4ab65d80f4744addc3b54a5f8bdb7",
      "5f4768eb854641aea6f3f26714b285f2"
     ]
    },
    "colab_type": "code",
    "id": "iQ9PYY1A9BL4",
    "outputId": "f1042dd4-6ec1-42b9-806f-306b4a55a165"
   },
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "94bd21b6c89d47488ef43dd045ff282a",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, description='Downloading', max=481.0, style=ProgressStyle(description_…"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "5892924a220043ada70ecc5435672697",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, description='Downloading', max=898823.0, style=ProgressStyle(descripti…"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "09bba54f402943c38fae70019553586c",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, description='Downloading', max=456318.0, style=ProgressStyle(descripti…"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "2bb40cb966f144ba87c4595d508ede35",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, description='Downloading', max=657434796.0, style=ProgressStyle(descri…"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:transformers.modeling_tf_utils:Some weights of the model checkpoint at roberta-base were not used when initializing TFRobertaModel: ['lm_head']\n",
      "- This IS expected if you are initializing TFRobertaModel from the checkpoint of a model trained on another task or with another architecture (e.g. initializing a BertForSequenceClassification model from a BertForPretraining model).\n",
      "- This IS NOT expected if you are initializing TFRobertaModel from the checkpoint of a model that you expect to be exactly identical (initializing a BertForSequenceClassification model from a BertForSequenceClassification model).\n",
      "WARNING:transformers.modeling_tf_utils:All the weights of TFRobertaModel were initialized from the model checkpoint at roberta-base.\n",
      "If your task is similar to the task the model of the ckeckpoint was trained on, you can already use TFRobertaModel for predictions without further training.\n"
     ]
    }
   ],
   "source": [
    "import tensorflow as tf\n",
    "from transformers import RobertaConfig, RobertaTokenizer, TFRobertaModel\n",
    " \n",
    "# Load pretrained model/tokenizer\n",
    "config = RobertaConfig.from_pretrained('roberta-base', output_hidden_states=True)\n",
    "roberta_tokenizer = RobertaTokenizer.from_pretrained('roberta-base')\n",
    "roberta_model = TFRobertaModel.from_pretrained('roberta-base', config = config)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 50
    },
    "colab_type": "code",
    "id": "TLTW2MgUMDB2",
    "outputId": "57a3da99-6480-4a5d-ebc9-8cc2c934b4e5"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(5349, 512) (5349, 512) (730, 512) (730, 512)\n",
      "(5349, 300) (5349, 300) (730, 300) (730, 300)\n"
     ]
    }
   ],
   "source": [
    "# Tokenizing : title_body\n",
    "title_body_train_tokens = roberta_tokenizer.batch_encode_plus(title_body_train, max_length = 512, truncation = True, pad_to_max_length = True, return_tensors=\"tf\")\n",
    "title_body_cv_tokens = roberta_tokenizer.batch_encode_plus(title_body_cv, max_length = 512, truncation = True, pad_to_max_length = True, return_tensors=\"tf\")\n",
    "# tokens = bert_tokenizer.batch_encode_plus(final_train, max_length = 50, truncation = True, pad_to_max_length = True)\n",
    "\n",
    "tb_train_input_ids  = np.array(title_body_train_tokens['input_ids'])\n",
    "tb_train_attn_mask = np.array(title_body_train_tokens['attention_mask'])\n",
    " \n",
    "tb_cv_input_ids  = np.array(title_body_cv_tokens['input_ids'])\n",
    "tb_cv_attn_mask = np.array(title_body_cv_tokens['attention_mask'])\n",
    "\n",
    "print(tb_train_input_ids.shape, tb_train_attn_mask.shape, tb_cv_input_ids.shape, tb_cv_attn_mask.shape)\n",
    "\n",
    "# Tokenizing : answer\n",
    "ans_train_tokens = roberta_tokenizer.batch_encode_plus(answer_train, max_length = 300, truncation = True, pad_to_max_length = True, return_tensors=\"tf\")\n",
    "ans_cv_tokens = roberta_tokenizer.batch_encode_plus(answer_cv, max_length = 300, truncation = True, pad_to_max_length = True, return_tensors=\"tf\")\n",
    "# tokens = bert_tokenizer.batch_encode_plus(final_train, max_length = 50, truncation = True, pad_to_max_length = True)\n",
    " \n",
    "ans_train_input_ids  = np.array(ans_train_tokens['input_ids'])\n",
    "ans_train_attn_mask = np.array(ans_train_tokens['attention_mask'])\n",
    " \n",
    "ans_cv_input_ids  = np.array(ans_cv_tokens['input_ids'])\n",
    "ans_cv_attn_mask = np.array(ans_cv_tokens['attention_mask'])\n",
    "seq_len = ans_train_input_ids.shape[1]\n",
    " \n",
    "print(ans_train_input_ids.shape, ans_train_attn_mask.shape, ans_cv_input_ids.shape, ans_cv_attn_mask.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 146,
     "referenced_widgets": [
      "1318c81749d64b1ab4f58c712b4d9bb1",
      "d0cad404720740338f3472016dc53e2b",
      "c1a100c2dc334553a97f7fe99a66d5de",
      "4c4675d5429745f7a45bf52b79d136d0",
      "9e6ae79f17194fdfb08d517fa70110dd",
      "d9a0e85bd1274f988f7950abd23882b2",
      "9a11576090ab4ff592e2a10c2ee4b903",
      "07236e145d2844fd9b129dd3ee7e1718",
      "ce1cafb40f5845f1a622427efb9c91a6",
      "89d12b0a31c04a76bdcaf88fee1fd658",
      "d70dfff45a584559919f412e84d1e23d",
      "ff85249aaf8e428bbf3a51cd2d5dd87e",
      "8551901507524337b5b6828efbddd5e9",
      "f72eef90b33a42edb2a61185e6150881",
      "454e610c4cb745eba9607bc3c95574d1",
      "6307ebc36945404fb153514c702e0fa6"
     ]
    },
    "colab_type": "code",
    "id": "2524QqfEO6tw",
    "outputId": "24fe3d6e-aa6b-49db-e980-5614c769b87f"
   },
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "1318c81749d64b1ab4f58c712b4d9bb1",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=168.0), HTML(value='')))"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "(5349, 3072)\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "ce1cafb40f5845f1a622427efb9c91a6",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=23.0), HTML(value='')))"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "(730, 3072)\n"
     ]
    }
   ],
   "source": [
    "#  title_body : TRAIN RoBERTa embedings\n",
    "batch_size = 32\n",
    "l = tb_train_input_ids.shape[0]\n",
    "\n",
    "lst1 = []\n",
    "for i in tqdm(range((l//batch_size)+1)):\n",
    "  x = roberta_model([tb_train_input_ids[i*batch_size:(i+1)*batch_size], tb_train_attn_mask[i*batch_size:(i+1)*batch_size]])[2]\n",
    "\n",
    "  # getting hidden states of last 4 layers (avg of each timrstep)\n",
    "  lst = [i[:, 0, :] for i in x[-4:]]\n",
    "\n",
    "  # concat all 4 averaged hidden_states\n",
    "  conc = tf.concat(lst, axis = 1)\n",
    "  lst1.append(conc)\n",
    "tb_RoBERTa_train = tf.concat(lst1, axis = 0)\n",
    "print(tb_RoBERTa_train.shape)\n",
    "\n",
    "# title_body : CV RoBERTa embedings\n",
    "batch_size = 32\n",
    "l = tb_cv_input_ids.shape[0]\n",
    "\n",
    "lst1 = []\n",
    "for i in tqdm(range((l//batch_size)+1)):\n",
    "  x = roberta_model([tb_cv_input_ids[i*batch_size:(i+1)*batch_size], tb_cv_attn_mask[i*batch_size:(i+1)*batch_size]])[2]\n",
    "  # getting hidden states of last 4 layers (avg of each timrstep)\n",
    "  lst = [i[:, 0, :] for i in x[-4:]]\n",
    "  # lst = [i[:, 0, :] for i in x[-4:]]\n",
    "  conc = tf.concat(lst, axis = 1)\n",
    "  lst1.append(conc)\n",
    "tb_RoBERTa_cv = tf.concat(lst1, axis = 0)\n",
    "print(tb_RoBERTa_cv.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 146,
     "referenced_widgets": [
      "af2fd6f1a2f74a11b86ddaafa3ca7419",
      "dfb0d68084a04b549f1c06abe1edd6e7",
      "a84e42c2cf6f499cb78824fa2da60dd5",
      "1aaecf0b3b68444889b82106c452ed6f",
      "f88504163dd24c64843f76e5952b070a",
      "328beb7281604e378b867f29be0221fe",
      "efa1895edc5e4e619e4eb588b9127cae",
      "9e38d7ade99d4423b028f44161d045b0",
      "415069863abb454686fcd00a0971e555",
      "290b1788f1ac40d9b70348b5d0d5a1b0",
      "3f4b0e5ab79a4ffc81adc3c6966d7308",
      "67aafe24778f433486f1bf9345f81c6b",
      "88794d3fd37c43ffab4a6e275ee6e5db",
      "7325a03b04684879a54094ca3b246f90",
      "033a5e21fde447b592f6a655270e9704",
      "dc7bc3a76c7e40d990eccdddd2927a7a"
     ]
    },
    "colab_type": "code",
    "id": "hekWFr4PP7xj",
    "outputId": "2b5df6a6-1030-4c47-87ac-c1fe0f316fb8"
   },
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "af2fd6f1a2f74a11b86ddaafa3ca7419",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=168.0), HTML(value='')))"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "(5349, 3072)\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "415069863abb454686fcd00a0971e555",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=23.0), HTML(value='')))"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "(730, 3072)\n"
     ]
    }
   ],
   "source": [
    "#  ans : TRAIN RoBERTa embedings\n",
    "batch_size = 32\n",
    "l = ans_train_input_ids.shape[0]\n",
    "\n",
    "lst1 = []\n",
    "for i in tqdm(range((l//batch_size)+1)):\n",
    "  x = roberta_model([ans_train_input_ids[i*batch_size:(i+1)*batch_size], ans_train_attn_mask[i*batch_size:(i+1)*batch_size]])[2]\n",
    "  # getting hidden states of last 4 layers (avg of each timrstep)\n",
    "  lst = [i[:, 0, :] for i in x[-4:]]\n",
    "  conc = tf.concat(lst, axis = 1)\n",
    "  lst1.append(conc)\n",
    "ans_RoBERTa_train = tf.concat(lst1, axis = 0)\n",
    "print(ans_RoBERTa_train.shape)\n",
    "\n",
    "# ans : CV BERT embedings\n",
    "batch_size = 32\n",
    "l = ans_cv_input_ids.shape[0]\n",
    "lst1 = []\n",
    "for i in tqdm(range((l//batch_size)+1)):\n",
    "  x = roberta_model([ans_cv_input_ids[i*batch_size:(i+1)*batch_size], ans_cv_attn_mask[i*batch_size:(i+1)*batch_size]])[2]\n",
    "  # getting hidden states of last 4 layers (avg of each timrstep)\n",
    "  lst = [i[:, 0, :] for i in x[-4:]]\n",
    "  conc = tf.concat(lst, axis = 1)\n",
    "  lst1.append(conc)\n",
    "ans_RoBERTa_cv = tf.concat(lst1, axis = 0)\n",
    "print(ans_RoBERTa_cv.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "-G6cTx9aNzR1"
   },
   "source": [
    "# 5.4. Sentence Embedding : XLNet"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 262,
     "referenced_widgets": [
      "3f77953ca9df4f6aa35efabd78c538c5",
      "33b74b9e2a094c0a8dc4fed58a8a0efa",
      "969d400ea2484a99b54a1e14de00103c",
      "94f91259526f4c539da1f88701a8d244",
      "1f8550f8dc224d3fb2723b6df6bfc6f3",
      "2c3f2b79a0ee4d9993c070b679692a47",
      "430481b1567548bf995b1a21b8f1296b",
      "41d6d529dc604841907652d578b392bb",
      "be3d89e9334c44d1a6acded33e08cee3",
      "e635e2918eee490fad0cf98561a3397e",
      "77b09c5619064d58a2796023448389cd",
      "be836e272611471c9dae7b5770f48793",
      "88ee5b27999c40f399bb8ca4a756c069",
      "be720b8098fb4accaca6710416a15451",
      "21a3a032149a49528625e7d0441d9585",
      "0c43c5aab6084565a6e6c607c686dcc0",
      "fdec9f8d91464284a3ef937691258d43",
      "5ad382ea213f4eee8d8d35811975ffdc",
      "3a789e90935647a3860665cc8618dd5c",
      "30201ac2d0174912b4c30bd374ca878f",
      "7a8ae20bce954983a509c835d6b796f2",
      "078378bbad8a41f697ef44eacfb88037",
      "22c888cfe57a43f396145fadc60c40e0",
      "a8f61cd4acac4b8583314fef0faceab4"
     ]
    },
    "colab_type": "code",
    "id": "csjpdrBoNzSI",
    "outputId": "a71d309f-1543-4ac5-e776-51ade31bbe3b"
   },
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "3f77953ca9df4f6aa35efabd78c538c5",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, description='Downloading', max=760.0, style=ProgressStyle(description_…"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "be3d89e9334c44d1a6acded33e08cee3",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, description='Downloading', max=798011.0, style=ProgressStyle(descripti…"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "fdec9f8d91464284a3ef937691258d43",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, description='Downloading', max=565485600.0, style=ProgressStyle(descri…"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:transformers.modeling_tf_utils:Some weights of the model checkpoint at xlnet-base-cased were not used when initializing TFXLNetModel: ['lm_loss']\n",
      "- This IS expected if you are initializing TFXLNetModel from the checkpoint of a model trained on another task or with another architecture (e.g. initializing a BertForSequenceClassification model from a BertForPretraining model).\n",
      "- This IS NOT expected if you are initializing TFXLNetModel from the checkpoint of a model that you expect to be exactly identical (initializing a BertForSequenceClassification model from a BertForSequenceClassification model).\n",
      "WARNING:transformers.modeling_tf_utils:All the weights of TFXLNetModel were initialized from the model checkpoint at xlnet-base-cased.\n",
      "If your task is similar to the task the model of the ckeckpoint was trained on, you can already use TFXLNetModel for predictions without further training.\n"
     ]
    }
   ],
   "source": [
    "import tensorflow as tf\n",
    "from transformers import XLNetConfig, XLNetTokenizer, TFXLNetModel\n",
    " \n",
    "# Load pretrained model/tokenizer\n",
    "config = XLNetConfig.from_pretrained('xlnet-base-cased', output_hidden_states=True)\n",
    "xlnet_tokenizer = XLNetTokenizer.from_pretrained('xlnet-base-cased')\n",
    "xlnet_model = TFXLNetModel.from_pretrained('xlnet-base-cased', config = config)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 50
    },
    "colab_type": "code",
    "id": "XO40tarqNzSX",
    "outputId": "8801a080-b4f4-4c21-e523-cef79d16ef4b"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(5349, 512) (5349, 512) (5349, 512) (730, 512) (730, 512) (730, 512)\n",
      "(5349, 300) (5349, 300) (5349, 300) (730, 300) (730, 300) (730, 300)\n"
     ]
    }
   ],
   "source": [
    "# Tokenizing : title_body\n",
    "title_body_train_tokens = xlnet_tokenizer.batch_encode_plus(title_body_train, max_length = 512, truncation = True, pad_to_max_length = True, return_tensors=\"tf\")\n",
    "title_body_cv_tokens = xlnet_tokenizer.batch_encode_plus(title_body_cv, max_length = 512, truncation = True, pad_to_max_length = True, return_tensors=\"tf\")\n",
    "# tokens = bert_tokenizer.batch_encode_plus(final_train, max_length = 50, truncation = True, pad_to_max_length = True)\n",
    "\n",
    "tb_train_input_ids  = np.array(title_body_train_tokens['input_ids'])\n",
    "tb_train_attn_mask = np.array(title_body_train_tokens['attention_mask'])\n",
    "tb_train_token_typ_ids = np.array(title_body_train_tokens['token_type_ids'])\n",
    " \n",
    "tb_cv_input_ids  = np.array(title_body_cv_tokens['input_ids'])\n",
    "tb_cv_attn_mask = np.array(title_body_cv_tokens['attention_mask'])\n",
    "tb_cv_token_typ_ids = np.array(title_body_cv_tokens['token_type_ids'])\n",
    "\n",
    "print(tb_train_input_ids.shape, tb_train_attn_mask.shape, tb_train_token_typ_ids.shape, tb_cv_input_ids.shape, tb_cv_attn_mask.shape, tb_cv_token_typ_ids.shape)\n",
    "\n",
    "# Tokenizing : answer\n",
    "ans_train_tokens = xlnet_tokenizer.batch_encode_plus(answer_train, max_length = 300, truncation = True, pad_to_max_length = True, return_tensors=\"tf\")\n",
    "ans_cv_tokens = xlnet_tokenizer.batch_encode_plus(answer_cv, max_length = 300, truncation = True, pad_to_max_length = True, return_tensors=\"tf\")\n",
    "ans_train_input_ids  = np.array(ans_train_tokens['input_ids'])\n",
    "ans_train_attn_mask = np.array(ans_train_tokens['attention_mask'])\n",
    "ans_train_token_typ_ids = np.array(ans_train_tokens['token_type_ids'])\n",
    " \n",
    "ans_cv_input_ids  = np.array(ans_cv_tokens['input_ids'])\n",
    "ans_cv_attn_mask = np.array(ans_cv_tokens['attention_mask'])\n",
    "ans_cv_token_typ_ids = np.array(ans_cv_tokens['token_type_ids'])\n",
    "seq_len = ans_train_input_ids.shape[1]\n",
    " \n",
    "print(ans_train_input_ids.shape, ans_train_attn_mask.shape, ans_train_token_typ_ids.shape, ans_cv_input_ids.shape, ans_cv_attn_mask.shape, ans_cv_token_typ_ids.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 146,
     "referenced_widgets": [
      "1d6bd0fb338e49e1acbbf3e51d2f1d7b",
      "76f3b9de46da468b824e3f49d39b83f2",
      "cb49ecbea18648928c9a12827ef3b1f8",
      "d96e164e55214e1ebd3c9a7b7842f76e",
      "a7d35e5de40f4b979c35f9e9d9b71b57",
      "0375dbb6cfd14ac59a93b531a2a974f9",
      "44c33f30b8264061a402d84632661071",
      "38937e83918147fa8e4f246439550ec8",
      "b8caaa40985b4dcbb41fe671817c389a",
      "9b5b1fc1cb85429c86655f717a9b1bb9",
      "d4cf8d837172408ab631f98779eae95c",
      "63ed3558c4a14df085e09d74deb889b8",
      "1b4dbe64fa2b4bb39b2304110e8903d3",
      "5e3f496954a94d2c97ca68a2358e4843",
      "48a9d939ab494c18b2b744c1f8d668fb",
      "9c52a6ea14b04a22a349847aa33f08df"
     ]
    },
    "colab_type": "code",
    "id": "Xyag1bP3-1mz",
    "outputId": "6b3712e0-048a-40b4-fdf7-6fd88a17c205"
   },
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "1d6bd0fb338e49e1acbbf3e51d2f1d7b",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=168.0), HTML(value='')))"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "(5349, 3072)\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "b8caaa40985b4dcbb41fe671817c389a",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=23.0), HTML(value='')))"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "(730, 3072)\n"
     ]
    }
   ],
   "source": [
    "# train : title_body BERT embedings\n",
    "batch_size = 32\n",
    "l = tb_train_input_ids.shape[0]\n",
    "\n",
    "lst1 = []\n",
    "for i in tqdm(range((l//batch_size)+1)):\n",
    "  x = xlnet_model([tb_train_input_ids[i*batch_size:(i+1)*batch_size], tb_train_attn_mask[i*batch_size:(i+1)*batch_size]])[1]\n",
    "\n",
    "  # getting hidden states of last 4 layers (avg of each timrstep)\n",
    "  lst = [tf.reduce_mean(i, axis = 1) for i in x[-4:]]\n",
    "\n",
    "  # concat all 4 averaged hidden_states\n",
    "  conc = tf.concat(lst, axis = 1)\n",
    "  lst1.append(conc)\n",
    "tb_XLNet_train = tf.concat(lst1, axis = 0)\n",
    "print(tb_XLNet_train.shape)\n",
    "\n",
    "# cv : title_body BERT embedings\n",
    "batch_size = 32\n",
    "l = tb_cv_input_ids.shape[0]\n",
    "\n",
    "lst1 = []\n",
    "for i in tqdm(range((l//batch_size)+1)):\n",
    "  x = xlnet_model([tb_cv_input_ids[i*batch_size:(i+1)*batch_size], tb_cv_attn_mask[i*batch_size:(i+1)*batch_size]])[1]\n",
    "  # getting hidden states of last 4 layers (avg of each timrstep)\n",
    "  lst = [tf.reduce_mean(i, axis = 1) for i in x[-4:]]\n",
    "  # lst = [i[:, 0, :] for i in x[-4:]]\n",
    "  conc = tf.concat(lst, axis = 1)\n",
    "  lst1.append(conc)\n",
    "tb_XLNet_cv = tf.concat(lst1, axis = 0)\n",
    "print(tb_XLNet_cv.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 146,
     "referenced_widgets": [
      "173231b0a0d34ab29897e44b2ad461df",
      "cb45c7c08787447bb99ee370fd8ba716",
      "749b7e4362964aa09aea33315ba6f14b",
      "f5b71821098948598e73e61b025db362",
      "baa37fa43b5a4f2987f09277249659df",
      "03371146acc44a65981ee3fcbc91377e",
      "3a4fa61568e542a581a987d6b03c544c",
      "0a8d1371d7744db18edf19fa41f85555",
      "62d36481dba74df180179d104988c71b",
      "cfa56e32bb79478393524302d4a931bc",
      "64501f87ef5e475497291fae69e5d0bc",
      "34027535904b4b4d9c4bbd19c58306a3",
      "37342f5fe94448989775ccfd2ead99ae",
      "52723b099ee8460091e9ff5db914d0e3",
      "f196815db4574391a3c27e0bf5e9da8b",
      "4be40858751347649165a1f1b054321c"
     ]
    },
    "colab_type": "code",
    "id": "sQuyfTNd-1nE",
    "outputId": "106baa80-6f87-4983-86ea-2b1a15c254e3"
   },
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "173231b0a0d34ab29897e44b2ad461df",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=168.0), HTML(value='')))"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "(5349, 3072)\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "62d36481dba74df180179d104988c71b",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=23.0), HTML(value='')))"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "(730, 3072)\n"
     ]
    }
   ],
   "source": [
    "# train : ans BERT embedings\n",
    "batch_size = 32\n",
    "l = ans_train_input_ids.shape[0]\n",
    "\n",
    "lst1 = []\n",
    "for i in tqdm(range((l//batch_size)+1)):\n",
    "  x = xlnet_model([ans_train_input_ids[i*batch_size:(i+1)*batch_size], ans_train_attn_mask[i*batch_size:(i+1)*batch_size]])[1]\n",
    "  # getting hidden states of last 4 layers (avg of each timrstep)\n",
    "  lst = [tf.reduce_mean(i, axis = 1) for i in x[-4:]]\n",
    "  conc = tf.concat(lst, axis = 1)\n",
    "  lst1.append(conc)\n",
    "ans_XLNet_train = tf.concat(lst1, axis = 0)\n",
    "print(ans_XLNet_train.shape)\n",
    "\n",
    "# cv : ans BERT embedings\n",
    "batch_size = 32\n",
    "l = ans_cv_input_ids.shape[0]\n",
    "lst1 = []\n",
    "for i in tqdm(range((l//batch_size)+1)):\n",
    "  x = xlnet_model([ans_cv_input_ids[i*batch_size:(i+1)*batch_size], ans_cv_attn_mask[i*batch_size:(i+1)*batch_size]])[1]\n",
    "  # getting hidden states of last 4 layers (avg of each timrstep)\n",
    "  lst = [tf.reduce_mean(i, axis = 1) for i in x[-4:]]\n",
    "  conc = tf.concat(lst, axis = 1)\n",
    "  lst1.append(conc)\n",
    "ans_XLNet_cv = tf.concat(lst1, axis = 0)\n",
    "print(ans_XLNet_cv.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "rSMZ4FjpL4zU"
   },
   "outputs": [],
   "source": [
    "np.savez_compressed('/content/drive/My Drive/AAIC Course/Personal case study 2  - Google Quest/bert_outputs', tb_BERT_train = tb_BERT_train.numpy(), tb_BERT_cv = tb_BERT_cv.numpy(), \n",
    "                    ans_BERT_train = ans_BERT_train.numpy(), ans_BERT_cv = ans_BERT_cv.numpy())\n",
    "\n",
    "np.savez_compressed('/content/drive/My Drive/AAIC Course/Personal case study 2  - Google Quest/use_outputs', tb_USE_train = tb_USE_train, tb_USE_cv = tb_USE_cv, \n",
    "                    ans_USE_train = ans_USE_train, ans_USE_cv = ans_USE_cv)\n",
    "\n",
    "np.savez_compressed('/content/drive/My Drive/AAIC Course/Personal case study 2  - Google Quest/roberta_outputs', tb_RoBERTa_train = tb_RoBERTa_train.numpy(), tb_RoBERTa_cv = tb_RoBERTa_cv.numpy(), \n",
    "                    ans_RoBERTa_train = ans_RoBERTa_train.numpy(), ans_RoBERTa_cv = ans_RoBERTa_cv.numpy())\n",
    "\n",
    "np.savez_compressed('/content/drive/My Drive/AAIC Course/Personal case study 2  - Google Quest/xlnet_outputs', tb_XLNet_train = tb_XLNet_train.numpy(), tb_XLNet_cv = tb_XLNet_cv.numpy(), \n",
    "                    ans_XLNet_train = ans_XLNet_train.numpy(), ans_XLNet_cv = ans_XLNet_cv.numpy())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "o4__knI8ltqP"
   },
   "source": [
    "# 5. Final Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "86mCipknlxDb"
   },
   "outputs": [],
   "source": [
    "import tensorflow as tf\n",
    "tf.compat.v1.enable_eager_execution()\n",
    "from tensorflow.keras.layers import Input, Dense, Dropout, Concatenate, BatchNormalization\n",
    "from tensorflow.keras.models import Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 50
    },
    "colab_type": "code",
    "id": "BIf8Oz04l7g6",
    "outputId": "8251f5a1-eb2b-441a-a28b-3fae8a33b56e"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 3.49 s, sys: 360 ms, total: 3.85 s\n",
      "Wall time: 11.2 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "tb_BERT_train = np.load('/content/drive/My Drive/AAIC Course/Personal case study 2  - Google Quest/bert_outputs.npz')['tb_BERT_train']\n",
    "tb_BERT_cv = np.load('/content/drive/My Drive/AAIC Course/Personal case study 2  - Google Quest/bert_outputs.npz')['tb_BERT_cv']\n",
    "ans_BERT_train = np.load('/content/drive/My Drive/AAIC Course/Personal case study 2  - Google Quest/bert_outputs.npz')['ans_BERT_train']\n",
    "ans_BERT_cv = np.load('/content/drive/My Drive/AAIC Course/Personal case study 2  - Google Quest/bert_outputs.npz')['ans_BERT_cv']\n",
    "\n",
    "tb_USE_train = np.load('/content/drive/My Drive/AAIC Course/Personal case study 2  - Google Quest/use_outputs.npz')['tb_USE_train']\n",
    "tb_USE_cv = np.load('/content/drive/My Drive/AAIC Course/Personal case study 2  - Google Quest/use_outputs.npz')['tb_USE_cv']\n",
    "ans_USE_train = np.load('/content/drive/My Drive/AAIC Course/Personal case study 2  - Google Quest/use_outputs.npz')['ans_USE_train']\n",
    "ans_USE_cv = np.load('/content/drive/My Drive/AAIC Course/Personal case study 2  - Google Quest/use_outputs.npz')['ans_USE_cv']\n",
    "\n",
    "tb_RoBERTa_train = np.load('/content/drive/My Drive/AAIC Course/Personal case study 2  - Google Quest/roberta_outputs.npz')['tb_RoBERTa_train']\n",
    "tb_RoBERTa_cv = np.load('/content/drive/My Drive/AAIC Course/Personal case study 2  - Google Quest/roberta_outputs.npz')['tb_RoBERTa_cv']\n",
    "ans_RoBERTa_train = np.load('/content/drive/My Drive/AAIC Course/Personal case study 2  - Google Quest/roberta_outputs.npz')['ans_RoBERTa_train']\n",
    "ans_RoBERTa_cv = np.load('/content/drive/My Drive/AAIC Course/Personal case study 2  - Google Quest/roberta_outputs.npz')['ans_RoBERTa_cv']\n",
    "\n",
    "tb_XLNet_train = np.load('/content/drive/My Drive/AAIC Course/Personal case study 2  - Google Quest/xlnet_outputs.npz')['tb_XLNet_train']\n",
    "tb_XLNet_cv = np.load('/content/drive/My Drive/AAIC Course/Personal case study 2  - Google Quest/xlnet_outputs.npz')['tb_XLNet_cv']\n",
    "ans_XLNet_train = np.load('/content/drive/My Drive/AAIC Course/Personal case study 2  - Google Quest/xlnet_outputs.npz')['ans_XLNet_train']\n",
    "ans_XLNet_cv = np.load('/content/drive/My Drive/AAIC Course/Personal case study 2  - Google Quest/xlnet_outputs.npz')['ans_XLNet_cv']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "tgRsF52ymu9h"
   },
   "outputs": [],
   "source": [
    "class block(tf.keras.layers.Layer):\n",
    "  def __init__(self, d1_rate = 0.2, dense_1_units = 1024, d2_rate = 0.2, dense_2_units = 512, d3_rate = 0.1, seed = 42):\n",
    "    super().__init__()\n",
    "    self.d1_rate = d1_rate\n",
    "    self.dense_1_units = dense_1_units\n",
    "    self.d2_rate = d2_rate\n",
    "    self.dense_2_units = dense_2_units\n",
    "    self.d3_rate = d3_rate\n",
    "    self.seed = seed\n",
    "\n",
    "  def build(self, input_shape):\n",
    "    # self.inp = Input(name = 'ans_bert_out', shape = (768*4,), dtype = 'float32')\n",
    "    self.dropout_1 =  Dropout(rate =self.d1_rate, seed = self.seed)\n",
    "\n",
    "    self.dense_1 = Dense(units = self.dense_1_units, activation = 'relu', kernel_initializer = tf.keras.initializers.he_normal(seed = self.seed))\n",
    "    self.dropout_2 = Dropout(rate = self.d2_rate, seed = self.seed)\n",
    "\n",
    "    self.dense_2 = Dense(units = self.dense_2_units, activation = 'relu', kernel_initializer = tf.keras.initializers.he_normal(seed = self.seed))\n",
    "    self.dropout_3 = Dropout(rate = self.d3_rate, seed = self.seed)\n",
    "\n",
    "  def call(self, X):\n",
    "    dropout_1 = self.dropout_1(X)\n",
    "\n",
    "    dense_1 = self.dense_1(dropout_1)\n",
    "    dropout_2 = self.dropout_2(dense_1)\n",
    "\n",
    "    dense_2 = self.dense_2(dropout_2)\n",
    "    dropout_3 = self.dropout_3(dense_2)\n",
    "    return dropout_3"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "colab_type": "code",
    "id": "5WTGGroGtKgh",
    "outputId": "00c27c7b-4631-4a2f-f2e5-6e6aee737643"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"functional_1\"\n",
      "__________________________________________________________________________________________________\n",
      "Layer (type)                    Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      "tb_use_out (InputLayer)         [(None, 512)]        0                                            \n",
      "__________________________________________________________________________________________________\n",
      "tb_roberta_out (InputLayer)     [(None, 3072)]       0                                            \n",
      "__________________________________________________________________________________________________\n",
      "block_1 (block)                 (None, 512)          1050112     tb_use_out[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "block_2 (block)                 (None, 512)          3671552     tb_roberta_out[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "dense_1 (Dense)                 (None, 21)           10773       block_1[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "dense_2 (Dense)                 (None, 21)           10773       block_2[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "ans_use_out (InputLayer)        [(None, 512)]        0                                            \n",
      "__________________________________________________________________________________________________\n",
      "ans_roberta_out (InputLayer)    [(None, 3072)]       0                                            \n",
      "__________________________________________________________________________________________________\n",
      "tb_bert_out (InputLayer)        [(None, 3072)]       0                                            \n",
      "__________________________________________________________________________________________________\n",
      "tb_xlnet_out (InputLayer)       [(None, 3072)]       0                                            \n",
      "__________________________________________________________________________________________________\n",
      "ans_bert_out (InputLayer)       [(None, 3072)]       0                                            \n",
      "__________________________________________________________________________________________________\n",
      "tf_op_layer_concat (TensorFlowO [(None, 533)]        0           dense_1[0][0]                    \n",
      "                                                                 ans_use_out[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "tf_op_layer_concat_1 (TensorFlo [(None, 3093)]       0           dense_2[0][0]                    \n",
      "                                                                 ans_roberta_out[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "ans_xlnet_out (InputLayer)      [(None, 3072)]       0                                            \n",
      "__________________________________________________________________________________________________\n",
      "block (block)                   (None, 512)          3671552     tb_bert_out[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "block_3 (block)                 (None, 512)          3671552     tb_xlnet_out[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "block_4 (block)                 (None, 512)          7342592     ans_bert_out[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "block_5 (block)                 (None, 512)          1071616     tf_op_layer_concat[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block_6 (block)                 (None, 512)          3693056     tf_op_layer_concat_1[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "block_7 (block)                 (None, 512)          3671552     ans_xlnet_out[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "dense (Dense)                   (None, 21)           10773       block[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "dense_3 (Dense)                 (None, 21)           10773       block_3[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "dense_4 (Dense)                 (None, 9)            4617        block_4[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "dense_5 (Dense)                 (None, 9)            4617        block_5[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "dense_6 (Dense)                 (None, 9)            4617        block_6[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "dense_7 (Dense)                 (None, 9)            4617        block_7[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "tf_op_layer_Mean/input (TensorF [(4, None, 21)]      0           dense[0][0]                      \n",
      "                                                                 dense_1[0][0]                    \n",
      "                                                                 dense_2[0][0]                    \n",
      "                                                                 dense_3[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "tf_op_layer_Mean_1/input (Tenso [(4, None, 9)]       0           dense_4[0][0]                    \n",
      "                                                                 dense_5[0][0]                    \n",
      "                                                                 dense_6[0][0]                    \n",
      "                                                                 dense_7[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "tf_op_layer_Mean (TensorFlowOpL [(None, 21)]         0           tf_op_layer_Mean/input[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "tf_op_layer_Mean_1 (TensorFlowO [(None, 9)]          0           tf_op_layer_Mean_1/input[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "concatenate (Concatenate)       (None, 30)           0           tf_op_layer_Mean[0][0]           \n",
      "                                                                 tf_op_layer_Mean_1[0][0]         \n",
      "==================================================================================================\n",
      "Total params: 27,905,144\n",
      "Trainable params: 27,905,144\n",
      "Non-trainable params: 0\n",
      "__________________________________________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "tf.keras.backend.clear_session()\n",
    "seed = 42\n",
    "\n",
    "# *-----------------title_body-----------------*\n",
    "tb_bert_input = Input(name = 'tb_bert_out', shape = (768*4,), dtype = 'float32')\n",
    "tb_bert_block = block(d1_rate = 0.2, dense_1_units = 1024, d2_rate = 0.2, dense_2_units = 512, d3_rate = 0.1, seed = 42)(tb_bert_input)\n",
    "tb_bert_out = Dense(units = 21, activation = 'sigmoid', kernel_initializer = tf.keras.initializers.he_normal(seed = seed))(tb_bert_block)\n",
    "\n",
    "tb_use_input = Input(name = 'tb_use_out', shape = (512,), dtype = 'float32')\n",
    "tb_use_block = block(d1_rate = 0.2, dense_1_units = 1024, d2_rate = 0.2, dense_2_units = 512, d3_rate = 0.2, seed = 42)(tb_use_input)\n",
    "tb_use_out = Dense(units = 21, activation = 'sigmoid', kernel_initializer = tf.keras.initializers.he_normal(seed = seed))(tb_use_block)\n",
    "\n",
    "tb_roberta_input = Input(name = 'tb_roberta_out', shape = (768*4,), dtype = 'float32')\n",
    "tb_roberta_block = block(d1_rate = 0.2, dense_1_units = 1024, d2_rate = 0.2, dense_2_units = 512, d3_rate = 0.1, seed = 42)(tb_roberta_input)\n",
    "tb_roberta_out = Dense(units = 21, activation = 'sigmoid', kernel_initializer = tf.keras.initializers.he_normal(seed = seed))(tb_roberta_block)\n",
    "\n",
    "tb_xlnet_input = Input(name = 'tb_xlnet_out', shape = (768*4,), dtype = 'float32')\n",
    "tb_xlnet_block = block(d1_rate = 0.2, dense_1_units = 1024, d2_rate = 0.2, dense_2_units = 512, d3_rate = 0.1, seed = 42)(tb_xlnet_input)\n",
    "tb_xlnet_out = Dense(units = 21, activation = 'sigmoid', kernel_initializer = tf.keras.initializers.he_normal(seed = seed))(tb_xlnet_block)\n",
    "\n",
    "tb_out = tf.reduce_mean([tb_bert_out ,tb_use_out ,tb_roberta_out ,tb_xlnet_out], axis = 0)\n",
    "\n",
    "# *----------------- answer -----------------*\n",
    "ans_bert_input = Input(name = 'ans_bert_out', shape = (768*4,), dtype = 'float32')\n",
    "ans_bert_block = block(d1_rate = 0.2, dense_1_units = 2048, d2_rate = 0.2, dense_2_units = 512, d3_rate = 0.1, seed = 42)(ans_bert_input)\n",
    "ans_bert_out = Dense(units = 9, activation = 'sigmoid', kernel_initializer = tf.keras.initializers.he_normal(seed = seed))(ans_bert_block)\n",
    "\n",
    "ans_use_input = Input(name = 'ans_use_out', shape = (512,), dtype = 'float32')\n",
    "ans_use_concat = tf.concat([tb_use_out, ans_use_input], axis = -1) \n",
    "ans_use_block = block(d1_rate = 0.2, dense_1_units = 1024, d2_rate = 0.2, dense_2_units = 512, d3_rate = 0.2, seed = 42)(ans_use_concat)\n",
    "ans_use_out = Dense(units = 9, activation = 'sigmoid', kernel_initializer = tf.keras.initializers.he_normal(seed = seed))(ans_use_block)\n",
    "\n",
    "ans_roberta_input = Input(name = 'ans_roberta_out', shape = (768*4,), dtype = 'float32')\n",
    "ans_roberta_concat = tf.concat([tb_roberta_out, ans_roberta_input], axis = -1) \n",
    "ans_roberta_block = block(d1_rate = 0.2, dense_1_units = 1024, d2_rate = 0.2, dense_2_units = 512, d3_rate = 0.1, seed = 42)(ans_roberta_concat)\n",
    "ans_roberta_out = Dense(units = 9, activation = 'sigmoid', kernel_initializer = tf.keras.initializers.he_normal(seed = seed))(ans_roberta_block)\n",
    "\n",
    "ans_xlnet_input = Input(name = 'ans_xlnet_out', shape = (768*4,), dtype = 'float32')\n",
    "ans_xlnet_block = block(d1_rate = 0.2, dense_1_units = 1024, d2_rate = 0.2, dense_2_units = 512, d3_rate = 0.1,  seed = 42)(ans_xlnet_input)\n",
    "ans_xlnet_out = Dense(units = 9, activation = 'sigmoid', kernel_initializer = tf.keras.initializers.he_normal(seed = seed))(ans_xlnet_block)\n",
    "\n",
    "ans_out = tf.reduce_mean([ans_bert_out ,ans_use_out ,ans_roberta_out ,ans_xlnet_out], axis = 0)\n",
    "\n",
    "# *----------------- concat -----------------*\n",
    "out =  Concatenate(axis = -1)([tb_out, ans_out])\n",
    "\n",
    "model = Model(inputs = [tb_bert_input, tb_use_input, tb_roberta_input, tb_xlnet_input, ans_bert_input, ans_use_input, ans_roberta_input, ans_xlnet_input], outputs = out)\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "n2x0Chu0w9pG"
   },
   "outputs": [],
   "source": [
    "# post processing : binning\n",
    "def return_bins(arr):\n",
    "  val = np.unique(arr)\n",
    "  bins = []\n",
    "  for i in range(len(val)):\n",
    "    if i > 0:\n",
    "      bins.append((val[i-1] + val[i])/2)\n",
    "  return bins\n",
    "  \n",
    "unique_val_30 = [np.unique(df_train[tar_features].values[:, i]) for i in range(30)]\n",
    "bins_30 = [return_bins(df_train[tar_features].values[:, i]) for i in range(30)]\n",
    " \n",
    "def binned_out(y_pred):\n",
    "  col = y_pred.shape[1]\n",
    "  final_pred = np.zeros(y_pred.shape)\n",
    "  for i in range(col):\n",
    "    idx = np.digitize(y_pred[:, i], bins_30[i])\n",
    "    final_pred[:, i] = unique_val_30[i][idx]\n",
    "  return final_pred"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 33
    },
    "colab_type": "code",
    "id": "9skvAyafxAoF",
    "outputId": "e369a7bc-b4b5-44ec-9996-d4fcb922904f"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:`write_grads` will be ignored in TensorFlow 2.0 for the `TensorBoard` Callback.\n"
     ]
    }
   ],
   "source": [
    "# Defining callbacks\n",
    "!rm -r '/content/saved models'\n",
    "!rm -r '/content/logs'\n",
    "!mkdir '/content/saved models'\n",
    "!mkdir '/content/logs/'\n",
    " \n",
    "# tensorboard callback\n",
    "import datetime\n",
    "log_dir=\"logs/\" + datetime.datetime.now().strftime(\"%Y-%m-%d %H_%M_%S\")\n",
    "tensorboard_callback = tf.keras.callbacks.TensorBoard(log_dir = log_dir, histogram_freq=1, write_graph=True, write_grads=True)\n",
    " \n",
    "# spearman function\n",
    "from scipy.stats import pearsonr, spearmanr\n",
    "def compute_spearman(y_true, y_pred, final_pred):\n",
    "  col = y_true.shape[1]\n",
    "  lst = []\n",
    "  for i in range(col):\n",
    "    # p = round(spearmanr(y_true[:, i], y_pred[:, i])[0], 5)\n",
    "    p = round(spearmanr(y_true[:, i], final_pred[:, i])[0], 5)\n",
    "    p = round(p, 5)\n",
    "    # if np.isnan(p):\n",
    "    #   p = round(spearmanr(y_true[:, i], y_pred[:, i])[0], 5)\n",
    "    lst.append(p)\n",
    "  return np.array(lst), round(np.nanmean(lst), 5)\n",
    " \n",
    "# Custom spearman metric\n",
    "class print_spearman(tf.keras.callbacks.Callback):\n",
    "    def __init__(self, train_data, validation_data):\n",
    "        super(tf.keras.callbacks.Callback, self).__init__()\n",
    "        self.x, self.y = train_data\n",
    "        self.val_x, self.val_y = validation_data\n",
    "    \n",
    "    def on_train_begin(self, logs={}):\n",
    "        self.all_feat_spearman = []\n",
    "        self.spearman_dict = {'train_spearman' :[], 'val_spearman' :[]}\n",
    " \n",
    "    def on_epoch_end(self, epoch, logs={}):\n",
    "        self.epoch = epoch\n",
    "        # 1. Test_set evaluation\n",
    "        print('\\nspearman :')\n",
    "        y_pred = self.model.predict(x = self.x)\n",
    "        y_pred_val = self.model.predict(x = self.val_x)\n",
    " \n",
    "        final_pred = binned_out(y_pred)\n",
    "        final_pred_val = binned_out(y_pred_val)\n",
    " \n",
    "        train_spear_lst, train_spearman = compute_spearman(self.y, y_pred, final_pred)\n",
    "        val_spear_lst, val_spearman = compute_spearman(self.val_y, y_pred_val, final_pred_val)\n",
    " \n",
    "        self.all_feat_spearman.append({'train_spearman' : train_spear_lst, 'val_spearman' : val_spear_lst})\n",
    " \n",
    "        self.spearman_dict['train_spearman'].append(train_spearman)\n",
    "        self.spearman_dict['val_spearman'].append(val_spearman)\n",
    "        prev_epoch_lr  = tf.keras.backend.eval(self.model.optimizer.lr)\n",
    "        print(\"train_spearman : {} | val_spearman : {} | Learning_Rate : {}\".format(train_spearman, val_spearman, round(prev_epoch_lr, 6)))\n",
    "        print('train_spear_lst : ', train_spear_lst, '\\n' 'val_spear_lst :', val_spear_lst)\n",
    " \n",
    "reduce_lr = tf.keras.callbacks.ReduceLROnPlateau(monitor= 'val_loss', factor=np.sqrt(0.1), patience=7, verbose=1)\n",
    " \n",
    "checkpt = tf.keras.callbacks.ModelCheckpoint('/content/saved models/weights-{epoch:03d}-{val_loss:.5f}.hdf5', monitor='val_loss', verbose=1, save_best_only=False, save_weights_only=True)\n",
    " \n",
    "print_spearman_fn = print_spearman(train_data = ([tb_BERT_train, tb_USE_train, tb_RoBERTa_train, tb_XLNet_train, ans_BERT_train, ans_USE_train, ans_RoBERTa_train, ans_XLNet_train],\n",
    "                                                 y_train.values),\n",
    "                                 validation_data = ([tb_BERT_cv, tb_USE_cv, tb_RoBERTa_cv, tb_XLNet_cv, ans_BERT_cv, ans_USE_cv, ans_RoBERTa_cv, ans_XLNet_cv],\n",
    "                                                    y_cv.values))\n",
    "callbacks = [print_spearman_fn, reduce_lr, checkpt, tensorboard_callback]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "colab_type": "code",
    "id": "Ym_4iKfj8Yd5",
    "outputId": "d350f6cb-6f50-4b53-89db-377435b40f86"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:Model failed to serialize as JSON. Ignoring... Layer block has arguments in `__init__` and therefore must override `get_config`.\n",
      "Epoch 1/60\n",
      " 2/84 [..............................] - ETA: 7s - loss: 0.7399 - root_mean_squared_error: 0.4389WARNING:tensorflow:Callbacks method `on_train_batch_end` is slow compared to the batch time (batch time: 0.0174s vs `on_train_batch_end` time: 0.1618s). Check your callbacks.\n",
      "84/84 [==============================] - ETA: 0s - loss: 0.4501 - root_mean_squared_error: 0.2596\n",
      "spearman :\n",
      "train_spearman : 0.26985 | val_spearman : 0.30619 | Learning_Rate : 7.999999797903001e-05\n",
      "train_spear_lst :  [    nan 0.35633     nan 0.07715 0.01595 0.23984 0.24237 0.38826 0.16534\n",
      "     nan 0.13429 0.20635     nan     nan     nan     nan 0.58921 0.06609\n",
      " 0.17864     nan 0.33282     nan 0.22662     nan     nan     nan 0.60715\n",
      "     nan 0.49122     nan] \n",
      "val_spear_lst : [    nan 0.37459     nan 0.11295     nan 0.30036 0.28224 0.42985 0.16297\n",
      "     nan 0.1287  0.19789     nan     nan     nan     nan 0.58208 0.12031\n",
      " 0.21832     nan 0.37638     nan 0.20278     nan     nan     nan 0.61749\n",
      "     nan 0.48588     nan]\n",
      "\n",
      "Epoch 00001: saving model to /content/saved models/weights.-001-0.41327\n",
      "84/84 [==============================] - 6s 72ms/step - loss: 0.4501 - root_mean_squared_error: 0.2596 - val_loss: 0.4133 - val_root_mean_squared_error: 0.2319\n",
      "Epoch 2/60\n",
      "84/84 [==============================] - ETA: 0s - loss: 0.4107 - root_mean_squared_error: 0.2333\n",
      "spearman :\n",
      "train_spearman : 0.33118 | val_spearman : 0.32434 | Learning_Rate : 7.999999797903001e-05\n",
      "train_spear_lst :  [    nan 0.36343 0.31536 0.13747 0.13881 0.32856 0.30205 0.4595  0.2964\n",
      "     nan 0.18111 0.26572     nan     nan     nan     nan 0.62974 0.15488\n",
      " 0.33993     nan 0.38385     nan 0.16619     nan     nan     nan 0.64911\n",
      "     nan 0.51791     nan] \n",
      "val_spear_lst : [    nan 0.38098 0.26435 0.14196 0.1123  0.37351 0.2861  0.485   0.29906\n",
      "     nan 0.20023 0.2514      nan     nan     nan     nan 0.63554 0.06868\n",
      " 0.36945     nan 0.38981     nan 0.12238     nan     nan     nan 0.62354\n",
      "     nan 0.50955     nan]\n",
      "\n",
      "Epoch 00002: saving model to /content/saved models/weights.-002-0.40150\n",
      "84/84 [==============================] - 5s 58ms/step - loss: 0.4107 - root_mean_squared_error: 0.2333 - val_loss: 0.4015 - val_root_mean_squared_error: 0.2263\n",
      "Epoch 3/60\n",
      "84/84 [==============================] - ETA: 0s - loss: 0.4030 - root_mean_squared_error: 0.2286\n",
      "spearman :\n",
      "train_spearman : 0.3277 | val_spearman : 0.33977 | Learning_Rate : 7.999999797903001e-05\n",
      "train_spear_lst :  [ 0.07827  0.38524  0.40103  0.14817  0.17492  0.33837  0.30765  0.47677\n",
      "  0.32793      nan  0.27251  0.26818      nan      nan      nan  0.27773\n",
      "  0.65632  0.18438  0.41853      nan  0.40977      nan  0.22604      nan\n",
      " -0.00659      nan  0.6671       nan  0.54176      nan] \n",
      "val_spear_lst : [0.13653 0.41298 0.4622  0.1468  0.18801 0.36688 0.26546 0.49012 0.32193\n",
      "     nan 0.26615 0.28996     nan     nan     nan 0.11755 0.65654 0.09462\n",
      " 0.40138     nan 0.42261     nan 0.21742     nan     nan     nan 0.66313\n",
      "     nan 0.53535     nan]\n",
      "\n",
      "Epoch 00003: saving model to /content/saved models/weights.-003-0.39565\n",
      "84/84 [==============================] - 5s 59ms/step - loss: 0.4030 - root_mean_squared_error: 0.2286 - val_loss: 0.3956 - val_root_mean_squared_error: 0.2222\n",
      "Epoch 4/60\n",
      "81/84 [===========================>..] - ETA: 0s - loss: 0.3972 - root_mean_squared_error: 0.2245\n",
      "spearman :\n",
      "train_spearman : 0.33474 | val_spearman : 0.32972 | Learning_Rate : 7.999999797903001e-05\n",
      "train_spear_lst :  [0.18977 0.40568 0.44503 0.17502 0.22944 0.38247 0.31621 0.48054 0.35532\n",
      "     nan 0.31461 0.39481     nan     nan 0.38887 0.3836  0.67127 0.24288\n",
      " 0.44459     nan 0.42158     nan 0.24735 0.07103 0.0922  0.08845 0.69059\n",
      " 0.04618 0.55629     nan] \n",
      "val_spear_lst : [0.1739  0.43876 0.44233 0.18815 0.23175 0.40021 0.30521 0.50032 0.33671\n",
      "     nan 0.34677 0.41826     nan     nan 0.3691  0.29456 0.67654 0.15336\n",
      " 0.40508     nan 0.42466     nan 0.24274 0.07179 0.08015 0.13368 0.66653\n",
      " 0.07227 0.54051     nan]\n",
      "\n",
      "Epoch 00004: saving model to /content/saved models/weights.-004-0.39159\n",
      "84/84 [==============================] - 5s 65ms/step - loss: 0.3972 - root_mean_squared_error: 0.2245 - val_loss: 0.3916 - val_root_mean_squared_error: 0.2194\n",
      "Epoch 5/60\n",
      "81/84 [===========================>..] - ETA: 0s - loss: 0.3918 - root_mean_squared_error: 0.2206\n",
      "spearman :\n",
      "train_spearman : 0.33867 | val_spearman : 0.33682 | Learning_Rate : 7.999999797903001e-05\n",
      "train_spear_lst :  [0.22159 0.42005 0.44549 0.19057 0.25735 0.39883 0.31839 0.48668 0.35636\n",
      "     nan 0.33222 0.4662      nan     nan 0.43746 0.42256 0.67397 0.25238\n",
      " 0.4723      nan 0.43503 0.01502 0.31799 0.07223 0.12479 0.22962 0.70352\n",
      " 0.13086 0.58934 0.03451] \n",
      "val_spear_lst : [0.19689 0.44523 0.40426 0.17095 0.24496 0.40235 0.30624 0.50021 0.35824\n",
      "     nan 0.36418 0.47938     nan     nan 0.44081 0.27032 0.67998 0.17661\n",
      " 0.44013     nan 0.43892     nan 0.28175 0.04817 0.06341 0.26624 0.68272\n",
      " 0.10067 0.56548 0.09232]\n",
      "\n",
      "Epoch 00005: saving model to /content/saved models/weights.-005-0.38762\n",
      "84/84 [==============================] - 5s 63ms/step - loss: 0.3921 - root_mean_squared_error: 0.2207 - val_loss: 0.3876 - val_root_mean_squared_error: 0.2165\n",
      "Epoch 6/60\n",
      "81/84 [===========================>..] - ETA: 0s - loss: 0.3883 - root_mean_squared_error: 0.2180\n",
      "spearman :\n",
      "train_spearman : 0.35058 | val_spearman : 0.35375 | Learning_Rate : 7.999999797903001e-05\n",
      "train_spear_lst :  [0.21519 0.42182 0.47418 0.19405 0.3015  0.41864 0.3223  0.49494 0.36705\n",
      "     nan 0.36677 0.4896  0.10311     nan 0.4651  0.45099 0.69537 0.27788\n",
      " 0.5197      nan 0.4433  0.06982 0.30779 0.13092 0.15102 0.2559  0.72451\n",
      " 0.17917 0.59378 0.03127] \n",
      "val_spear_lst : [0.2088  0.43756 0.44994 0.18749 0.2916  0.42782 0.31601 0.50764 0.38159\n",
      "     nan 0.35178 0.4997      nan     nan 0.45157 0.30028 0.68429 0.18305\n",
      " 0.5022      nan 0.43468 0.05355 0.26234 0.07896 0.09582 0.30385 0.70784\n",
      " 0.15669 0.56873     nan]\n",
      "\n",
      "Epoch 00006: saving model to /content/saved models/weights.-006-0.38418\n",
      "84/84 [==============================] - 5s 59ms/step - loss: 0.3882 - root_mean_squared_error: 0.2178 - val_loss: 0.3842 - val_root_mean_squared_error: 0.2135\n",
      "Epoch 7/60\n",
      "84/84 [==============================] - ETA: 0s - loss: 0.3849 - root_mean_squared_error: 0.2152\n",
      "spearman :\n",
      "train_spearman : 0.37516 | val_spearman : 0.36296 | Learning_Rate : 7.999999797903001e-05\n",
      "train_spear_lst :  [0.20895 0.43629 0.46832 0.18388 0.30845 0.43451 0.32443 0.49545 0.41244\n",
      "     nan 0.37379 0.58005 0.3174      nan 0.51395 0.46828 0.70506 0.29691\n",
      " 0.53372     nan 0.45669 0.11849 0.31154 0.1447  0.17047 0.28383 0.73408\n",
      " 0.2127  0.61212 0.02273] \n",
      "val_spear_lst : [0.18893 0.45024 0.47361 0.15824 0.30615 0.44581 0.31437 0.51464 0.41546\n",
      "     nan 0.37196 0.52832 0.1584      nan 0.39731 0.35993 0.69571 0.21735\n",
      " 0.50602     nan 0.44264 0.14495 0.28266 0.09489 0.18075 0.3305  0.72109\n",
      " 0.14688 0.59012     nan]\n",
      "\n",
      "Epoch 00007: saving model to /content/saved models/weights.-007-0.38194\n",
      "84/84 [==============================] - 5s 58ms/step - loss: 0.3849 - root_mean_squared_error: 0.2152 - val_loss: 0.3819 - val_root_mean_squared_error: 0.2116\n",
      "Epoch 8/60\n",
      "81/84 [===========================>..] - ETA: 0s - loss: 0.3823 - root_mean_squared_error: 0.2129\n",
      "spearman :\n",
      "train_spearman : 0.39392 | val_spearman : 0.38766 | Learning_Rate : 7.999999797903001e-05\n",
      "train_spear_lst :  [0.23532 0.44317 0.48151 0.20664 0.3174  0.45116 0.32672 0.49959 0.42015\n",
      "     nan 0.39307 0.60247 0.37565     nan 0.52981 0.49102 0.72848 0.31953\n",
      " 0.56394     nan 0.46259 0.18707 0.34425 0.15522 0.16779 0.29165 0.74595\n",
      " 0.26031 0.61805 0.01737] \n",
      "val_spear_lst : [0.236   0.46437 0.4366  0.19618 0.31384 0.46941 0.32725 0.51435 0.4322\n",
      "     nan 0.37158 0.55975 0.31856     nan 0.51004 0.36532 0.72544 0.20753\n",
      " 0.55073     nan 0.45999 0.1754  0.3356  0.09903 0.13268 0.34118 0.73308\n",
      " 0.19972 0.60341     nan]\n",
      "\n",
      "Epoch 00008: saving model to /content/saved models/weights.-008-0.37977\n",
      "84/84 [==============================] - 7s 83ms/step - loss: 0.3822 - root_mean_squared_error: 0.2131 - val_loss: 0.3798 - val_root_mean_squared_error: 0.2102\n",
      "Epoch 9/60\n",
      "84/84 [==============================] - ETA: 0s - loss: 0.3799 - root_mean_squared_error: 0.2112\n",
      "spearman :\n",
      "train_spearman : 0.40441 | val_spearman : 0.39735 | Learning_Rate : 7.999999797903001e-05\n",
      "train_spear_lst :  [0.26314 0.46079 0.49097 0.21196 0.3383  0.4504  0.32166 0.50442 0.43413\n",
      "     nan 0.40757 0.63802 0.4269      nan 0.55889 0.48988 0.71977 0.32802\n",
      " 0.55225     nan 0.47308 0.21121 0.32437 0.15587 0.18615 0.31637 0.7497\n",
      " 0.26226 0.62996 0.01306] \n",
      "val_spear_lst : [0.22787 0.46859 0.42493 0.1905  0.34029 0.46716 0.30946 0.51599 0.4552\n",
      "     nan 0.39176 0.57538 0.44982     nan 0.45397 0.4155  0.72569 0.21111\n",
      " 0.52934     nan 0.46106 0.21723 0.34762 0.11683 0.15389 0.34534 0.73451\n",
      " 0.18946 0.61252     nan]\n",
      "\n",
      "Epoch 00009: saving model to /content/saved models/weights.-009-0.37820\n",
      "84/84 [==============================] - 5s 61ms/step - loss: 0.3799 - root_mean_squared_error: 0.2112 - val_loss: 0.3782 - val_root_mean_squared_error: 0.2089\n",
      "Epoch 10/60\n",
      "84/84 [==============================] - ETA: 0s - loss: 0.3781 - root_mean_squared_error: 0.2099\n",
      "spearman :\n",
      "train_spearman : 0.4215 | val_spearman : 0.39014 | Learning_Rate : 7.999999797903001e-05\n",
      "train_spear_lst :  [0.28538 0.46651 0.49498 0.24368 0.34234 0.47079 0.32904 0.50937 0.45871\n",
      "     nan 0.42688 0.63755 0.48269     nan 0.58014 0.50728 0.73293 0.33873\n",
      " 0.58284     nan 0.47833 0.23672 0.37129 0.17139 0.18167 0.33036 0.75589\n",
      " 0.28115 0.63934 0.04464] \n",
      "val_spear_lst : [0.23339 0.4864  0.41213 0.2252  0.31478 0.47833 0.31632 0.52425 0.45025\n",
      "     nan 0.41811 0.59036 0.49107     nan 0.52686 0.40342 0.72163 0.22125\n",
      " 0.54954     nan 0.46218 0.21375 0.35985 0.11602 0.07363 0.36342 0.74209\n",
      " 0.17404 0.61215 0.05323]\n",
      "\n",
      "Epoch 00010: saving model to /content/saved models/weights.-010-0.37659\n",
      "84/84 [==============================] - 6s 77ms/step - loss: 0.3781 - root_mean_squared_error: 0.2099 - val_loss: 0.3766 - val_root_mean_squared_error: 0.2076\n",
      "Epoch 11/60\n",
      "83/84 [============================>.] - ETA: 0s - loss: 0.3759 - root_mean_squared_error: 0.2082\n",
      "spearman :\n",
      "train_spearman : 0.42993 | val_spearman : 0.4029 | Learning_Rate : 7.999999797903001e-05\n",
      "train_spear_lst :  [0.31447 0.4805  0.50114 0.24016 0.34321 0.47015 0.3316  0.5111  0.46242\n",
      "     nan 0.41521 0.65989 0.47809     nan 0.58394 0.52371 0.7358  0.33678\n",
      " 0.5951      nan 0.47739 0.25187 0.33998 0.19391 0.20596 0.33955 0.75203\n",
      " 0.28019 0.64585 0.138  ] \n",
      "val_spear_lst : [0.25792 0.49997 0.45816 0.25795 0.32026 0.47873 0.31612 0.52353 0.45012\n",
      "     nan 0.39118 0.60093 0.49107     nan 0.53581 0.40562 0.72647 0.21975\n",
      " 0.57756     nan 0.46628 0.27451 0.32997 0.12097 0.15433 0.37051 0.72318\n",
      " 0.15831 0.61627 0.15282]\n",
      "\n",
      "Epoch 00011: saving model to /content/saved models/weights.-011-0.37673\n",
      "84/84 [==============================] - 8s 99ms/step - loss: 0.3759 - root_mean_squared_error: 0.2083 - val_loss: 0.3767 - val_root_mean_squared_error: 0.2075\n",
      "Epoch 12/60\n",
      "81/84 [===========================>..] - ETA: 0s - loss: 0.3744 - root_mean_squared_error: 0.2071\n",
      "spearman :\n",
      "train_spearman : 0.44109 | val_spearman : 0.40933 | Learning_Rate : 7.999999797903001e-05\n",
      "train_spear_lst :  [0.32711 0.48441 0.50429 0.25849 0.34992 0.48415 0.3306  0.51676 0.46757\n",
      "     nan 0.43098 0.66046 0.48207     nan 0.62253 0.53181 0.75443 0.34205\n",
      " 0.59689     nan 0.48539 0.26027 0.35265 0.20675 0.22091 0.3489  0.76545\n",
      " 0.2988  0.66322 0.16244] \n",
      "val_spear_lst : [0.25468 0.49809 0.42553 0.25325 0.31145 0.4798  0.32071 0.53034 0.46509\n",
      "     nan 0.39193 0.61678 0.45885     nan 0.54997 0.46094 0.74861 0.25444\n",
      " 0.5872      nan 0.47359 0.28243 0.30675 0.13415 0.15918 0.3654  0.73339\n",
      " 0.20976 0.61095 0.16878]\n",
      "\n",
      "Epoch 00012: saving model to /content/saved models/weights.-012-0.37475\n",
      "84/84 [==============================] - 11s 134ms/step - loss: 0.3745 - root_mean_squared_error: 0.2071 - val_loss: 0.3748 - val_root_mean_squared_error: 0.2061\n",
      "Epoch 13/60\n",
      "84/84 [==============================] - ETA: 0s - loss: 0.3734 - root_mean_squared_error: 0.2062\n",
      "spearman :\n",
      "train_spearman : 0.44915 | val_spearman : 0.40964 | Learning_Rate : 7.999999797903001e-05\n",
      "train_spear_lst :  [0.33741 0.49109 0.52445 0.25303 0.37593 0.47889 0.33574 0.51394 0.48375\n",
      "     nan 0.44569 0.6739  0.50178     nan 0.62474 0.54096 0.75314 0.34248\n",
      " 0.60048     nan 0.48876 0.27597 0.37709 0.19946 0.24051 0.3622  0.76881\n",
      " 0.30169 0.65938 0.17591] \n",
      "val_spear_lst : [0.24822 0.50314 0.43669 0.23801 0.31762 0.47746 0.33049 0.53108 0.47108\n",
      "     nan 0.41951 0.61406 0.43137     nan 0.52589 0.47055 0.74725 0.27148\n",
      " 0.59001     nan 0.46514 0.24927 0.35299 0.06432 0.16162 0.35699 0.73345\n",
      " 0.24749 0.62011 0.18511]\n",
      "\n",
      "Epoch 00013: saving model to /content/saved models/weights.-013-0.37454\n",
      "84/84 [==============================] - 7s 80ms/step - loss: 0.3734 - root_mean_squared_error: 0.2062 - val_loss: 0.3745 - val_root_mean_squared_error: 0.2056\n",
      "Epoch 14/60\n",
      "81/84 [===========================>..] - ETA: 0s - loss: 0.3719 - root_mean_squared_error: 0.2051\n",
      "spearman :\n",
      "train_spearman : 0.45577 | val_spearman : 0.42091 | Learning_Rate : 7.999999797903001e-05\n",
      "train_spear_lst :  [0.37012 0.49696 0.51972 0.26899 0.36768 0.48758 0.33796 0.51814 0.48315\n",
      "     nan 0.45558 0.67727 0.50661     nan 0.62742 0.5744  0.75954 0.3441\n",
      " 0.60675     nan 0.49124 0.29531 0.37332 0.22078 0.23013 0.35309 0.77007\n",
      " 0.3144  0.67043 0.18509] \n",
      "val_spear_lst : [0.27975 0.51132 0.40629 0.27392 0.31181 0.48223 0.31389 0.547   0.46794\n",
      "     nan 0.43775 0.61547 0.47747     nan 0.53249 0.50776 0.75296 0.24449\n",
      " 0.56932     nan 0.45559 0.28225 0.34777 0.15636 0.21566 0.38315 0.73662\n",
      " 0.23276 0.61764 0.20498]\n",
      "\n",
      "Epoch 00014: saving model to /content/saved models/weights.-014-0.37411\n",
      "84/84 [==============================] - 10s 115ms/step - loss: 0.3720 - root_mean_squared_error: 0.2052 - val_loss: 0.3741 - val_root_mean_squared_error: 0.2052\n",
      "Epoch 15/60\n",
      "84/84 [==============================] - ETA: 0s - loss: 0.3708 - root_mean_squared_error: 0.2042\n",
      "spearman :\n",
      "train_spearman : 0.46318 | val_spearman : 0.42636 | Learning_Rate : 7.999999797903001e-05\n",
      "train_spear_lst :  [0.36632 0.50473 0.51512 0.28066 0.37084 0.4963  0.34076 0.52535 0.49972\n",
      "     nan 0.45589 0.67962 0.51329     nan 0.63679 0.58802 0.75176 0.34717\n",
      " 0.60965     nan 0.50482 0.30039 0.4025  0.23741 0.24345 0.36941 0.77439\n",
      " 0.29751 0.67942 0.21457] \n",
      "val_spear_lst : [0.27998 0.52276 0.45507 0.2756  0.32296 0.48177 0.343   0.55078 0.48336\n",
      "     nan 0.443   0.62975 0.46876     nan 0.5129  0.52532 0.73723 0.24307\n",
      " 0.57951     nan 0.47116 0.25761 0.41317 0.13435 0.20684 0.38716 0.74013\n",
      " 0.15696 0.63077 0.25862]\n",
      "\n",
      "Epoch 00015: saving model to /content/saved models/weights.-015-0.37290\n",
      "84/84 [==============================] - 11s 137ms/step - loss: 0.3708 - root_mean_squared_error: 0.2042 - val_loss: 0.3729 - val_root_mean_squared_error: 0.2043\n",
      "Epoch 16/60\n",
      "82/84 [============================>.] - ETA: 0s - loss: 0.3697 - root_mean_squared_error: 0.2034\n",
      "spearman :\n",
      "train_spearman : 0.47194 | val_spearman : 0.42638 | Learning_Rate : 7.999999797903001e-05\n",
      "train_spear_lst :  [0.37246 0.51388 0.5369  0.28678 0.39106 0.49042 0.35324 0.5286  0.50743\n",
      "     nan 0.47559 0.69642 0.52228     nan 0.65749 0.58812 0.75502 0.36346\n",
      " 0.62595     nan 0.50755 0.31778 0.40786 0.2301  0.25404 0.38348 0.77411\n",
      " 0.31856 0.68283 0.20105] \n",
      "val_spear_lst : [0.28002 0.52871 0.45236 0.28835 0.30992 0.47839 0.3441  0.54844 0.46703\n",
      "     nan 0.42168 0.62661 0.48938     nan 0.52864 0.55751 0.74158 0.25853\n",
      " 0.58494     nan 0.45614 0.25534 0.41072 0.15828 0.2198  0.3796  0.73857\n",
      " 0.14878 0.6129  0.22607]\n",
      "\n",
      "Epoch 00016: saving model to /content/saved models/weights.-016-0.37292\n",
      "84/84 [==============================] - 5s 60ms/step - loss: 0.3696 - root_mean_squared_error: 0.2033 - val_loss: 0.3729 - val_root_mean_squared_error: 0.2043\n",
      "Epoch 17/60\n",
      "84/84 [==============================] - ETA: 0s - loss: 0.3688 - root_mean_squared_error: 0.2025\n",
      "spearman :\n",
      "train_spearman : 0.4764 | val_spearman : 0.42263 | Learning_Rate : 7.999999797903001e-05\n",
      "train_spear_lst :  [0.37609 0.52189 0.53974 0.29479 0.39728 0.49557 0.36301 0.52549 0.51469\n",
      "     nan 0.47141 0.70245 0.53415     nan 0.65189 0.59716 0.76174 0.35596\n",
      " 0.62309     nan 0.51126 0.32429 0.39133 0.25001 0.242   0.39352 0.77844\n",
      " 0.32603 0.68516 0.23424] \n",
      "val_spear_lst : [0.27913 0.53729 0.4443  0.27456 0.32607 0.46976 0.33896 0.53386 0.47848\n",
      "     nan 0.42429 0.64105 0.47733     nan 0.51628 0.56037 0.74195 0.25828\n",
      " 0.57525     nan 0.46538 0.22596 0.34219 0.13082 0.2063  0.39975 0.74131\n",
      " 0.14857 0.61888 0.25471]\n",
      "\n",
      "Epoch 00017: saving model to /content/saved models/weights.-017-0.37333\n",
      "84/84 [==============================] - 7s 83ms/step - loss: 0.3688 - root_mean_squared_error: 0.2025 - val_loss: 0.3733 - val_root_mean_squared_error: 0.2044\n",
      "Epoch 18/60\n",
      "81/84 [===========================>..] - ETA: 0s - loss: 0.3678 - root_mean_squared_error: 0.2022\n",
      "spearman :\n",
      "train_spearman : 0.47876 | val_spearman : 0.43089 | Learning_Rate : 7.999999797903001e-05\n",
      "train_spear_lst :  [0.37371 0.51654 0.53895 0.30077 0.38166 0.50387 0.36221 0.5263  0.50981\n",
      "     nan 0.47099 0.70306 0.54856     nan 0.66639 0.60487 0.7642  0.36673\n",
      " 0.62767     nan 0.50792 0.32238 0.40457 0.25033 0.24332 0.38731 0.77669\n",
      " 0.34425 0.68341 0.24007] \n",
      "val_spear_lst : [0.2835  0.52968 0.44086 0.27743 0.33194 0.48978 0.35414 0.54751 0.47992\n",
      "     nan 0.43309 0.64025 0.51066     nan 0.53309 0.58044 0.74913 0.25252\n",
      " 0.59976     nan 0.47054 0.26755 0.32625 0.14337 0.19753 0.39949 0.74434\n",
      " 0.20444 0.61982 0.22699]\n",
      "\n",
      "Epoch 00018: saving model to /content/saved models/weights.-018-0.37211\n",
      "84/84 [==============================] - 10s 122ms/step - loss: 0.3680 - root_mean_squared_error: 0.2020 - val_loss: 0.3721 - val_root_mean_squared_error: 0.2036\n",
      "Epoch 19/60\n",
      "83/84 [============================>.] - ETA: 0s - loss: 0.3669 - root_mean_squared_error: 0.2011\n",
      "spearman :\n",
      "train_spearman : 0.48661 | val_spearman : 0.42693 | Learning_Rate : 7.999999797903001e-05\n",
      "train_spear_lst :  [0.38849 0.52184 0.55662 0.30968 0.40311 0.50808 0.37127 0.53583 0.52361\n",
      "     nan 0.46737 0.69658 0.54517     nan 0.67273 0.61181 0.76815 0.36702\n",
      " 0.62997     nan 0.51939 0.33688 0.43516 0.25785 0.28833 0.40297 0.77579\n",
      " 0.31067 0.69318 0.24093] \n",
      "val_spear_lst : [0.29327 0.53764 0.4341  0.24705 0.32563 0.47996 0.34359 0.54859 0.47314\n",
      "     nan 0.41467 0.64464 0.4861      nan 0.53516 0.56022 0.75131 0.2627\n",
      " 0.59068     nan 0.46869 0.25423 0.38291 0.14658 0.18692 0.37591 0.74379\n",
      " 0.14675 0.63309 0.2599 ]\n",
      "\n",
      "Epoch 00019: saving model to /content/saved models/weights.-019-0.37234\n",
      "84/84 [==============================] - 7s 79ms/step - loss: 0.3669 - root_mean_squared_error: 0.2011 - val_loss: 0.3723 - val_root_mean_squared_error: 0.2036\n",
      "Epoch 20/60\n",
      "83/84 [============================>.] - ETA: 0s - loss: 0.3663 - root_mean_squared_error: 0.2004\n",
      "spearman :\n",
      "train_spearman : 0.46991 | val_spearman : 0.42659 | Learning_Rate : 7.999999797903001e-05\n",
      "train_spear_lst :  [0.38804 0.52312 0.55774 0.3118  0.40266 0.50491 0.36737 0.53749 0.53034\n",
      " 0.13085 0.49665 0.7044  0.54344 0.21334 0.6717  0.61707 0.76922 0.36684\n",
      " 0.63605     nan 0.51914 0.31937 0.44084 0.2625  0.30687 0.41885 0.78422\n",
      " 0.35945 0.69056 0.25258] \n",
      "val_spear_lst : [0.28896 0.52991 0.41867 0.3     0.32385 0.48486 0.36102 0.53506 0.47079\n",
      "     nan 0.41626 0.63157 0.46598     nan 0.56507 0.59835 0.75364 0.256\n",
      " 0.5909      nan 0.48193 0.25876 0.42807 0.06756 0.11781 0.37329 0.74265\n",
      " 0.18132 0.62693 0.24877]\n",
      "\n",
      "Epoch 00020: saving model to /content/saved models/weights.-020-0.37204\n",
      "84/84 [==============================] - 10s 115ms/step - loss: 0.3661 - root_mean_squared_error: 0.2004 - val_loss: 0.3720 - val_root_mean_squared_error: 0.2033\n",
      "Epoch 21/60\n",
      "81/84 [===========================>..] - ETA: 0s - loss: 0.3655 - root_mean_squared_error: 0.2000\n",
      "spearman :\n",
      "train_spearman : 0.47413 | val_spearman : 0.43158 | Learning_Rate : 7.999999797903001e-05\n",
      "train_spear_lst :  [0.38924 0.53691 0.56202 0.31922 0.41057 0.5096  0.37039 0.5459  0.53971\n",
      " 0.13085 0.50226 0.71464 0.55396 0.21334 0.67407 0.61586 0.76769 0.37984\n",
      " 0.63009     nan 0.52654 0.34847 0.43113 0.29014 0.27493 0.41741 0.78285\n",
      " 0.35775 0.70118 0.25333] \n",
      "val_spear_lst : [0.29079 0.548   0.43957 0.27905 0.3371  0.47896 0.36611 0.54916 0.4921\n",
      "     nan 0.4285  0.65024 0.4861      nan 0.52557 0.56786 0.75507 0.23671\n",
      " 0.58        nan 0.4813  0.24159 0.3718  0.11261 0.21259 0.3869  0.74276\n",
      " 0.21817 0.62876 0.24542]\n",
      "\n",
      "Epoch 00021: saving model to /content/saved models/weights.-021-0.37174\n",
      "84/84 [==============================] - 11s 135ms/step - loss: 0.3652 - root_mean_squared_error: 0.1997 - val_loss: 0.3717 - val_root_mean_squared_error: 0.2031\n",
      "Epoch 22/60\n",
      "84/84 [==============================] - ETA: 0s - loss: 0.3645 - root_mean_squared_error: 0.1993\n",
      "spearman :\n",
      "train_spearman : 0.47808 | val_spearman : 0.43506 | Learning_Rate : 7.999999797903001e-05\n",
      "train_spear_lst :  [0.40078 0.54112 0.56346 0.32482 0.41363 0.50164 0.37405 0.53347 0.53585\n",
      " 0.13085 0.49879 0.72105 0.55264 0.21334 0.68893 0.61945 0.76989 0.37352\n",
      " 0.63907     nan 0.53412 0.33717 0.44193 0.29184 0.30313 0.42779 0.78953\n",
      " 0.37112 0.70378 0.2677 ] \n",
      "val_spear_lst : [0.31474 0.54655 0.42401 0.30435 0.33237 0.45769 0.37053 0.53902 0.4939\n",
      "     nan 0.44329 0.64813 0.4861      nan 0.55247 0.59931 0.75269 0.26022\n",
      " 0.58996     nan 0.48014 0.27365 0.40666 0.11651 0.12193 0.38798 0.73777\n",
      " 0.20722 0.63273 0.26669]\n",
      "\n",
      "Epoch 00022: saving model to /content/saved models/weights.-022-0.37142\n",
      "84/84 [==============================] - 5s 60ms/step - loss: 0.3645 - root_mean_squared_error: 0.1993 - val_loss: 0.3714 - val_root_mean_squared_error: 0.2027\n",
      "Epoch 23/60\n",
      "82/84 [============================>.] - ETA: 0s - loss: 0.3635 - root_mean_squared_error: 0.1985\n",
      "spearman :\n",
      "train_spearman : 0.48163 | val_spearman : 0.43186 | Learning_Rate : 7.999999797903001e-05\n",
      "train_spear_lst :  [0.39691 0.54041 0.57384 0.32379 0.41541 0.51269 0.38094 0.5505  0.52795\n",
      " 0.13085 0.501   0.71522 0.55612 0.21334 0.69064 0.62443 0.77403 0.38004\n",
      " 0.64323     nan 0.53442 0.35974 0.44012 0.29754 0.29978 0.43271 0.79339\n",
      " 0.37338 0.70767 0.27727] \n",
      "val_spear_lst : [0.29266 0.54951 0.43326 0.30228 0.34096 0.47106 0.36356 0.54895 0.48954\n",
      "     nan 0.42756 0.63749 0.4861      nan 0.5454  0.56374 0.75827 0.26064\n",
      " 0.59718     nan 0.47806 0.25366 0.38717 0.11631 0.18169 0.37174 0.74393\n",
      " 0.18566 0.62761 0.24611]\n",
      "\n",
      "Epoch 00023: saving model to /content/saved models/weights.-023-0.37116\n",
      "84/84 [==============================] - 7s 86ms/step - loss: 0.3638 - root_mean_squared_error: 0.1987 - val_loss: 0.3712 - val_root_mean_squared_error: 0.2026\n",
      "Epoch 24/60\n",
      "82/84 [============================>.] - ETA: 0s - loss: 0.3628 - root_mean_squared_error: 0.1980\n",
      "spearman :\n",
      "train_spearman : 0.48491 | val_spearman : 0.43292 | Learning_Rate : 7.999999797903001e-05\n",
      "train_spear_lst :  [0.39775 0.54588 0.58637 0.32812 0.42029 0.50204 0.37903 0.53731 0.55379\n",
      " 0.13085 0.49914 0.72682 0.55673 0.21334 0.69505 0.63691 0.77578 0.37994\n",
      " 0.638       nan 0.53654 0.36774 0.45082 0.28421 0.30495 0.43777 0.78963\n",
      " 0.3949  0.7178  0.27485] \n",
      "val_spear_lst : [0.28909 0.54533 0.42623 0.2878  0.33898 0.44633 0.37171 0.54514 0.51324\n",
      "     nan 0.41052 0.64222 0.49741     nan 0.5381  0.60671 0.75942 0.26715\n",
      " 0.58203     nan 0.48389 0.23424 0.41513 0.14896 0.18339 0.37998 0.7424\n",
      " 0.20023 0.62341 0.20983]\n",
      "\n",
      "Epoch 00024: saving model to /content/saved models/weights.-024-0.37117\n",
      "84/84 [==============================] - 10s 114ms/step - loss: 0.3631 - root_mean_squared_error: 0.1980 - val_loss: 0.3712 - val_root_mean_squared_error: 0.2024\n",
      "Epoch 25/60\n",
      "81/84 [===========================>..] - ETA: 0s - loss: 0.3626 - root_mean_squared_error: 0.1976\n",
      "spearman :\n",
      "train_spearman : 0.48803 | val_spearman : 0.43398 | Learning_Rate : 7.999999797903001e-05\n",
      "train_spear_lst :  [0.40533 0.54288 0.57833 0.32529 0.40918 0.51414 0.37842 0.54347 0.5587\n",
      " 0.13085 0.50546 0.72539 0.55605 0.21334 0.69886 0.63227 0.78008 0.38404\n",
      " 0.64604     nan 0.54131 0.36777 0.45792 0.31136 0.32279 0.4497  0.79749\n",
      " 0.38239 0.71237 0.28174] \n",
      "val_spear_lst : [0.31082 0.55389 0.43831 0.28511 0.33066 0.46407 0.37352 0.53421 0.48593\n",
      "     nan 0.42668 0.6474  0.50939     nan 0.53452 0.58823 0.76625 0.25162\n",
      " 0.59244     nan 0.49074 0.24526 0.39845 0.13094 0.17308 0.38866 0.74771\n",
      " 0.17691 0.62668 0.24602]\n",
      "\n",
      "Epoch 00025: saving model to /content/saved models/weights.-025-0.37073\n",
      "84/84 [==============================] - 9s 110ms/step - loss: 0.3624 - root_mean_squared_error: 0.1975 - val_loss: 0.3707 - val_root_mean_squared_error: 0.2023\n",
      "Epoch 26/60\n",
      "83/84 [============================>.] - ETA: 0s - loss: 0.3617 - root_mean_squared_error: 0.1969\n",
      "spearman :\n",
      "train_spearman : 0.49558 | val_spearman : 0.43653 | Learning_Rate : 7.999999797903001e-05\n",
      "train_spear_lst :  [0.4134  0.54805 0.59343 0.34612 0.42984 0.50284 0.38361 0.54401 0.55728\n",
      " 0.13085 0.51452 0.73273 0.57114 0.21334 0.70499 0.64578 0.78108 0.38952\n",
      " 0.65419     nan 0.54267 0.38554 0.46022 0.30912 0.34391 0.4581  0.79632\n",
      " 0.39829 0.72508 0.29578] \n",
      "val_spear_lst : [0.30923 0.55771 0.43599 0.29451 0.34146 0.46432 0.36754 0.5409  0.50899\n",
      "     nan 0.43517 0.66244 0.4861      nan 0.54118 0.61011 0.7679  0.27335\n",
      " 0.58851     nan 0.48649 0.23674 0.36615 0.1468  0.18586 0.37511 0.74781\n",
      " 0.17726 0.62352 0.2552 ]\n",
      "\n",
      "Epoch 00026: saving model to /content/saved models/weights.-026-0.37073\n",
      "84/84 [==============================] - 5s 61ms/step - loss: 0.3617 - root_mean_squared_error: 0.1970 - val_loss: 0.3707 - val_root_mean_squared_error: 0.2022\n",
      "Epoch 27/60\n",
      "82/84 [============================>.] - ETA: 0s - loss: 0.3611 - root_mean_squared_error: 0.1965\n",
      "spearman :\n",
      "train_spearman : 0.49912 | val_spearman : 0.43495 | Learning_Rate : 7.999999797903001e-05\n",
      "train_spear_lst :  [0.41307 0.55563 0.58631 0.3489  0.43424 0.53013 0.38381 0.54895 0.5627\n",
      " 0.13085 0.51991 0.74335 0.57462 0.21334 0.70252 0.6461  0.78023 0.38298\n",
      " 0.65287     nan 0.54959 0.38423 0.4642  0.34105 0.3513  0.46195 0.80105\n",
      " 0.40533 0.71231 0.29298] \n",
      "val_spear_lst : [0.31162 0.55487 0.44281 0.28901 0.33963 0.47771 0.35492 0.54169 0.49186\n",
      "     nan 0.43087 0.66124 0.50311     nan 0.55897 0.58176 0.76451 0.25779\n",
      " 0.59078     nan 0.49427 0.26453 0.40134 0.07977 0.14903 0.38364 0.75746\n",
      " 0.19376 0.62677 0.24003]\n",
      "\n",
      "Epoch 00027: saving model to /content/saved models/weights.-027-0.37012\n",
      "84/84 [==============================] - 8s 96ms/step - loss: 0.3610 - root_mean_squared_error: 0.1964 - val_loss: 0.3701 - val_root_mean_squared_error: 0.2015\n",
      "Epoch 28/60\n",
      "84/84 [==============================] - ETA: 0s - loss: 0.3602 - root_mean_squared_error: 0.1958\n",
      "spearman :\n",
      "train_spearman : 0.50027 | val_spearman : 0.43508 | Learning_Rate : 7.999999797903001e-05\n",
      "train_spear_lst :  [0.40808 0.5542  0.59723 0.35117 0.43361 0.51246 0.38414 0.54951 0.56528\n",
      " 0.13085 0.51541 0.74243 0.56969 0.21334 0.70398 0.64795 0.78272 0.39462\n",
      " 0.65158     nan 0.5576  0.39847 0.47624 0.32809 0.33451 0.46685 0.80346\n",
      " 0.41071 0.72467 0.29901] \n",
      "val_spear_lst : [0.32346 0.56033 0.44583 0.28672 0.35246 0.43684 0.36357 0.53094 0.50292\n",
      "     nan 0.42659 0.66519 0.49741     nan 0.5299  0.59658 0.76943 0.24906\n",
      " 0.60213     nan 0.49023 0.23944 0.39353 0.12613 0.2159  0.37376 0.74649\n",
      " 0.18163 0.63348 0.20719]\n",
      "\n",
      "Epoch 00028: saving model to /content/saved models/weights.-028-0.37080\n",
      "84/84 [==============================] - 8s 100ms/step - loss: 0.3602 - root_mean_squared_error: 0.1958 - val_loss: 0.3708 - val_root_mean_squared_error: 0.2022\n",
      "Epoch 29/60\n",
      "81/84 [===========================>..] - ETA: 0s - loss: 0.3598 - root_mean_squared_error: 0.1955\n",
      "spearman :\n",
      "train_spearman : 0.5068 | val_spearman : 0.43928 | Learning_Rate : 7.999999797903001e-05\n",
      "train_spear_lst :  [0.40387 0.55739 0.60061 0.35113 0.44589 0.53745 0.38387 0.5527  0.56531\n",
      " 0.18477 0.52823 0.73372 0.57493 0.21334 0.71088 0.65563 0.78411 0.40564\n",
      " 0.65494     nan 0.55391 0.40137 0.48062 0.34334 0.32137 0.47388 0.8075\n",
      " 0.41704 0.73645 0.31719] \n",
      "val_spear_lst : [0.29525 0.56127 0.43134 0.29542 0.34734 0.49799 0.36989 0.54178 0.4941\n",
      "     nan 0.43278 0.6648  0.4861      nan 0.54637 0.59992 0.76889 0.26226\n",
      " 0.5951      nan 0.48059 0.26537 0.39703 0.12564 0.23523 0.38028 0.75873\n",
      " 0.16873 0.62503 0.23321]\n",
      "\n",
      "Epoch 00029: saving model to /content/saved models/weights.-029-0.37055\n",
      "84/84 [==============================] - 11s 135ms/step - loss: 0.3597 - root_mean_squared_error: 0.1953 - val_loss: 0.3706 - val_root_mean_squared_error: 0.2021\n",
      "Epoch 30/60\n",
      "81/84 [===========================>..] - ETA: 0s - loss: 0.3586 - root_mean_squared_error: 0.1944\n",
      "spearman :\n",
      "train_spearman : 0.50877 | val_spearman : 0.43641 | Learning_Rate : 7.999999797903001e-05\n",
      "train_spear_lst :  [0.42103 0.56341 0.60874 0.357   0.44583 0.51383 0.38681 0.55853 0.56546\n",
      " 0.13085 0.53326 0.73923 0.58809 0.25354 0.71426 0.64989 0.7789  0.41106\n",
      " 0.65817     nan 0.56072 0.39969 0.48084 0.33822 0.33049 0.47824 0.80785\n",
      " 0.42395 0.72874 0.32762] \n",
      "val_spear_lst : [0.31719 0.56522 0.42554 0.29157 0.34569 0.42901 0.3598  0.53796 0.48272\n",
      "     nan 0.43924 0.66828 0.53746     nan 0.54128 0.57672 0.75035 0.26612\n",
      " 0.61294     nan 0.48657 0.21975 0.42126 0.12741 0.21957 0.40065 0.74017\n",
      " 0.1881  0.63218 0.20034]\n",
      "\n",
      "Epoch 00030: saving model to /content/saved models/weights.-030-0.37129\n",
      "84/84 [==============================] - 10s 120ms/step - loss: 0.3589 - root_mean_squared_error: 0.1946 - val_loss: 0.3713 - val_root_mean_squared_error: 0.2025\n",
      "Epoch 31/60\n",
      "84/84 [==============================] - ETA: 0s - loss: 0.3590 - root_mean_squared_error: 0.1947\n",
      "spearman :\n",
      "train_spearman : 0.51215 | val_spearman : 0.4314 | Learning_Rate : 7.999999797903001e-05\n",
      "train_spear_lst :  [0.42345 0.57007 0.61437 0.36343 0.45104 0.51356 0.39067 0.55366 0.57594\n",
      " 0.13085 0.53111 0.7431  0.58853 0.26921 0.7161  0.65808 0.78931 0.4111\n",
      " 0.66014     nan 0.5647  0.41987 0.48362 0.35467 0.33761 0.48572 0.801\n",
      " 0.41396 0.71852 0.31889] \n",
      "val_spear_lst : [0.32541 0.57263 0.42517 0.28532 0.34949 0.41505 0.37109 0.52293 0.49497\n",
      "     nan 0.42076 0.6715  0.53746     nan 0.54913 0.58215 0.77157 0.25675\n",
      " 0.61107     nan 0.48365 0.21315 0.37959 0.08539 0.1659  0.38744 0.75143\n",
      " 0.1649  0.63313 0.22074]\n",
      "\n",
      "Epoch 00031: saving model to /content/saved models/weights.-031-0.37136\n",
      "84/84 [==============================] - 5s 59ms/step - loss: 0.3590 - root_mean_squared_error: 0.1947 - val_loss: 0.3714 - val_root_mean_squared_error: 0.2027\n",
      "Epoch 32/60\n",
      "83/84 [============================>.] - ETA: 0s - loss: 0.3579 - root_mean_squared_error: 0.1939\n",
      "spearman :\n",
      "train_spearman : 0.51778 | val_spearman : 0.43233 | Learning_Rate : 7.999999797903001e-05\n",
      "train_spear_lst :  [0.41705 0.57801 0.61281 0.36557 0.45273 0.52443 0.38761 0.55524 0.57864\n",
      " 0.18477 0.54284 0.74953 0.59153 0.21966 0.71143 0.65563 0.79532 0.40594\n",
      " 0.65861     nan 0.55957 0.41881 0.49168 0.36309 0.37095 0.50366 0.81258\n",
      " 0.44401 0.74139 0.3225 ] \n",
      "val_spear_lst : [0.30919 0.5754  0.46096 0.27531 0.34382 0.45196 0.35185 0.52979 0.50055\n",
      "     nan 0.42566 0.66727 0.51066     nan 0.516   0.56844 0.76667 0.24831\n",
      " 0.58952     nan 0.49455 0.24497 0.42195 0.12352 0.15356 0.37674 0.74555\n",
      " 0.16158 0.62801 0.23111]\n",
      "\n",
      "Epoch 00032: saving model to /content/saved models/weights.-032-0.37032\n",
      "84/84 [==============================] - 9s 105ms/step - loss: 0.3579 - root_mean_squared_error: 0.1939 - val_loss: 0.3703 - val_root_mean_squared_error: 0.2017\n",
      "Epoch 33/60\n",
      "81/84 [===========================>..] - ETA: 0s - loss: 0.3574 - root_mean_squared_error: 0.1938\n",
      "spearman :\n",
      "train_spearman : 0.52199 | val_spearman : 0.43435 | Learning_Rate : 7.999999797903001e-05\n",
      "train_spear_lst :  [0.41988 0.57611 0.61642 0.37238 0.45119 0.52978 0.38375 0.56327 0.58594\n",
      " 0.18477 0.53317 0.74888 0.59232 0.26921 0.71568 0.66691 0.7915  0.41132\n",
      " 0.66452     nan 0.56872 0.43235 0.48791 0.35875 0.36503 0.50241 0.81121\n",
      " 0.44408 0.74284 0.34727] \n",
      "val_spear_lst : [0.3044  0.57337 0.46824 0.26731 0.33473 0.46275 0.36956 0.53734 0.50534\n",
      "     nan 0.41304 0.67524 0.54432     nan 0.516   0.55752 0.76754 0.26389\n",
      " 0.59818     nan 0.49801 0.24074 0.39028 0.10485 0.16158 0.37976 0.75647\n",
      " 0.17559 0.634   0.22748]\n",
      "\n",
      "Epoch 00033: saving model to /content/saved models/weights.-033-0.37019\n",
      "84/84 [==============================] - 10s 116ms/step - loss: 0.3573 - root_mean_squared_error: 0.1935 - val_loss: 0.3702 - val_root_mean_squared_error: 0.2012\n",
      "Epoch 34/60\n",
      "83/84 [============================>.] - ETA: 0s - loss: 0.3565 - root_mean_squared_error: 0.1928\n",
      "spearman :\n",
      "train_spearman : 0.52751 | val_spearman : 0.43321 | Learning_Rate : 7.999999797903001e-05\n",
      "train_spear_lst :  [0.43125 0.5788  0.61687 0.37166 0.45724 0.54819 0.39167 0.55998 0.57935\n",
      " 0.18477 0.54555 0.74552 0.59304 0.26921 0.71658 0.66261 0.79788 0.40731\n",
      " 0.67124     nan 0.56767 0.43119 0.49425 0.41151 0.40452 0.50919 0.81161\n",
      " 0.46962 0.74398 0.32541] \n",
      "val_spear_lst : [0.31136 0.57223 0.44395 0.27941 0.35195 0.49873 0.3667  0.53525 0.50026\n",
      "     nan 0.42609 0.67    0.54183     nan 0.52565 0.57339 0.76629 0.2384\n",
      " 0.59875     nan 0.50857 0.22303 0.39538 0.10763 0.12627 0.37485 0.74218\n",
      " 0.16402 0.62958 0.22491]\n",
      "\n",
      "Epoch 00034: ReduceLROnPlateau reducing learning rate to 2.529822064226021e-05.\n",
      "\n",
      "Epoch 00034: saving model to /content/saved models/weights.-034-0.37092\n",
      "84/84 [==============================] - 6s 74ms/step - loss: 0.3564 - root_mean_squared_error: 0.1927 - val_loss: 0.3709 - val_root_mean_squared_error: 0.2020\n",
      "Epoch 35/60\n",
      "83/84 [============================>.] - ETA: 0s - loss: 0.3552 - root_mean_squared_error: 0.1917\n",
      "spearman :\n",
      "train_spearman : 0.5297 | val_spearman : 0.43335 | Learning_Rate : 2.499999936844688e-05\n",
      "train_spear_lst :  [0.43194 0.584   0.61761 0.38159 0.46227 0.53649 0.38973 0.55884 0.58165\n",
      " 0.18477 0.55282 0.7526  0.59554 0.26921 0.72756 0.67228 0.79649 0.42258\n",
      " 0.66947     nan 0.57113 0.44325 0.49638 0.38912 0.37522 0.51492 0.82106\n",
      " 0.45984 0.74684 0.35612] \n",
      "val_spear_lst : [0.3137  0.57569 0.43299 0.27504 0.36017 0.46205 0.3579  0.52931 0.50275\n",
      "     nan 0.43358 0.68121 0.54432 0.27965 0.58021 0.57531 0.77146 0.25927\n",
      " 0.59211     nan 0.49593 0.21909 0.43037 0.10513 0.19211 0.38387 0.75687\n",
      " 0.17006 0.62056 0.23303]\n",
      "\n",
      "Epoch 00035: saving model to /content/saved models/weights.-035-0.37016\n",
      "84/84 [==============================] - 9s 111ms/step - loss: 0.3552 - root_mean_squared_error: 0.1916 - val_loss: 0.3702 - val_root_mean_squared_error: 0.2013\n",
      "Epoch 36/60\n",
      "81/84 [===========================>..] - ETA: 0s - loss: 0.3546 - root_mean_squared_error: 0.1912\n",
      "spearman :\n",
      "train_spearman : 0.53266 | val_spearman : 0.42948 | Learning_Rate : 2.499999936844688e-05\n",
      "train_spear_lst :  [0.43123 0.58877 0.62285 0.3852  0.46554 0.53822 0.39072 0.55984 0.59124\n",
      " 0.18477 0.55013 0.75441 0.59554 0.26921 0.72599 0.66562 0.79962 0.42355\n",
      " 0.6687      nan 0.5749  0.45455 0.49453 0.40345 0.38992 0.52282 0.82259\n",
      " 0.46032 0.75454 0.35828] \n",
      "val_spear_lst : [0.3075  0.57732 0.44315 0.30297 0.34604 0.47114 0.36306 0.53196 0.49617\n",
      "     nan 0.42585 0.67962 0.54432 0.27965 0.54128 0.56917 0.76709 0.23748\n",
      " 0.59561     nan 0.50199 0.23721 0.4283  0.09415 0.15035 0.36076 0.75455\n",
      " 0.17252 0.6371  0.20907]\n",
      "\n",
      "Epoch 00036: saving model to /content/saved models/weights.-036-0.36982\n",
      "84/84 [==============================] - 9s 104ms/step - loss: 0.3547 - root_mean_squared_error: 0.1914 - val_loss: 0.3698 - val_root_mean_squared_error: 0.2010\n",
      "Epoch 37/60\n",
      "84/84 [==============================] - ETA: 0s - loss: 0.3540 - root_mean_squared_error: 0.1907\n",
      "spearman :\n",
      "train_spearman : 0.53532 | val_spearman : 0.43013 | Learning_Rate : 2.499999936844688e-05\n",
      "train_spear_lst :  [0.43098 0.58786 0.62155 0.39015 0.46636 0.53852 0.38847 0.55967 0.59259\n",
      " 0.18477 0.55025 0.75541 0.59588 0.28412 0.72591 0.66486 0.79651 0.42283\n",
      " 0.66986     nan 0.57645 0.46358 0.49072 0.40675 0.40925 0.52423 0.8246\n",
      " 0.47841 0.75511 0.36862] \n",
      "val_spear_lst : [0.30599 0.58224 0.44605 0.29647 0.35682 0.4688  0.35117 0.53617 0.49611\n",
      "     nan 0.4226  0.67747 0.54432 0.27965 0.55981 0.56738 0.77835 0.23883\n",
      " 0.60887     nan 0.4994  0.25047 0.38367 0.09519 0.15794 0.3703  0.75153\n",
      " 0.15729 0.63013 0.23063]\n",
      "\n",
      "Epoch 00037: saving model to /content/saved models/weights.-037-0.37009\n",
      "84/84 [==============================] - 7s 81ms/step - loss: 0.3540 - root_mean_squared_error: 0.1907 - val_loss: 0.3701 - val_root_mean_squared_error: 0.2013\n",
      "Epoch 38/60\n",
      "82/84 [============================>.] - ETA: 0s - loss: 0.3539 - root_mean_squared_error: 0.1905\n",
      "spearman :\n",
      "train_spearman : 0.53755 | val_spearman : 0.42995 | Learning_Rate : 2.499999936844688e-05\n",
      "train_spear_lst :  [0.43306 0.58671 0.6241  0.3922  0.47018 0.5392  0.39134 0.55949 0.59081\n",
      " 0.18477 0.55486 0.75583 0.59554 0.28412 0.72424 0.67536 0.79833 0.42673\n",
      " 0.67193     nan 0.57717 0.46251 0.50557 0.42195 0.41438 0.52901 0.82155\n",
      " 0.4774  0.75485 0.36587] \n",
      "val_spear_lst : [0.30664 0.57943 0.44371 0.29367 0.35872 0.46095 0.34871 0.5305  0.51297\n",
      "     nan 0.42813 0.67323 0.54432 0.27965 0.54924 0.55436 0.77494 0.24454\n",
      " 0.60084     nan 0.49921 0.24186 0.41112 0.09224 0.16694 0.37941 0.74286\n",
      " 0.16286 0.63226 0.2254 ]\n",
      "\n",
      "Epoch 00038: saving model to /content/saved models/weights.-038-0.37036\n",
      "84/84 [==============================] - 8s 100ms/step - loss: 0.3538 - root_mean_squared_error: 0.1906 - val_loss: 0.3704 - val_root_mean_squared_error: 0.2014\n",
      "Epoch 39/60\n",
      "82/84 [============================>.] - ETA: 0s - loss: 0.3535 - root_mean_squared_error: 0.1904\n",
      "spearman :\n",
      "train_spearman : 0.54003 | val_spearman : 0.42972 | Learning_Rate : 2.499999936844688e-05\n",
      "train_spear_lst :  [0.43844 0.58787 0.62484 0.38754 0.47059 0.53927 0.39083 0.55798 0.59658\n",
      " 0.26049 0.55594 0.75618 0.59746 0.26594 0.7251  0.6679  0.79842 0.42448\n",
      " 0.6731      nan 0.58092 0.47319 0.50143 0.39637 0.40491 0.53404 0.82476\n",
      " 0.48923 0.76246 0.3746 ] \n",
      "val_spear_lst : [0.30804 0.57782 0.44044 0.28191 0.36215 0.4644  0.34684 0.53197 0.48768\n",
      "     nan 0.42694 0.67034 0.53432 0.27965 0.55532 0.56844 0.76811 0.24147\n",
      " 0.58905     nan 0.50805 0.2083  0.41758 0.10837 0.16594 0.38571 0.75226\n",
      " 0.18475 0.62329 0.24297]\n",
      "\n",
      "Epoch 00039: saving model to /content/saved models/weights.-039-0.37009\n",
      "84/84 [==============================] - 9s 108ms/step - loss: 0.3536 - root_mean_squared_error: 0.1904 - val_loss: 0.3701 - val_root_mean_squared_error: 0.2011\n",
      "Epoch 40/60\n",
      "81/84 [===========================>..] - ETA: 0s - loss: 0.3536 - root_mean_squared_error: 0.1904\n",
      "spearman :\n",
      "train_spearman : 0.53959 | val_spearman : 0.42784 | Learning_Rate : 2.499999936844688e-05\n",
      "train_spear_lst :  [0.44171 0.59541 0.62384 0.38865 0.47131 0.54203 0.39028 0.5623  0.59452\n",
      " 0.18477 0.55784 0.75755 0.59848 0.26594 0.72328 0.67686 0.79714 0.42416\n",
      " 0.67168     nan 0.58245 0.47505 0.50116 0.4134  0.41573 0.53508 0.82654\n",
      " 0.49325 0.76021 0.3776 ] \n",
      "val_spear_lst : [0.30729 0.58517 0.43403 0.27348 0.3545  0.47182 0.3445  0.52416 0.50418\n",
      "     nan 0.43154 0.67316 0.53432 0.27965 0.55525 0.55932 0.77611 0.24045\n",
      " 0.59832     nan 0.4904  0.20714 0.39864 0.09962 0.1691  0.37244 0.74844\n",
      " 0.18282 0.63252 0.23105]\n",
      "\n",
      "Epoch 00040: saving model to /content/saved models/weights.-040-0.37018\n",
      "84/84 [==============================] - 11s 128ms/step - loss: 0.3535 - root_mean_squared_error: 0.1904 - val_loss: 0.3702 - val_root_mean_squared_error: 0.2014\n",
      "Epoch 41/60\n",
      "81/84 [===========================>..] - ETA: 0s - loss: 0.3532 - root_mean_squared_error: 0.1900\n",
      "spearman :\n",
      "train_spearman : 0.54113 | val_spearman : 0.42891 | Learning_Rate : 2.499999936844688e-05\n",
      "train_spear_lst :  [0.43982 0.59197 0.62367 0.39612 0.47095 0.5417  0.39086 0.56158 0.59859\n",
      " 0.18477 0.5603  0.75829 0.59937 0.28412 0.73173 0.66787 0.79905 0.42839\n",
      " 0.67565     nan 0.58426 0.47959 0.50171 0.42295 0.40594 0.53103 0.82944\n",
      " 0.48346 0.76343 0.38629] \n",
      "val_spear_lst : [0.30555 0.58041 0.43635 0.28018 0.36106 0.47751 0.35257 0.53194 0.50031\n",
      "     nan 0.42134 0.67459 0.53432 0.27965 0.54924 0.56854 0.77372 0.24318\n",
      " 0.59643     nan 0.50689 0.21523 0.40426 0.08733 0.1648  0.39215 0.75114\n",
      " 0.15356 0.6345  0.23277]\n",
      "\n",
      "Epoch 00041: saving model to /content/saved models/weights.-041-0.37015\n",
      "84/84 [==============================] - 5s 61ms/step - loss: 0.3532 - root_mean_squared_error: 0.1901 - val_loss: 0.3701 - val_root_mean_squared_error: 0.2013\n",
      "Epoch 42/60\n",
      "84/84 [==============================] - ETA: 0s - loss: 0.3528 - root_mean_squared_error: 0.1897\n",
      "spearman :\n",
      "train_spearman : 0.54257 | val_spearman : 0.42759 | Learning_Rate : 2.499999936844688e-05\n",
      "train_spear_lst :  [0.43602 0.59309 0.62444 0.39623 0.4739  0.53194 0.39052 0.56171 0.60098\n",
      " 0.18477 0.56017 0.75926 0.59921 0.28412 0.73203 0.67685 0.79983 0.42665\n",
      " 0.67458     nan 0.58366 0.47936 0.50854 0.43719 0.414   0.53669 0.83226\n",
      " 0.49342 0.76387 0.37921] \n",
      "val_spear_lst : [0.30979 0.57843 0.4309  0.29146 0.35694 0.45422 0.34852 0.53213 0.51169\n",
      "     nan 0.42797 0.67576 0.53432 0.27965 0.53761 0.56439 0.77135 0.25278\n",
      " 0.60388     nan 0.50763 0.2133  0.41172 0.07625 0.14672 0.36882 0.74559\n",
      " 0.17667 0.62421 0.23995]\n",
      "\n",
      "Epoch 00042: saving model to /content/saved models/weights.-042-0.36990\n",
      "84/84 [==============================] - 11s 126ms/step - loss: 0.3528 - root_mean_squared_error: 0.1897 - val_loss: 0.3699 - val_root_mean_squared_error: 0.2011\n",
      "Epoch 43/60\n",
      "83/84 [============================>.] - ETA: 0s - loss: 0.3530 - root_mean_squared_error: 0.1899\n",
      "spearman :\n",
      "train_spearman : 0.54419 | val_spearman : 0.42693 | Learning_Rate : 2.499999936844688e-05\n",
      "train_spear_lst :  [0.4389  0.59618 0.62781 0.39216 0.47224 0.54155 0.393   0.56408 0.60232\n",
      " 0.18477 0.56673 0.76025 0.59921 0.26594 0.73231 0.66784 0.8016  0.42913\n",
      " 0.6737      nan 0.58361 0.48975 0.50927 0.44109 0.42786 0.54041 0.83281\n",
      " 0.49731 0.76663 0.38315] \n",
      "val_spear_lst : [0.30356 0.57874 0.44044 0.29382 0.3602  0.47277 0.35066 0.53429 0.4984\n",
      "     nan 0.42877 0.6799  0.53432 0.27965 0.54816 0.56854 0.76739 0.24164\n",
      " 0.60407     nan 0.50397 0.21267 0.39743 0.07648 0.13925 0.37454 0.75\n",
      " 0.15516 0.63076 0.22843]\n",
      "\n",
      "Epoch 00043: ReduceLROnPlateau reducing learning rate to 7.999999867444209e-06.\n",
      "\n",
      "Epoch 00043: saving model to /content/saved models/weights.-043-0.36999\n",
      "84/84 [==============================] - 6s 73ms/step - loss: 0.3530 - root_mean_squared_error: 0.1899 - val_loss: 0.3700 - val_root_mean_squared_error: 0.2012\n",
      "Epoch 44/60\n",
      "84/84 [==============================] - ETA: 0s - loss: 0.3521 - root_mean_squared_error: 0.1891\n",
      "spearman :\n",
      "train_spearman : 0.54478 | val_spearman : 0.42791 | Learning_Rate : 7.999999979801942e-06\n",
      "train_spear_lst :  [0.43841 0.59416 0.62612 0.39996 0.47315 0.54412 0.39098 0.56224 0.60202\n",
      " 0.18477 0.56365 0.75984 0.59921 0.26594 0.73231 0.66883 0.8026  0.42909\n",
      " 0.67355     nan 0.58724 0.49059 0.51499 0.43828 0.42448 0.54116 0.83353\n",
      " 0.50347 0.76578 0.38808] \n",
      "val_spear_lst : [0.30761 0.57946 0.44044 0.27763 0.36021 0.47798 0.34866 0.53313 0.50731\n",
      "     nan 0.4257  0.67789 0.54183 0.27965 0.55525 0.56738 0.76793 0.23841\n",
      " 0.60097     nan 0.50369 0.21429 0.41121 0.08492 0.15351 0.36962 0.74446\n",
      " 0.1562  0.63191 0.22422]\n",
      "\n",
      "Epoch 00044: saving model to /content/saved models/weights.-044-0.36989\n",
      "84/84 [==============================] - 12s 143ms/step - loss: 0.3521 - root_mean_squared_error: 0.1891 - val_loss: 0.3699 - val_root_mean_squared_error: 0.2011\n",
      "Epoch 45/60\n",
      "81/84 [===========================>..] - ETA: 0s - loss: 0.3523 - root_mean_squared_error: 0.1893\n",
      "spearman :\n",
      "train_spearman : 0.54597 | val_spearman : 0.4275 | Learning_Rate : 7.999999979801942e-06\n",
      "train_spear_lst :  [0.43763 0.5964  0.62888 0.39885 0.47374 0.54412 0.39088 0.56292 0.6028\n",
      " 0.18477 0.56472 0.75972 0.59921 0.26594 0.73361 0.67995 0.80155 0.42976\n",
      " 0.6755      nan 0.58809 0.48873 0.50935 0.4499  0.4306  0.54597 0.83366\n",
      " 0.50027 0.76542 0.39021] \n",
      "val_spear_lst : [0.3059  0.5792  0.43635 0.28171 0.36269 0.47521 0.34665 0.53608 0.50573\n",
      "     nan 0.42622 0.67443 0.53432 0.27965 0.55431 0.56554 0.7697  0.25351\n",
      " 0.60098     nan 0.51161 0.21386 0.39591 0.07599 0.15708 0.36283 0.75167\n",
      " 0.15402 0.63218 0.22672]\n",
      "\n",
      "Epoch 00045: saving model to /content/saved models/weights.-045-0.36974\n",
      "84/84 [==============================] - 5s 58ms/step - loss: 0.3521 - root_mean_squared_error: 0.1892 - val_loss: 0.3697 - val_root_mean_squared_error: 0.2009\n",
      "Epoch 46/60\n",
      "83/84 [============================>.] - ETA: 0s - loss: 0.3517 - root_mean_squared_error: 0.1890\n",
      "spearman :\n",
      "train_spearman : 0.54547 | val_spearman : 0.4292 | Learning_Rate : 7.999999979801942e-06\n",
      "train_spear_lst :  [0.44087 0.59685 0.6272  0.40079 0.47642 0.54445 0.39161 0.56313 0.60311\n",
      " 0.18477 0.56885 0.76029 0.59921 0.26594 0.73246 0.67531 0.8021  0.42901\n",
      " 0.678       nan 0.58799 0.49083 0.5108  0.43715 0.41409 0.54435 0.8328\n",
      " 0.50269 0.7651  0.3925 ] \n",
      "val_spear_lst : [0.30627 0.5856  0.43972 0.28508 0.35823 0.47376 0.35492 0.53222 0.51296\n",
      "     nan 0.4215  0.67795 0.54183 0.27965 0.54727 0.56844 0.77131 0.24191\n",
      " 0.5996      nan 0.50865 0.2132  0.40322 0.08601 0.17305 0.37676 0.75074\n",
      " 0.15605 0.63036 0.22136]\n",
      "\n",
      "Epoch 00046: saving model to /content/saved models/weights.-046-0.36983\n",
      "84/84 [==============================] - 13s 153ms/step - loss: 0.3517 - root_mean_squared_error: 0.1889 - val_loss: 0.3698 - val_root_mean_squared_error: 0.2009\n",
      "Epoch 47/60\n",
      "82/84 [============================>.] - ETA: 0s - loss: 0.3519 - root_mean_squared_error: 0.1890\n",
      "spearman :\n",
      "train_spearman : 0.54684 | val_spearman : 0.42896 | Learning_Rate : 7.999999979801942e-06\n",
      "train_spear_lst :  [0.43921 0.59752 0.63044 0.40118 0.47568 0.54425 0.39184 0.56603 0.60319\n",
      " 0.18477 0.56808 0.7613  0.59921 0.26594 0.72985 0.67198 0.8024  0.43142\n",
      " 0.67752     nan 0.58649 0.49546 0.50977 0.45079 0.42852 0.54796 0.83409\n",
      " 0.50337 0.76868 0.39146] \n",
      "val_spear_lst : [0.3089  0.58296 0.44605 0.2836  0.36225 0.4722  0.3505  0.53324 0.50569\n",
      "     nan 0.42419 0.67541 0.54183 0.27965 0.55981 0.5725  0.77034 0.24622\n",
      " 0.604       nan 0.51337 0.21066 0.39641 0.09083 0.16761 0.36213 0.75017\n",
      " 0.15441 0.62368 0.22215]\n",
      "\n",
      "Epoch 00047: saving model to /content/saved models/weights.-047-0.36975\n",
      "84/84 [==============================] - 5s 58ms/step - loss: 0.3518 - root_mean_squared_error: 0.1889 - val_loss: 0.3698 - val_root_mean_squared_error: 0.2009\n",
      "Epoch 48/60\n",
      "82/84 [============================>.] - ETA: 0s - loss: 0.3519 - root_mean_squared_error: 0.1890\n",
      "spearman :\n",
      "train_spearman : 0.54672 | val_spearman : 0.4295 | Learning_Rate : 7.999999979801942e-06\n",
      "train_spear_lst :  [0.438   0.59841 0.62888 0.39639 0.47591 0.5461  0.39102 0.565   0.6044\n",
      " 0.18477 0.56704 0.76162 0.59921 0.26594 0.72887 0.67545 0.80289 0.43041\n",
      " 0.67962     nan 0.58531 0.49426 0.51111 0.44973 0.42535 0.54654 0.83456\n",
      " 0.50611 0.76881 0.39311] \n",
      "val_spear_lst : [0.30771 0.58477 0.43635 0.28059 0.36262 0.47354 0.35559 0.53895 0.5039\n",
      "     nan 0.42373 0.67932 0.54183 0.27965 0.56065 0.56902 0.77321 0.2403\n",
      " 0.60127     nan 0.51354 0.21325 0.40107 0.08365 0.17695 0.36201 0.74609\n",
      " 0.15947 0.63379 0.2232 ]\n",
      "\n",
      "Epoch 00048: saving model to /content/saved models/weights.-048-0.36975\n",
      "84/84 [==============================] - 7s 80ms/step - loss: 0.3516 - root_mean_squared_error: 0.1889 - val_loss: 0.3698 - val_root_mean_squared_error: 0.2009\n",
      "Epoch 49/60\n",
      "81/84 [===========================>..] - ETA: 0s - loss: 0.3514 - root_mean_squared_error: 0.1889\n",
      "spearman :\n",
      "train_spearman : 0.54736 | val_spearman : 0.42891 | Learning_Rate : 7.999999979801942e-06\n",
      "train_spear_lst :  [0.44047 0.59956 0.62994 0.40136 0.47593 0.54439 0.39292 0.56396 0.60317\n",
      " 0.18477 0.56556 0.7624  0.59921 0.26594 0.72975 0.67765 0.80351 0.43108\n",
      " 0.67853     nan 0.58663 0.49384 0.51394 0.44598 0.4266  0.5464  0.8351\n",
      " 0.50964 0.77139 0.39373] \n",
      "val_spear_lst : [0.31052 0.58179 0.44272 0.2816  0.35824 0.47202 0.34871 0.53338 0.50446\n",
      "     nan 0.42319 0.6796  0.54183 0.27965 0.56065 0.56844 0.77074 0.24437\n",
      " 0.60279     nan 0.51531 0.20776 0.40627 0.07256 0.16669 0.37494 0.74837\n",
      " 0.14367 0.63565 0.23358]\n",
      "\n",
      "Epoch 00049: saving model to /content/saved models/weights.-049-0.36992\n",
      "84/84 [==============================] - 9s 102ms/step - loss: 0.3517 - root_mean_squared_error: 0.1889 - val_loss: 0.3699 - val_root_mean_squared_error: 0.2010\n",
      "Epoch 50/60\n",
      "82/84 [============================>.] - ETA: 0s - loss: 0.3515 - root_mean_squared_error: 0.1886\n",
      "spearman :\n",
      "train_spearman : 0.54809 | val_spearman : 0.42968 | Learning_Rate : 7.999999979801942e-06\n",
      "train_spear_lst :  [0.4389  0.59821 0.62999 0.40508 0.47729 0.54256 0.39308 0.56532 0.60643\n",
      " 0.18477 0.56759 0.76157 0.59921 0.26594 0.7339  0.67883 0.80419 0.4318\n",
      " 0.67491     nan 0.58803 0.49433 0.51327 0.4448  0.43272 0.55063 0.83661\n",
      " 0.50764 0.76978 0.39715] \n",
      "val_spear_lst : [0.31326 0.58515 0.44605 0.27449 0.35712 0.47701 0.35609 0.53518 0.50862\n",
      "     nan 0.42351 0.67948 0.54183 0.27965 0.55431 0.56896 0.77216 0.24437\n",
      " 0.60099     nan 0.51636 0.21808 0.40155 0.06808 0.17127 0.3834  0.74634\n",
      " 0.15776 0.62853 0.22158]\n",
      "\n",
      "Epoch 00050: ReduceLROnPlateau reducing learning rate to 2.5298221217475167e-06.\n",
      "\n",
      "Epoch 00050: saving model to /content/saved models/weights.-050-0.36999\n",
      "84/84 [==============================] - 12s 141ms/step - loss: 0.3514 - root_mean_squared_error: 0.1886 - val_loss: 0.3700 - val_root_mean_squared_error: 0.2011\n",
      "Epoch 51/60\n",
      "84/84 [==============================] - ETA: 0s - loss: 0.3512 - root_mean_squared_error: 0.1885\n",
      "spearman :\n",
      "train_spearman : 0.5489 | val_spearman : 0.42951 | Learning_Rate : 3.000000106112566e-06\n",
      "train_spear_lst :  [0.43941 0.59951 0.62824 0.40242 0.47613 0.54261 0.39317 0.56431 0.60647\n",
      " 0.18477 0.56917 0.76286 0.59921 0.26594 0.73758 0.6764  0.80338 0.43267\n",
      " 0.67752     nan 0.58717 0.4969  0.51546 0.45615 0.43689 0.55006 0.83554\n",
      " 0.51424 0.77146 0.39242] \n",
      "val_spear_lst : [0.31392 0.58473 0.44605 0.28508 0.35764 0.4771  0.35609 0.53578 0.50687\n",
      "     nan 0.42404 0.67799 0.54183 0.27965 0.55431 0.56552 0.77307 0.24437\n",
      " 0.60073     nan 0.51417 0.21847 0.40663 0.07298 0.16959 0.36549 0.7469\n",
      " 0.15262 0.62921 0.22534]\n",
      "\n",
      "Epoch 00051: saving model to /content/saved models/weights.-051-0.36983\n",
      "84/84 [==============================] - 5s 60ms/step - loss: 0.3512 - root_mean_squared_error: 0.1885 - val_loss: 0.3698 - val_root_mean_squared_error: 0.2009\n",
      "Epoch 52/60\n",
      "82/84 [============================>.] - ETA: 0s - loss: 0.3515 - root_mean_squared_error: 0.1887\n",
      "spearman :\n",
      "train_spearman : 0.54848 | val_spearman : 0.42892 | Learning_Rate : 3.000000106112566e-06\n",
      "train_spear_lst :  [0.43978 0.59926 0.62999 0.4025  0.4772  0.54436 0.39242 0.56386 0.60326\n",
      " 0.18477 0.56857 0.76324 0.59921 0.26594 0.73586 0.68004 0.80436 0.43317\n",
      " 0.67806     nan 0.58863 0.49863 0.51312 0.45027 0.43279 0.55205 0.83496\n",
      " 0.51009 0.7685  0.39115] \n",
      "val_spear_lst : [0.31065 0.58535 0.44605 0.28286 0.35931 0.4723  0.3557  0.53578 0.50744\n",
      "     nan 0.41964 0.6756  0.54183 0.27965 0.55431 0.56896 0.77186 0.24117\n",
      " 0.59879     nan 0.51332 0.21765 0.40116 0.06665 0.17214 0.37233 0.74751\n",
      " 0.16148 0.63015 0.22012]\n",
      "\n",
      "Epoch 00052: saving model to /content/saved models/weights.-052-0.36991\n",
      "84/84 [==============================] - 8s 101ms/step - loss: 0.3513 - root_mean_squared_error: 0.1886 - val_loss: 0.3699 - val_root_mean_squared_error: 0.2010\n",
      "Epoch 53/60\n",
      "83/84 [============================>.] - ETA: 0s - loss: 0.3513 - root_mean_squared_error: 0.1887\n",
      "spearman :\n",
      "train_spearman : 0.54902 | val_spearman : 0.42846 | Learning_Rate : 3.000000106112566e-06\n",
      "train_spear_lst :  [0.43967 0.59935 0.63114 0.40333 0.47658 0.54544 0.39343 0.56375 0.60319\n",
      " 0.18477 0.56898 0.76233 0.59921 0.26594 0.735   0.6764  0.80359 0.43292\n",
      " 0.67703     nan 0.58835 0.49904 0.51484 0.45242 0.43665 0.55155 0.8359\n",
      " 0.51345 0.77032 0.39715] \n",
      "val_spear_lst : [0.3072  0.58294 0.44944 0.2816  0.35154 0.47507 0.35609 0.53436 0.5034\n",
      "     nan 0.42406 0.67692 0.54183 0.27965 0.55431 0.56552 0.7725  0.24437\n",
      " 0.60033     nan 0.51377 0.21535 0.40384 0.07675 0.17407 0.3656  0.73899\n",
      " 0.15934 0.63028 0.21779]\n",
      "\n",
      "Epoch 00053: saving model to /content/saved models/weights.-053-0.36983\n",
      "84/84 [==============================] - 8s 100ms/step - loss: 0.3514 - root_mean_squared_error: 0.1887 - val_loss: 0.3698 - val_root_mean_squared_error: 0.2009\n",
      "Epoch 54/60\n",
      "84/84 [==============================] - ETA: 0s - loss: 0.3513 - root_mean_squared_error: 0.1886\n",
      "spearman :\n",
      "train_spearman : 0.54894 | val_spearman : 0.42896 | Learning_Rate : 3.000000106112566e-06\n",
      "train_spear_lst :  [0.44054 0.6003  0.63114 0.40325 0.47788 0.54448 0.39306 0.56519 0.60742\n",
      " 0.18477 0.57001 0.76183 0.59921 0.26594 0.73101 0.67618 0.80393 0.43227\n",
      " 0.67807     nan 0.58889 0.50045 0.51531 0.45248 0.43396 0.54993 0.83514\n",
      " 0.51052 0.77177 0.39437] \n",
      "val_spear_lst : [0.31283 0.58551 0.44944 0.27884 0.3517  0.47382 0.35841 0.53436 0.50683\n",
      "     nan 0.41949 0.67692 0.54183 0.27965 0.56065 0.56552 0.77224 0.24437\n",
      " 0.60249     nan 0.5165  0.21692 0.40384 0.07616 0.17551 0.3601  0.74266\n",
      " 0.15967 0.62329 0.22131]\n",
      "\n",
      "Epoch 00054: saving model to /content/saved models/weights.-054-0.36980\n",
      "84/84 [==============================] - 8s 93ms/step - loss: 0.3513 - root_mean_squared_error: 0.1886 - val_loss: 0.3698 - val_root_mean_squared_error: 0.2009\n",
      "Epoch 55/60\n",
      "84/84 [==============================] - ETA: 0s - loss: 0.3511 - root_mean_squared_error: 0.1885\n",
      "spearman :\n",
      "train_spearman : 0.55188 | val_spearman : 0.42886 | Learning_Rate : 3.000000106112566e-06\n",
      "train_spear_lst :  [0.43957 0.60052 0.63276 0.40438 0.47748 0.54527 0.39314 0.56492 0.60663\n",
      " 0.26049 0.56992 0.76112 0.59921 0.26594 0.73586 0.67483 0.8031  0.43296\n",
      " 0.67964     nan 0.58907 0.49781 0.51524 0.44953 0.43624 0.55105 0.8356\n",
      " 0.51496 0.77049 0.39683] \n",
      "val_spear_lst : [0.30381 0.58815 0.44944 0.2816  0.35598 0.47352 0.35281 0.53578 0.50687\n",
      "     nan 0.4244  0.67692 0.54183 0.27965 0.55431 0.56552 0.77348 0.24618\n",
      " 0.60397     nan 0.51714 0.21207 0.40276 0.06818 0.17214 0.37027 0.74257\n",
      " 0.16113 0.6279  0.21968]\n",
      "\n",
      "Epoch 00055: saving model to /content/saved models/weights.-055-0.36981\n",
      "84/84 [==============================] - 9s 104ms/step - loss: 0.3511 - root_mean_squared_error: 0.1885 - val_loss: 0.3698 - val_root_mean_squared_error: 0.2009\n",
      "Epoch 56/60\n",
      "84/84 [==============================] - ETA: 0s - loss: 0.3511 - root_mean_squared_error: 0.1882\n",
      "spearman :\n",
      "train_spearman : 0.54952 | val_spearman : 0.42834 | Learning_Rate : 3.000000106112566e-06\n",
      "train_spear_lst :  [0.44136 0.60019 0.6329  0.40406 0.47741 0.54521 0.3929  0.56575 0.60636\n",
      " 0.18477 0.57034 0.76175 0.59921 0.26594 0.73672 0.67536 0.80346 0.43242\n",
      " 0.67979     nan 0.58823 0.49905 0.51481 0.45282 0.43532 0.55154 0.83637\n",
      " 0.51366 0.77157 0.3968 ] \n",
      "val_spear_lst : [0.31149 0.58586 0.44944 0.28233 0.35376 0.47224 0.35383 0.53586 0.50862\n",
      "     nan 0.42419 0.67692 0.54183 0.27965 0.55431 0.56552 0.7735  0.24437\n",
      " 0.60065     nan 0.51414 0.21055 0.40223 0.06877 0.17407 0.36523 0.74454\n",
      " 0.16273 0.62232 0.21448]\n",
      "\n",
      "Epoch 00056: saving model to /content/saved models/weights.-056-0.36979\n",
      "84/84 [==============================] - 10s 116ms/step - loss: 0.3511 - root_mean_squared_error: 0.1882 - val_loss: 0.3698 - val_root_mean_squared_error: 0.2009\n",
      "Epoch 57/60\n",
      "83/84 [============================>.] - ETA: 0s - loss: 0.3513 - root_mean_squared_error: 0.1886\n",
      "spearman :\n",
      "train_spearman : 0.55234 | val_spearman : 0.42792 | Learning_Rate : 3.000000106112566e-06\n",
      "train_spear_lst :  [0.44129 0.60014 0.6352  0.40469 0.47827 0.5462  0.39215 0.56562 0.60588\n",
      " 0.26049 0.56853 0.76247 0.59921 0.26594 0.73586 0.67483 0.80416 0.4333\n",
      " 0.67996     nan 0.5884  0.49923 0.51597 0.45043 0.43592 0.55091 0.83633\n",
      " 0.51732 0.77113 0.39805] \n",
      "val_spear_lst : [0.30699 0.58328 0.44944 0.2818  0.35358 0.47369 0.35383 0.53586 0.50375\n",
      "     nan 0.42119 0.67692 0.54183 0.27965 0.55431 0.56552 0.77385 0.24117\n",
      " 0.60396     nan 0.51488 0.20967 0.40438 0.07108 0.17424 0.36898 0.744\n",
      " 0.14877 0.62696 0.21816]\n",
      "\n",
      "Epoch 00057: ReduceLROnPlateau reducing learning rate to 7.999999723640471e-07.\n",
      "\n",
      "Epoch 00057: saving model to /content/saved models/weights.-057-0.36981\n",
      "84/84 [==============================] - 6s 76ms/step - loss: 0.3513 - root_mean_squared_error: 0.1886 - val_loss: 0.3698 - val_root_mean_squared_error: 0.2009\n",
      "Epoch 58/60\n",
      "81/84 [===========================>..] - ETA: 0s - loss: 0.3515 - root_mean_squared_error: 0.1888\n",
      "spearman :\n",
      "train_spearman : 0.55233 | val_spearman : 0.42868 | Learning_Rate : 9.999999974752427e-07\n",
      "train_spear_lst :  [0.44026 0.60019 0.63359 0.40384 0.4791  0.54499 0.39224 0.56628 0.60621\n",
      " 0.26049 0.5687  0.76158 0.59921 0.26594 0.73586 0.67424 0.80421 0.4333\n",
      " 0.67881     nan 0.58862 0.49963 0.51617 0.45133 0.43734 0.55097 0.83667\n",
      " 0.51841 0.7716  0.39776] \n",
      "val_spear_lst : [0.30826 0.58551 0.44944 0.27831 0.35598 0.47372 0.35383 0.53586 0.50375\n",
      "     nan 0.42223 0.67692 0.54183 0.27965 0.55431 0.56552 0.77444 0.24117\n",
      " 0.59885     nan 0.51488 0.21542 0.40382 0.07149 0.17602 0.36922 0.74452\n",
      " 0.16126 0.62647 0.22033]\n",
      "\n",
      "Epoch 00058: saving model to /content/saved models/weights.-058-0.36981\n",
      "84/84 [==============================] - 8s 99ms/step - loss: 0.3515 - root_mean_squared_error: 0.1886 - val_loss: 0.3698 - val_root_mean_squared_error: 0.2009\n",
      "Epoch 59/60\n",
      "81/84 [===========================>..] - ETA: 0s - loss: 0.3513 - root_mean_squared_error: 0.1887\n",
      "spearman :\n",
      "train_spearman : 0.55006 | val_spearman : 0.42862 | Learning_Rate : 9.999999974752427e-07\n",
      "train_spear_lst :  [0.44091 0.60046 0.63359 0.40505 0.4793  0.54565 0.3925  0.56637 0.60682\n",
      " 0.18477 0.56899 0.7611  0.59921 0.26594 0.73586 0.67718 0.80412 0.43271\n",
      " 0.67968     nan 0.58962 0.49857 0.51616 0.45335 0.43904 0.55074 0.83591\n",
      " 0.51846 0.77173 0.39809] \n",
      "val_spear_lst : [0.30826 0.58538 0.44944 0.28286 0.35598 0.47378 0.35383 0.53586 0.50375\n",
      "     nan 0.41975 0.67654 0.54183 0.27965 0.55431 0.56552 0.77414 0.24117\n",
      " 0.60357     nan 0.51121 0.21393 0.41001 0.06955 0.17407 0.36822 0.74405\n",
      " 0.15947 0.62701 0.21816]\n",
      "\n",
      "Epoch 00059: saving model to /content/saved models/weights.-059-0.36983\n",
      "84/84 [==============================] - 11s 130ms/step - loss: 0.3515 - root_mean_squared_error: 0.1888 - val_loss: 0.3698 - val_root_mean_squared_error: 0.2009\n",
      "Epoch 60/60\n",
      "82/84 [============================>.] - ETA: 0s - loss: 0.3509 - root_mean_squared_error: 0.1882\n",
      "spearman :\n",
      "train_spearman : 0.55251 | val_spearman : 0.42875 | Learning_Rate : 9.999999974752427e-07\n",
      "train_spear_lst :  [0.44076 0.60085 0.6352  0.40409 0.4787  0.54596 0.39194 0.56565 0.60753\n",
      " 0.26049 0.56885 0.76096 0.59921 0.26594 0.73586 0.67659 0.80439 0.4332\n",
      " 0.67903     nan 0.58892 0.49861 0.51625 0.4535  0.43805 0.55218 0.83611\n",
      " 0.51672 0.77083 0.39648] \n",
      "val_spear_lst : [0.30826 0.58595 0.44944 0.28286 0.35838 0.47529 0.35383 0.53586 0.50375\n",
      "     nan 0.42503 0.67692 0.54183 0.27965 0.55431 0.56552 0.77326 0.24117\n",
      " 0.60163     nan 0.51121 0.21503 0.4033  0.07125 0.17602 0.36534 0.74311\n",
      " 0.15768 0.62869 0.22033]\n",
      "\n",
      "Epoch 00060: saving model to /content/saved models/weights.-060-0.36984\n",
      "84/84 [==============================] - 5s 65ms/step - loss: 0.3512 - root_mean_squared_error: 0.1884 - val_loss: 0.3698 - val_root_mean_squared_error: 0.2009\n"
     ]
    }
   ],
   "source": [
    "# best_model : training a model\n",
    "tf.keras.backend.clear_session()\n",
    "opt = tf.keras.optimizers.Adam(learning_rate = 0.00008)\n",
    "rmse = tf.keras.metrics.RootMeanSquaredError()\n",
    " \n",
    "model.compile(loss = 'binary_crossentropy', optimizer = opt,  metrics = [rmse])\n",
    "history = model.fit(x = [tb_BERT_train, tb_USE_train, tb_RoBERTa_train, tb_XLNet_train, ans_BERT_train, ans_USE_train, ans_RoBERTa_train, ans_XLNet_train],\n",
    "                    y =  y_train.values,\n",
    "                    validation_data = ([tb_BERT_cv, tb_USE_cv, tb_RoBERTa_cv, tb_XLNet_cv, ans_BERT_cv, ans_USE_cv, ans_RoBERTa_cv, ans_XLNet_cv],\n",
    "                                       y_cv.values),\n",
    "                    batch_size = 64, epochs = 60, callbacks = callbacks)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 33
    },
    "colab_type": "code",
    "id": "t5cn15HCCsMr",
    "outputId": "7893e072-2937-4d3a-83a0-7e04551837da"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(0.43928, 28)"
      ]
     },
     "execution_count": 42,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# best_results : (val_spearman, epoch_no)\n",
    "max(print_spearman_fn.spearman_dict['val_spearman']), np.argmax(print_spearman_fn.spearman_dict['val_spearman'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 106,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "QpsYLtas8CW1"
   },
   "outputs": [],
   "source": [
    "# saving best model\n",
    "!cp '/content/saved models/weights.-029-0.37055.data-00000-of-00001' '/content/drive/My Drive/AAIC Course/Personal case study 2  - Google Quest/'\n",
    "!cp '/content/saved models/weights.-029-0.37055.index' '/content/drive/My Drive/AAIC Course/Personal case study 2  - Google Quest/'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 107,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 33
    },
    "colab_type": "code",
    "id": "-XJpPkQ6OdDG",
    "outputId": "11f61a3c-cef8-483f-fcad-902dad9b9e52"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<tensorflow.python.training.tracking.util.CheckpointLoadStatus at 0x7fbfd0853e48>"
      ]
     },
     "execution_count": 107,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.load_weights('/content/saved models/weights.-029-0.37055')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "HnupQMa-IX94"
   },
   "outputs": [],
   "source": [
    "%load_ext tensorboard\n",
    "%tensorboard --logdir /content/logs/"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "0Zo007WZ0Z90"
   },
   "source": [
    "# 6. Error Analysis :"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 163
    },
    "colab_type": "code",
    "id": "Q6sal3yRGOYN",
    "outputId": "d8d03737-2b4f-41ae-a1d5-40fc531f81ad"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train_spear_lst : [0.42345 0.57007 0.61437 0.36343 0.45104 0.51356 0.39067 0.55366 0.57594\n",
      " 0.13085 0.53111 0.7431  0.58853 0.26921 0.7161  0.65808 0.78931 0.4111\n",
      " 0.66014 0.      0.5647  0.41987 0.48362 0.35467 0.33761 0.48572 0.801\n",
      " 0.41396 0.71852 0.31889] \n",
      "\n",
      "val_spear_lst : [0.32541 0.57263 0.42517 0.28532 0.34949 0.41505 0.37109 0.52293 0.49497\n",
      " 0.      0.42076 0.6715  0.53746 0.      0.54913 0.58215 0.77157 0.25675\n",
      " 0.61107 0.      0.48365 0.21315 0.37959 0.08539 0.1659  0.38744 0.75143\n",
      " 0.1649  0.63313 0.22074]\n"
     ]
    }
   ],
   "source": [
    "train_spear_lst, val_spear_lst = print_spearman_fn.all_feat_spearman[30]['train_spearman'], print_spearman_fn.all_feat_spearman[30]['val_spearman']\n",
    "train_spear_lst[np.isnan(train_spear_lst)] = 0.0\n",
    "val_spear_lst[np.isnan(val_spear_lst)] = 0.0\n",
    "\n",
    "print('train_spear_lst :', train_spear_lst, '\\n\\nval_spear_lst :', val_spear_lst)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 82
    },
    "colab_type": "code",
    "id": "4J2HQ1hS5aIq",
    "outputId": "f3d25529-9c92-4ba5-9150-f6aa5699c380"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "featurewise change : [ 0.05264  0.01583 -0.03435  0.07532  0.06585  0.01477  0.12049  0.07294\n",
      "  0.05678 -0.04555  0.06477  0.0679  -0.01409 -0.0979   0.04741  0.12682\n",
      "  0.0246   0.03874  0.07825 -0.03014 -0.00643 -0.02999  0.09617  0.00517\n",
      " -0.0181   0.08856  0.10426  0.02537  0.09874  0.11337]\n"
     ]
    }
   ],
   "source": [
    "# 1. diference between baseline_val_spearman and current model val_spearman\n",
    "baseline_val_spear = np.array([0.27277, 0.5568 , 0.45952, 0.21   , 0.28364, 0.40028, 0.2506 , 0.44999, 0.43819,\n",
    " 0.04555, 0.35599, 0.6036 , 0.55155, 0.0979 , 0.50172, 0.45533, 0.74697, 0.21801,\n",
    " 0.53282, 0.03014, 0.49008, 0.24314, 0.28342, 0.08022, 0.184  , 0.29888, 0.64717,\n",
    " 0.13953, 0.53439, 0.10737])\n",
    "print('featurewise change :', val_spear_lst - baseline_val_spear)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 487
    },
    "colab_type": "code",
    "id": "ijj1MyJG2lHa",
    "outputId": "c81f224d-9d73-4181-b2c9-7cb3e6519221"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1152x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# 2. plot spearman for each feature :\n",
    "plt.figure(figsize = (16, 8))\n",
    "plt.plot(train_spear_lst, 'o')\n",
    "plt.plot(val_spear_lst, 'o')\n",
    "plt.plot(train_spear_lst, label = 'train_spearman')\n",
    "plt.plot(val_spear_lst, label = 'val_spearman')\n",
    "plt.grid()\n",
    "plt.legend()\n",
    "plt.xticks(ticks = range(30), labels = range(1, 31))\n",
    "plt.yticks(ticks = np.arange(0, 1.1, 0.1), labels =  np.arange(0, 1.1, 0.1, dtype = np.float32))\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 50
    },
    "colab_type": "code",
    "id": "OIr3RO803CJ0",
    "outputId": "fed6a7ed-523c-4f1c-c701-f3ef34511f95"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "mean spearman for best target features : 0.4679825 \n",
      "mean spearman for worst target features : 0.069365\n"
     ]
    }
   ],
   "source": [
    "# 3. compute val_spearman with best and worst feature (threshold is val_spearman > 0.20 = good target feature)\n",
    "below_20_idx = [i for i in range(len(val_spear_lst)) if val_spear_lst[i] < 0.2]\n",
    "above_20_idx = [i for i in range(30)]\n",
    "for i in below_20_idx:\n",
    "  above_20_idx.remove(i) \n",
    "\n",
    "print('mean spearman for best target features :', np.mean(val_spear_lst[above_20_idx]), '\\nmean spearman for worst target features :', np.mean(val_spear_lst[below_20_idx]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 261
    },
    "colab_type": "code",
    "id": "fX93WKBD3jBH",
    "outputId": "59df5d81-3db9-47af-a8d2-4fd83181a18c"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "best target features : ['question_asker_intent_understanding' 'question_body_critical'\n",
      " 'question_conversational' 'question_expect_short_answer'\n",
      " 'question_fact_seeking' 'question_has_commonly_accepted_answer'\n",
      " 'question_interestingness_others' 'question_interestingness_self'\n",
      " 'question_multi_intent' 'question_opinion_seeking' 'question_type_choice'\n",
      " 'question_type_compare' 'question_type_definition' 'question_type_entity'\n",
      " 'question_type_instructions' 'question_type_procedure'\n",
      " 'question_type_reason_explanation' 'question_well_written'\n",
      " 'answer_helpful' 'answer_level_of_information' 'answer_satisfaction'\n",
      " 'answer_type_instructions' 'answer_type_reason_explanation'\n",
      " 'answer_well_written'] \n",
      "\n",
      "worst target features : ['question_not_really_a_question' 'question_type_consequence'\n",
      " 'question_type_spelling' 'answer_plausible' 'answer_relevance'\n",
      " 'answer_type_procedure']\n"
     ]
    }
   ],
   "source": [
    "# 4. names of good and bad performing target features\n",
    "print('best target features :', np.array(tar_features)[above_20_idx], '\\n\\nworst target features :', np.array(tar_features)[below_20_idx])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "IxJsfjNzQonX"
   },
   "source": [
    "**6.1. Q. Why these worst features are not performing well?**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 277
    },
    "colab_type": "code",
    "id": "Z8E_d_3QOC5p",
    "outputId": "b42a2063-8c7d-464a-fe20-f301f81c6d4b"
   },
   "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>question_not_really_a_question</th>\n",
       "      <th>question_type_consequence</th>\n",
       "      <th>question_type_spelling</th>\n",
       "      <th>answer_plausible</th>\n",
       "      <th>answer_relevance</th>\n",
       "      <th>answer_type_procedure</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>6079.000000</td>\n",
       "      <td>6079.000000</td>\n",
       "      <td>6079.000000</td>\n",
       "      <td>6079.000000</td>\n",
       "      <td>6079.000000</td>\n",
       "      <td>6079.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>0.004469</td>\n",
       "      <td>0.010035</td>\n",
       "      <td>0.000823</td>\n",
       "      <td>0.960054</td>\n",
       "      <td>0.968626</td>\n",
       "      <td>0.130641</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>0.045782</td>\n",
       "      <td>0.074240</td>\n",
       "      <td>0.020489</td>\n",
       "      <td>0.086926</td>\n",
       "      <td>0.074631</td>\n",
       "      <td>0.225718</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.333333</td>\n",
       "      <td>0.333333</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.333333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.666667</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       question_not_really_a_question  ...  answer_type_procedure\n",
       "count                     6079.000000  ...            6079.000000\n",
       "mean                         0.004469  ...               0.130641\n",
       "std                          0.045782  ...               0.225718\n",
       "min                          0.000000  ...               0.000000\n",
       "25%                          0.000000  ...               0.000000\n",
       "50%                          0.000000  ...               0.000000\n",
       "75%                          0.000000  ...               0.333333\n",
       "max                          1.000000  ...               1.000000\n",
       "\n",
       "[8 rows x 6 columns]"
      ]
     },
     "execution_count": 29,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# -> 1. basics\n",
    "df_train[np.array(tar_features)[below_20_idx]].describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 605
    },
    "colab_type": "code",
    "id": "dfZCjWh1OrVD",
    "outputId": "3f90176b-333d-4c15-8d6f-759759070c3a"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 2160x792 with 6 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# -> 2. pdfs\n",
    "plt.figure(figsize = (30, 11))\n",
    "for i in range(6):\n",
    "  plt.subplot(2, 3, i+1)\n",
    "  column = np.array(tar_features)[below_20_idx][i]\n",
    "  sns.distplot(df_train[column].values, hist = True, kde=True)\n",
    "  \n",
    "  plt.xlabel(column)\n",
    "  plt.ylabel('count')\n",
    "  plt.grid()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 797
    },
    "colab_type": "code",
    "id": "9mmwadeuPYAW",
    "outputId": "d71b9870-7058-4958-907e-e2f6a0121270"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.000000    6013\n",
      "0.333333      48\n",
      "0.500000      11\n",
      "0.666667       4\n",
      "1.000000       3\n",
      "Name: question_not_really_a_question, dtype: int64 \n",
      "\n",
      "0.000000    5948\n",
      "0.333333      86\n",
      "0.666667      26\n",
      "1.000000      11\n",
      "0.500000       8\n",
      "Name: question_type_consequence, dtype: int64 \n",
      "\n",
      "0.000000    6068\n",
      "0.333333       7\n",
      "0.666667       4\n",
      "Name: question_type_spelling, dtype: int64 \n",
      "\n",
      "1.000000    4710\n",
      "0.888889     740\n",
      "0.666667     215\n",
      "0.833333     199\n",
      "0.777778     191\n",
      "0.333333      10\n",
      "0.555556       8\n",
      "0.500000       4\n",
      "0.444444       2\n",
      "Name: answer_plausible, dtype: int64 \n",
      "\n",
      "1.000000    4913\n",
      "0.888889     713\n",
      "0.777778     184\n",
      "0.833333     138\n",
      "0.666667     111\n",
      "0.555556       7\n",
      "0.333333       5\n",
      "0.500000       5\n",
      "0.444444       3\n",
      "Name: answer_relevance, dtype: int64 \n",
      "\n",
      "0.000000    4281\n",
      "0.333333    1252\n",
      "0.666667     323\n",
      "0.500000     123\n",
      "1.000000     100\n",
      "Name: answer_type_procedure, dtype: int64 \n",
      "\n"
     ]
    }
   ],
   "source": [
    "# 3. -> bins\n",
    "for i in np.array(tar_features)[below_20_idx]:\n",
    "  print(df_train[i].value_counts(), '\\n')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "GCAFXFtoInyW"
   },
   "source": [
    "**6.2. Q. Why these best features are performing well?**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 66
    },
    "colab_type": "code",
    "id": "YuCE-Fw1IslP",
    "outputId": "b38fd2ac-aeb8-44c0-f2c6-6288cacd804b"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "top_6 target features : ['question_type_entity' 'question_type_reason_explanation'\n",
      " 'answer_type_reason_explanation' 'question_type_choice'\n",
      " 'answer_type_instructions' 'question_type_instructions']\n"
     ]
    }
   ],
   "source": [
    "# top 6 features\n",
    "top_06_idx = np.argsort(val_spear_lst)[-6:]\n",
    "\n",
    "# names of top 6 features target features\n",
    "print('top_6 target features :', np.array(tar_features)[top_06_idx])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 277
    },
    "colab_type": "code",
    "id": "oiMhDjqeKBZy",
    "outputId": "cbab3ccb-dab6-4395-8f8e-96caf1a479f9"
   },
   "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>question_type_entity</th>\n",
       "      <th>question_type_reason_explanation</th>\n",
       "      <th>answer_type_reason_explanation</th>\n",
       "      <th>question_type_choice</th>\n",
       "      <th>answer_type_instructions</th>\n",
       "      <th>question_type_instructions</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>6079.000000</td>\n",
       "      <td>6079.000000</td>\n",
       "      <td>6079.000000</td>\n",
       "      <td>6079.000000</td>\n",
       "      <td>6079.000000</td>\n",
       "      <td>6079.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>0.065225</td>\n",
       "      <td>0.386385</td>\n",
       "      <td>0.502468</td>\n",
       "      <td>0.284915</td>\n",
       "      <td>0.479547</td>\n",
       "      <td>0.497587</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>0.197582</td>\n",
       "      <td>0.383384</td>\n",
       "      <td>0.407097</td>\n",
       "      <td>0.368826</td>\n",
       "      <td>0.422921</td>\n",
       "      <td>0.423138</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.333333</td>\n",
       "      <td>0.500000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.500000</td>\n",
       "      <td>0.666667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.666667</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.666667</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       question_type_entity  ...  question_type_instructions\n",
       "count           6079.000000  ...                 6079.000000\n",
       "mean               0.065225  ...                    0.497587\n",
       "std                0.197582  ...                    0.423138\n",
       "min                0.000000  ...                    0.000000\n",
       "25%                0.000000  ...                    0.000000\n",
       "50%                0.000000  ...                    0.666667\n",
       "75%                0.000000  ...                    1.000000\n",
       "max                1.000000  ...                    1.000000\n",
       "\n",
       "[8 rows x 6 columns]"
      ]
     },
     "execution_count": 70,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# -> 1. basics\n",
    "df_train[np.array(tar_features)[top_06_idx]].describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 610
    },
    "colab_type": "code",
    "id": "bppZJiPVKT-1",
    "outputId": "8c6ccfec-b98e-4d92-aa87-17967199d411"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 2160x792 with 6 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# -> 2. pdfs\n",
    "plt.figure(figsize = (30, 11))\n",
    "for i in range(6):\n",
    "  plt.subplot(2, 3, i+1)\n",
    "  column = np.array(tar_features)[top_06_idx][i]\n",
    "  sns.distplot(df_train[column].values, hist = True, kde=True)\n",
    "  \n",
    "  plt.xlabel(column)\n",
    "  plt.ylabel('count')\n",
    "  plt.grid()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 700
    },
    "colab_type": "code",
    "id": "sw2oJ47bKbwV",
    "outputId": "5e72c57d-8a09-4ba7-fa1b-968e917e927d"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.000000    5348\n",
      "0.333333     391\n",
      "0.666667     172\n",
      "1.000000     135\n",
      "0.500000      33\n",
      "Name: question_type_entity, dtype: int64 \n",
      "\n",
      "0.000000    2456\n",
      "0.333333    1316\n",
      "1.000000    1171\n",
      "0.666667    1027\n",
      "0.500000     109\n",
      "Name: question_type_reason_explanation, dtype: int64 \n",
      "\n",
      "1.000000    1899\n",
      "0.000000    1892\n",
      "0.666667    1107\n",
      "0.333333    1038\n",
      "0.500000     143\n",
      "Name: answer_type_reason_explanation, dtype: int64 \n",
      "\n",
      "0.000000    3363\n",
      "0.333333    1064\n",
      "1.000000     875\n",
      "0.666667     683\n",
      "0.500000      94\n",
      "Name: question_type_choice, dtype: int64 \n",
      "\n",
      "0.000000    2244\n",
      "1.000000    1917\n",
      "0.666667    1007\n",
      "0.333333     772\n",
      "0.500000     139\n",
      "Name: answer_type_instructions, dtype: int64 \n",
      "\n",
      "0.000000    2171\n",
      "1.000000    1985\n",
      "0.666667    1122\n",
      "0.333333     652\n",
      "0.500000     149\n",
      "Name: question_type_instructions, dtype: int64 \n",
      "\n"
     ]
    }
   ],
   "source": [
    "# 3. -> bins\n",
    "for i in np.array(tar_features)[top_06_idx]:\n",
    "  print(df_train[i].value_counts(), '\\n')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "QS_Bii7MMGeW"
   },
   "source": [
    "**6.3.  what are the types of data points giving best and worst predictions?**\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 128,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "j_EhwjEVMUiw"
   },
   "outputs": [],
   "source": [
    "y_pred_val = model.predict([tb_BERT_cv, tb_USE_cv, tb_RoBERTa_cv, tb_XLNet_cv, ans_BERT_cv, ans_USE_cv, ans_RoBERTa_cv, ans_XLNet_cv]) # cv_prediction\n",
    "final_pred_val = binned_out(y_pred_val) # binning the model output\n",
    "\n",
    "# compute no. of zeros in each datapoint prediction\n",
    "zero_lst = []\n",
    "for i in range(len(y_cv)):\n",
    "  zero_lst.append(len(np.where((y_cv.values - final_pred_val)[i] == 0.0)[0]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 130,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 277
    },
    "colab_type": "code",
    "id": "gTupm_fYYNEh",
    "outputId": "3c51e5a2-8baa-4e5e-dee0-40a8c0e77762"
   },
   "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>zero_lst</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>730.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>15.582192</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>3.026986</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>7.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>14.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>16.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>18.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>26.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         zero_lst\n",
       "count  730.000000\n",
       "mean    15.582192\n",
       "std      3.026986\n",
       "min      7.000000\n",
       "25%     14.000000\n",
       "50%     16.000000\n",
       "75%     18.000000\n",
       "max     26.000000"
      ]
     },
     "execution_count": 130,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 1. basics \n",
    "pd.DataFrame({'zero_lst' : zero_lst}).describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 191,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 752
    },
    "colab_type": "code",
    "id": "bBEnYMyeYVzD",
    "outputId": "0018a3bb-2b4c-4d01-f6c7-72b107b1fa73"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "best_10 title_body : ['javascript equivalent of php __invoke i am developing a small framework (in js) and for esthetic reasons and simplicity i was wondering if there could be a way to implement something like php \"__invoke\". for example: var myclass = function(config) { this.config = config; this.method = function(){}; this.execute = function() { return this.method.apply(this, arguments); } } var execcustom = new myclass({ wait: 100 }); execcustom.method = function() { console.log(\"called method with \"+arguments.length+\" argument(s):\"); for(var a in arguments) console.log(arguments[a]); return true; }; execcustom.execute(\"someval\",\"other\"); desired way to execute: execcustom(\"someval\",\"other\"); any ideas? thanks.'\n",
      " 'jquery - select exact class structure i am working on some functionality for a website. i am designing, and i am kind of stuck on something small. i want to hide divs that contain an exact class structure. for example, if i give it the class selector \".class1.class2.class3\", it will only hide elements that have exact class structure. what i am doing now would hide elements like \".class1.class2.class3.class4\", and i do not want that. any help would be greatly appreciated!'\n",
      " 'how to scale entire document including maths symbols? i am trying to make a cheat-sheet for my math class. i wonder is there a way to scale the entire document so that i can fit more formulas and theorems into it. it does not need to be super tiny though. i tried tiny font, but it still waste too much space. have anyone done it before could share me some experiences? thanks in advance.'\n",
      " 'how to convert particles to mesh i need to convert a particle system to a mesh object, so i can export it to unity 3d. how could i do this?'\n",
      " 'how to get src from cdata in rss? i am fetching data from rss feed with magpie. $rss[description] contains cdata with html elements: <![cdata[ <div> <a href=\\'url\\'> <img src=\\'img_url\\' alt=\\'aaa\\' title=\\'bbb\\' border=\\'0\\' width=\\'116\\' height=\\'116\\'> </a> </div>]]> some other text how can i fetch \"img_url\" from this? is preg_match() the only way? maybe i can use simplexml?'\n",
      " 'teletext subtitles on romanian television there is a teletext feature on the televisions in the hotel rooms. in england there was a way to get subtitles to appear by going to the teletext interface and entering 888. as far as i know this is no longer available. in romania, is there a special code you can enter into the teletext interface to enable subtitles? i have tried 888 but it does not do anything.'\n",
      " 'can i delete my fork of a bitbucket repository, after the owner has accepted and merged my pull request? i created a fork of a public repository on bitbucket, with the sole purpose of committing a pull request that fixed some bug. the pull request was accepted and merged. now, i want to delete the repository, that has outlived its purpose. if i do that, will the changes be lost? there is already a question about this same issue, but for github. apparently, on github you can delete the repository before the request has been even merged, as it gets automatically saved in the target repository. i guess that on bitbucket, things will be similar, but i wanted to be sure.'\n",
      " 'plot has red \\'film\\', no error message i am encountering really strange behavior: i wrote a function that plots values some data. after executing the code the resulting plot has the red \\'error\\' colour, but only momentarily (maybe 0.1 sec). there is no error message, so the plot seems fine. however, when i export the graphic, the red colour is exported too. there seems no way to get the graphic without the red. i am (in this case) not interested in what caused the error message but want to export the graphic without te red colour. does anyone know how to do this? i tried using quiet[] but that did not help. my code: plotbetaci[whole_,everyminute_]:= module[ {len,linearmodels,betas,lows,highs}, len=length[everyminute]; linearmodels=quiet@linearmodelfit[transpose[{#,whole}],x,x]&amp;/@everyminute; betas=transpose[{range[len],#[\"parameterconfidenceintervaltableentries\"][[2,1]]&amp;/@linearmodels}]; lows=transpose[{range[len],#[\"parameterconfidenceintervaltableentries\"][[2,3,1]]&amp;/@linearmodels}]; highs=transpose[{range[len],#[\"parameterconfidenceintervaltableentries\"][[2,3,2]]&amp;/@linearmodels}]; listplot[{betas,sequence@@transpose[{lows,highs}]},joined->{false,sequence@@constantarray[true,len]},plotstyle->black,plotmarkers->{graphics[{line[{{-1,-1},{1,1}}],line[{{-1,1},{1,-1}}]},imagesize->10],sequence@@constantarray[graphics[{line[{{0,0},{0.04,0}}]},imagesize->10],len]},imagesize->large,plotrange->{automatic,{automatic,1.5}}] ] the command to actually plot is (this produces the plot i am talking about) plotbetaci[rmssd, everyminutermssd] rmssd is a list of values. everyminutermssd has the same length, but depth 10.'\n",
      " 'what does mathematics have to do with programming? i just started a diploma in software development. right now we are starting out with basic java and such (so right from the bottom you might say) - which is fine, i have no programming experience apart from knowing how to do \"hello world\" in java. i keep hearing that mathematics is pertinent to coding, but how is it so? what general examples would show how mathematics and programming go together, or are reliant on one another? i apologize of my question is vague, i am barely starting to get a rough idea of the kind of world i am stepping into as a code monkey student...'\n",
      " 'get current activity with android i have a method that scans the position and gets called on every activity i have, everytime an actions trigger i need to show an alertdialog, this is the code: alertdialog.builder builder = new alertdialog.builder(mmainactivity.this); builder.setmessage(\"test\") .setcancelable(false) .setpositivebutton(\"go\", new dialoginterface.onclicklistener() { public void onclick(dialoginterface dialog, int id) { intent intent = new intent(mmainactivity.this,locationdetails.class); intent.putextra(\"placeid\",1); startactivity(intent); } }) .setnegativebutton(\"cancel\", new dialoginterface.onclicklistener() { public void onclick(dialoginterface dialog, int id) { dialog.cancel(); } }); alertdialog alert = builder.create(); alert.show(); but how can i create the alert on the current activity? because now it only gets created on alertdialog.builder(mmainactivity.this);'] \n",
      "\n",
      "worst_10 title_body : [\"what platform can i use to establish a users support forum? i was asked to develop a users support community forum for a small and young hi-tech company. this kind of forums is very popular with companies these days and is a great alternative to the official product support lines. an example of what we look for is the analog devices' engineer zone forums. i am totally noob to this area so any advice will be appreciated - how to start developing such a site, what platforms are available (preferably open-source/free) and what are the advantages and disadvantages of these platforms. is a stackexchange style forum appropriate for such site? * if you think there is a better se site to post this question, please let me know.\"\n",
      " 'potential php problems with ee 1.13 or ce 1.8 my client is on an old server running php 5.2.17, should i encourage him to upgrade to 5.3 or will it be ok?'\n",
      " 'multiple keys via hkdf - whats better, one or two applications of hkdf-extract assume for the sake of the question that i have two variable-length bit strings, each with 128 bit cryptographic randomness, and i want to extract two 128 bit keys via hkdf-sha256. which alternative is better (if any), and why? use a single hkdf-extract on the concatenation of the two strings, and two hkdf-expands with different info strings to get two 128 bit keys. use two hkdf-extract operations, one for each bit string, and use a single hkdf-expand on each to get two 128 bit keys. or in other words, is it better to hkdf-extract on a longer ikm string and use multiple hkdf-expands, or is it better to use hkdf on independent but shorter ikms. my intuition tells me that, if my randomness strings are really as good as i claim, then two independent hkdf extracts are better, but using a single one on the concatenation is, in practice, just as safe, and safer if my input randomness is not actually as good as assumed, so two hkdfs might be more robust in practice.'\n",
      " 'cs paper rejected for lack of technical contribution; what does this mean, and how can i fix it? i recently got two papers rejected with the reviewers asserting a lack of technical contribution. they contained good ideas, at least in my eyes, and used common methodologies in the experimentation. so my question related to the cs field is, what is the difference between a good idea that was tested and a technical contribution? what should an idea have to become a technical contribution? another observation i had is that other papers, that got accepted, were actually worse from the research methodologies and were previously published in similar form (so the idea was not very original) at that conference, so the contribution of these papers is much lower than mine would have been (again its just from my perspective). so i dont understand why those papers got accepted and if the reviewers decisions actually make sense, what is it that makes such papers accepted? i aim at enhancing my now rejected papers because i still think they contain original ideas that should be published, i just dont know where to start.'\n",
      " 'open problems/questions in representation theory and around? what are open problems in representation theory? what are the sources (books/papers/sites) discussing this? any kinds of problems/questions are welcome - big/small, vague/concrete. some estimation of difficulty and importance, as well as, small description, prerequisites and relevant references, ... are welcome. to the best of my knowledge, there are no good lists of representation theory problems on the web. e.g. the sites below contain lots of unsolved problem in other areas, but not in representation theory: http://en.wikipedia.org/wiki/unsolved_problems_in_mathematics http://garden.irmacs.sfu.ca/ http://maven.smith.edu/~orourke/topp/ mo questions also discuss other fields, but not representation theory: what are the big problems in probability theory? what are some open problems in algebraic geometry? what are some open problems in toric varieties? more open problems open problems with monetary rewards open problems in euclidean geometry? open questions in riemannian geometry what are some of the big open problems in 3-manifold theory? open problems in continued fractions theory'\n",
      " 'syntax highlighting language hints do you think it would be worthwhile to provide hints as to what language to use for the syntax highlighting? sometimes i find the highlighting on sql or vb.net answers is more distracting than helpful; for example: converting mysql select to postgresql retrieving data from a vb.net arraylist of objects to pick the two i have been looking at recently.'\n",
      " 'redirecting users to referrer page after logging in using custom login form i am using a custom wordpress login form using the following code. i have this form in the sidebar of my site: <?php if ( ! is_user_logged_in() ) { // display wordpress login form: $args = array( aredirect\\' => admin_url(), \\'form_id\\' => \\'loginform-custom\\', \\'label_username\\' => __( \\'username\\' ), \\'label_password\\' => __( \\'password ), \\'label_remember\\' => __( aremember me\\' ), \\'label_log_in\\' => __( \\'log in\\' ), aremember\\' => true ); wp_login_form( $args ); } else { // if logged in: echo \\'you are already logged in\\'; wp_loginout( home_url() ); // display \"log out\" link. echo \" | \"; wp_register(\\'\\', \\'\\'); // display \"site admin\" link. } ?> and in the functions.php page, i am using the following code to redirect users to the homepage after they log in: function my_login_redirect( request, $user ){ //is there a user to check? if( isset( user->roles ) ) { //check for admins if( in_array( \"administrator\", $user->roles ) ) { // redirect them to the default place return $redirect_to; } else { return home_url(); } } else { return $redirect_to; } } add_filter(\"login_redirect\", \"my_login_redirect\", 10, 3); my question is:, is there a way to redirect users to the areferrer page (page from where they logged in)\\' after they login by making changes to the above code or perhaps using a better code? thanks in advance.'\n",
      " 'when to use tools vs custom development? to learn dependency injection in a current side-project, i am writing my own dependency injection container. but this has lead me to wonder, at what point is it worthwhile to use a third party product vs \"roll my own\"? what would be the kind of questions you would ask to figure this out?'\n",
      " 'is there any granular way to change decisions made in me1/me2 to see different outcomes in me3? in me1 and me2, i had to make decisions not really understanding their impact on the series as a whole. now that i am partway through me3, i find myself curious as to what would have happened had i made a different choice previously. what if i would done x? what if i would let y live/die? i can not just replay me3, import my me2 save, and make different choices to see these outcomes - i would have to make changes to the way i played me/me2. likewise, importing no save would make wholesale changes to my experience that i am not sure i would appreciate. besides starting the entire series over again, and running through each game dozens of times to explore the various permutations, what can i do to change the way things play out in me3? do i get the option to edit the imported starting state on new game+? i am on the xbox, so save file editing or importing from the internet is likely to be at least challenging, if not impossible.'\n",
      " 'red flags of unpaid it internship i read the following question: tips for a first year cs student looking for a summer internship to gain experience? but rather than how to get and/or look for internships, my questions concerns of how to filter companies looking for free labor to do their website vs companies that might not offer a paid check, but it will offer you mentorship and skills. i have a cousin that is a college freshman and she is looking for sofware development internships, but at this point, she is desperately looking for any internship that is it related. besides the big software names and elite small software shops, how do you recognize a good internships in non-it-companies(large or small) doing it maintenance vs the bad ones? if you do not have work experience, what questions a college student should ask to prevent ripoff? are there any red flags that you can spot before accepting an offer?'] \n",
      "\n",
      "best_10 answer : ['just return a function that will form the public interface: function myclass(config) { var pubif = function() { return pubif.method.apply(pubif, arguments); }; pubif.config = config; pubif.method = function() { }; return pubif; } the rest of the code remains the same.'\n",
      " 'i you want to match objects that only have exactly those three class names and you want to be insensitive to the order of the class names or the amount of white space between them, then you can do it like this: $(\\'.class1.class2.class3\\').filter(function() { return(this.classname.replace(/class1|class2|class3/g, \"\").replace(/\\\\s*/g, \"\") == \\'\\'); }); if you want to only exclude a specific other class, then you can do it like this: $(\\'.class1.class2.class3\\').not(\\'.class4\\')'\n",
      " 'the best way to create some space in your document is to use a combination of geometry (for setting/removing margin dimensions) and some standard documentclass font settings: \\\\documentclass[10pt]{article}% default is 10pt font however, in terms of creating a cheat-sheet that should condense a number of pages on (say) a two-sided single piece of stock, i would suggest creating a document as per usual and then creating a second document that includes the pages of the first in \"n-up\" style, depending on your page count. let me explain... consider your original straight-forward document that has a bunch of content in it, including formulas and nifty how-to is: cheat-material.tex \\\\documentclass{article} \\\\usepackage[margin=1cm]{geometry}% http://ctan.org/pkg/geometry \\\\usepackage{blindtext}% http://ctan.org/pkg/blindtext \\\\usepackage[english]{babel}% http://ctan.org/pkg/babel \\\\begin{document} \\\\blinddocument% creates a 8-page document in the current style \\\\end{document} the above is just a mock-up of your document using the blindtext package. it creates 8 pages of wholesome cheat-sheet material. now i create my second document using the pdfpages package to insert output of cheat-material.tex in n-up style (i chose 4-up, since 8 is divisible by 4): cheat-sheet.tex \\\\documentclass{article} \\\\usepackage{pdfpages}% http://ctan.org/pkg/pdfpages \\\\begin{document} \\\\includepdf[nup=4,pages=-,frame]{cheat-material} \\\\end{document} pages=- includes all pages of cheat-material.pdf, and frame adds a page border around each included page. this way you are not bound as much by trying to modify things - you are merely creating a document (cheat-material) and squeezing it as needed into a specific shape (cheat-sheet). this is especially helpful if you have multiple pages that you want to condense, as seems to be the case for you.'\n",
      " \"the following assumes that you are using 'particle system-> render-> object -> dupli object'. option 1) you can press the convert button for the particles modifier. option 2) if you would like to keep the particle emitter intact, you can use the amake duplicates real' tool shift+ctrl+a. this will create object particles at every particle position just as if you applied the modifier but the particle emitter and its particle system will still be intact in case you would like to make further changes.\"\n",
      " 'cdata you have is a string containing html. so first treat it as text, but since this text is meant to contain html, parse this text using solution appropriate for parsing html. in other words: use appropriate tool (html parser) for the job (parsing html). there are already existing solutions for parsing html in php - eg. see answers to this questions: how to parse and process html with php? robust, mature html parser for php. so, basically, you already are able to get the html string from your xml. now parse the html and get source of the image using some of the solutions mentioned above. using preg_match() for parsing html is not a good idea, as it would need to be very complex to do a simple thing html parser is suitable for.'\n",
      " 'bad news. while 888 will get you the teletext subtitles in greece, belgium and other countries, and austria is 777, for example, romania is one of the few eu members to have none for teletext subtitling. source: state of subtitling access in eu report - european federation of hard of hearing people (page 20).'\n",
      " 'i would say if things work as expected on github, then they would work on bitbucket. if the pull request was \"accepted and merged\", then nothing you do to your personal repository will modify the repository that has merged in your changes.'\n",
      " 'the pink box shows a formatting error. you can disable highlighting of formatting errors for a specific object using style: style[ graphics[{disk[], garbage}], autostyleoptions -> {\"highlightformattingerrors\" -> false}] you can also change the setting globally using the preferences dialog (edit - preferences - messages - formatting error indications), or just for the current session using: setoptions[$frontendsession, autostyleoptions -> {\"highlightformattingerrors\" -> false}]'\n",
      " 'two examples that immediately come to mind are: functions - the idea of applying a transformation to input variables to produce an output variable is strongly rooted in mathematics. the notion of passing a function around as a parameter to another function even more so. in general, the notion of abstract thinking associated with programing parallels mathematics quite closely. bitmasks - this common programming approach to solving problems requires at least a basic understanding of boolean algebra to even grasp the concept.'\n",
      " 'have this alertdialog logic in a public class and make call using a method with passing context . for example: showalert(youractivitycontext); use the context as your current activity.'] \n",
      "\n",
      "worst_10 answer : ['a stackexchange style forum would be appropriate for this site (although there are many many other support forums out there). for a list of clones see this meta.stackexchange.com answer. the stackexchange model is designed around users helping users, but if your staff can take part then that is even better. if i were you i would tweak the code so that users that are staff are clearly identified as such.'\n",
      " 'i suggest to try with php 5.4. if it is ee installation please talk to support first as 5.4 is not officially supported yet (not fully tested) but works correctly since 1.13/1.8 based on the feedback i got.'\n",
      " 'realistically, it probably does not matter, if all of your premises are accurate. if it were me, i would probably concatenate the inputs, then apply a hkdf to the concatenation to derive two keys -- but honestly, it is unlikely to matter. this is very unlikely to be the weakest link in your system. pick something that is easy to implement and easy to understand, and move on: focus your energy on some other aspect of your system.'\n",
      " 'since you ask those questions, i assume you are relatively new to cs research. i do not know if you have previous research experience in any other scientific area, but let is assume you are new to research all together. correct me if i am wrong. when your paper gets rejected, the feeling of rejection sucks. it sucks a lot. but in research you have to deal with this rejection and move-on. it does not matter if other people is works that you consider worse have been accepted. maybe they are better than yours maybe not. but that is really irrelevant. focus on your work and how to improve it. the questions you should be asking: is this problem you are trying to solve interesting? is your solution substantially better than previous state-of-the-art? are you certain that you described related and recent literature thoroughly? are you up-to-date on related literature? did you make clear the benefits on your approach? is your paper well presented and written? have you explained your approach thoroughly (with pseudocode, figures) and not just with words? were your experiments thorough? did you use many datasets for your experiments? have you used the datasets that the related papers use or at least similar if not bigger in size and scope? have you compared your results with previous approaches? have you many charts demonstrating the various aspects of your solution? have you advertised your work by finding many test-cases for your problem? if any of those answer is no, then you must turn them to yes, before resubmitting your work. you must understand that if your paper or the problem you try to solve is boring, then it gets rejected. it is as simple as that. in that cases, the most usual reason for rejection is \"not enough contribution\". so, probably it is also your presentation that needs to be fixed. also, it seems rather strange that although you are probably new to research, you have submitted two papers at the same time. perhaps this is too ambitious and it would be better to focus on one of them, improve it and resubmit it. perhaps, they should be sent to different venues that are more tailored for each one of them. and last but not least. do not worry. it gets better with time. both the way you handle rejection and the way you write papers. so, improve what needs to be fixed and everything will work out.'\n",
      " 'there are many open problems on modular representation theory of finite groups. there is a well-known list of problems of r. brauer which date to around 1960, about ordinary and modular representations of finite groups. there are other major conjectures ( broue is abelian defect group conjecture has already been mentioned in comments). there are alperin is weight conjecture, dade is projective conjecture, the isaacs-navarro conjecture, and several related conjectures. this is only the tip of the iceberg (and only in one corner of representation theory).'\n",
      " 'scavenger is answer is a good idea, but # might be something that is valid in the code like a pre-processor command in c++ that can be part of the actual code sample. i suggest to have something that is specified before the code block in a special way. the code block is indented with 4 spaces, so the language specifier could be indented 3 spaces (and only mean something when appearing immediately before a code block). this follows the simplicity in the markup rules that already exists. &nbsp;&nbsp;&nbsp;vb.net &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;dim myvariable as string \\'here is the first code line &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;myvariable = \"hello world\"'\n",
      " 'the key part is right here-- aredirect\\' => admin_url(), in your first block of code. whatever you set that to is where you will redirect. i do not see a reason for the second block if you are redirecting everyone. there is no function that i can find for \"this page\" so you probably want $_server[ arequest_uri\\'] $args = array( aredirect\\' => esc_url($_server[ arequest_uri\\']), \\'form_id\\' => \\'loginform-custom\\', \\'label_username\\' => __( \\'username\\' ), \\'label_password\\' => __( \\'password\\' ), \\'label_remember\\' => __( aremember me\\' ), \\'label_log_in\\' => __( \\'log in\\' ), aremember\\' => true ); wp_login_form( $args ); be careful with $_server vars: http://markjaquith.wordpress.com/2009/09/21/php-server-vars-not-safe-in-forms-or-links/'\n",
      " 'to really understand a tool, i do like to create my own version of said tool. it helps me understand what the tool is actually doing and the hard parts of the problem it is trying to solve. understanding those things can help you make better decisions about how to use the tool. that being said i would rarely recommend using the roll-your-own over the tool in a real project. a few reasons for this are: your version has only had you use and test it it is hard to hire people that have knowledge of your custom tool you are the only person who can fix it if there is a problem'\n",
      " 'theoretically your best bet would be using a save game editor to change the flags regarding those pre-me3 decisions and outcomes. there was one for me2, but i admit i have not checked if me3 has one. alternately, i recall an archive of all possible me1 variant ending saves that people could import into me2 to get the start they wanted, hang on.. bingo. your second best bet would probably be to use that site, download the save you want to import, play through, download next save, etc.'\n",
      " 'fair warning - in my opinion, her biggest detriment is years of college experience. while it is quite true that you can be a great programmer and be a college freshman - big companies with established internship programs that offer the kind of things you are speaking of typically have a set of hardened criteria for reviewing applicants. one of them is years of college experience and most companies do not readily accept interns that are freshman - the earliest is usually after sophomore year. this is partially because interns - even if they are free - are an investment. they need computers, space to work, and the time of a senior person to supervise them. the cost of this has to be weighed against the value the intern will bring to the team. imo - most interns are a good deal, but you can not take everyone so a company has to find some set of criteria. so, knowing your cousin is a freshman tells me that if she wants a technical internship, she is already going to have to get creative - it is time to start using every under-the-wire resource and personal connection. you will want to maximize your contact with people who will see working with her as a plus - and that is a heck of a lot easier with people who know your cousin. the other venue that is always worth maximizing is her college is internship system. most colleges have internal internship job boards where the college and the internship providers have worked together to build a set of opportunities for students. in these cases, the companies realize that they need to give interns a good opportunity. i am assuming that at this point, though, your cousin has tried everything and she is really scrounging - going to just about every job board and trying just about every opportunity. assuming she gets an interview, she will want to remember that interviewing is as much about checking out the company as it is about showing potential. here is some things i would look for when going on an internship interview (now that i am well past my intern years and i know from the other side exactly what a manager can do with an intern!). i am keeping it broad as it is a heck of a lot more than just software work! what will i be working on? - good signs are work that involve either making things work or verifying that they are working. bad signs are doing paperwork while other people make things work. having to service customer problems (front line tech support) is not all bad, it may all she can snag, but if that is the case, she should be getting paid. who will i be working for? - preferably it is someone on the interview list or someone on the interview list is \"sample manager\" - ie a manager who needs interns, but the company may assign the candidate to a similar but different manager in the end - it is just that managers have to handle pools of interviews and often matchmaking is done later on. ask this manager what they expect from an intern and what they are willing to give. good signs are \"i expect you to take responsibility\", \"i expect you to ask lots of questions\", \"i expect you will keep me appraised of your status\", \"i expect that by the end of the internship you will be able to take on a 3-4 week chunk of work and work with minimal assistance\". bad signs are - \"you will follow me around and do exactly what i tell you\", or \"i will not have time to give much guidance, and i do not have anyone that can help, we will have to figure it out as we go\". how big is the technical team? - hopefully it is a team of 5-10 people with varying degrees of expertise, with at least a few senior people. also - how much time will the senior people have to spare to answer questions? good = 15 minutes a day, with a known rampup of longer than that early on. bad = \"what senior people?\" or \"it is just you\". what are the major goals of the team? what are the key ingredients to success on this project? - this is my particular way (when i do volunteer work) of figuring out whether i have found people i can work with. if it takes 20 minutes to answer this question, we have a failure. even if i am working alone, with a non-technical person, i may learn a lot of good stuff. but that requires that the people i am working with have a real clue of what they want to get done. it is amazing how many people really do not know what the problem they are trying to solve is, or what qualifies as a solution. without that core vision, no matter how technical the team is, it is likely that the internship will be frustrating. all that said - seriously - do not feel awful if an internship does not happen in the summer of freshman year. i had pretty great grades, 3 languages under my belt and a programmer mother and i still did not swing an internship my freshman summer. i used my touch typing skills to qualify as a temp and worked answering phones, typing and filing microfiche (boy am i old!) my freshman summer... at least i got to hang out in a.c.!'] \n",
      "\n"
     ]
    }
   ],
   "source": [
    "# 2. best and worst datapoints\n",
    "top_10_dp_idx = np.argsort(zero_lst)[-10:][::-1]\n",
    "bottom_10_dp_idx = np.argsort(zero_lst)[:10]\n",
    "\n",
    "print('best_10 title_body :', np.array(title_body_cv)[top_10_dp_idx], '\\n')\n",
    "print('worst_10 title_body :', np.array(title_body_cv)[bottom_10_dp_idx], '\\n')\n",
    "print('best_10 answer :', np.array(answer_cv)[top_10_dp_idx], '\\n')\n",
    "print('worst_10 answer :', np.array(answer_cv)[bottom_10_dp_idx], '\\n')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 189,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 82
    },
    "colab_type": "code",
    "id": "gK2lcHpsk1kW",
    "outputId": "1e115375-2822-4761-9e80-b88fe1729c40"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "best_10 title_body : [90, 80, 74, 30, 56, 74, 121, 154, 115, 100]\n",
      "worst_10 title_body : [128, 32, 187, 190, 155, 64, 231, 60, 195, 159]\n",
      "best_10 answer : [41, 68, 249, 84, 135, 54, 42, 61, 82, 28]\n",
      "worst_10 answer : [70, 38, 78, 410, 86, 103, 102, 113, 88, 932]\n"
     ]
    }
   ],
   "source": [
    "# 3. checking word length of best and worst datapoints\n",
    "print('best_10 title_body :', [len(i.split(' ')) for i in np.array(title_body_cv)[top_10_dp_idx]])\n",
    "print('worst_10 title_body :', [len(i.split(' ')) for i in np.array(title_body_cv)[bottom_10_dp_idx]])\n",
    "print('best_10 answer :',  [len(i.split(' ')) for i in np.array(answer_cv)[top_10_dp_idx]])\n",
    "print('worst_10 answer :',  [len(i.split(' ')) for i in np.array(answer_cv)[bottom_10_dp_idx]])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "WqdGrIm24Lv-"
   },
   "source": [
    "# Observations :\n",
    "1. All of the worst performing target features are highly imbalanced.\n",
    "2. Hence some of them are NOT at all learnt by model or model has OVERFIT them on train data.\n",
    "3. 3 out of 9 answer related features (33%) are worst learning target features.\n",
    "4. Top performing features are nicely distributed, mostly their mean and median values are mutually close.\n",
    "5. 2 out of 6 top performing targets are answer related.\n",
    "\n",
    "# Conclusion :\n",
    "1. Use (over)sampling techniques. SMOGN is very expensive because we have multi-target regression task, best suitable for single target regression task.\n",
    "2. Train 6 separate models for each of the worst target features.\n",
    "3. We need more data for good predictions.\n",
    "4. BERT, USE, RoBERTa, XLNet are pretrained on another datasets than stackexchange, all embeddings can be fine tune on stackexchange properties data, to get better improved results. [Note : pretraining or fine_tuning SOTA sentence embeddings is very expensive task.]\n",
    "\n",
    "# Difference between baseline model and final_model :\n",
    "1. 'answer_well_written' target feature is better learnt than baseline_model, hence we found improvement in val_score.\n",
    "2. for each taget_feature spearman_score is slightly or moderately improved."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "3-OAQC6RiTaO"
   },
   "source": [
    "### # Final Results :\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 180
    },
    "colab_type": "code",
    "id": "LtaK-hhNht_u",
    "outputId": "09302f7b-f444-430e-8a95-e60c1f3c33a9"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+------------------------+----------------+--------------+\n",
      "|         Model          | train_spearman | val_spearman |\n",
      "+------------------------+----------------+--------------+\n",
      "|     baseline_LSTM      |     0.4995     |   0.34899    |\n",
      "|          BERT          |    0.50081     |   0.39536    |\n",
      "|          USE           |    0.52245     |    0.4366    |\n",
      "|        RoBERTa         |    0.48059     |   0.41772    |\n",
      "|         XLNet          |    0.52237     |   0.37525    |\n",
      "| BERT_USE_RoBERTa_XLNet |     0.5068     |   0.43924    |\n",
      "+------------------------+----------------+--------------+\n"
     ]
    }
   ],
   "source": [
    "\n",
    "from prettytable import PrettyTable\n",
    "x = PrettyTable()\n",
    "\n",
    "x.field_names = [\"Model\", \"train_spearman\", \"val_spearman\"]\n",
    "x.add_row([\"baseline_LSTM\", 0.4995, 0.34899])\n",
    "x.add_row([\"BERT\", 0.50081, 0.39536])\n",
    "x.add_row([\"USE\",  0.52245, 0.4366])\n",
    "x.add_row([\"RoBERTa\", 0.48059, 0.41772])\n",
    "x.add_row([\"XLNet\", 0.52237, 0.37525])\n",
    "x.add_row([\"BERT_USE_RoBERTa_XLNet\", 0.5068, 0.43924])\n",
    "\n",
    "print(x)"
   ]
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "collapsed_sections": [],
   "name": "3.2. final_modelling.ipynb",
   "provenance": []
  },
  "kernelspec": {
   "display_name": "Python 3",
   "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": "3.7.3"
  },
  "widgets": {
   "application/vnd.jupyter.widget-state+json": {
    "01c5462b34a741deb508c4a6aa9c1308": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "FloatProgressModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "FloatProgressModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "ProgressView",
      "bar_style": "success",
      "description": "100%",
      "description_tooltip": null,
      "layout": "IPY_MODEL_fc3ee3f19d244a48bd1d145efd9d9264",
      "max": 23,
      "min": 0,
      "orientation": "horizontal",
      "style": "IPY_MODEL_15b4a5793e464151844a3ac0c6cacad5",
      "value": 23
     }
    },
    "03371146acc44a65981ee3fcbc91377e": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "033a5e21fde447b592f6a655270e9704": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "DescriptionStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "DescriptionStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "description_width": ""
     }
    },
    "0375dbb6cfd14ac59a93b531a2a974f9": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "04b1246c315743e885e28770d34fd83f": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "DescriptionStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "DescriptionStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "description_width": ""
     }
    },
    "055725b181bb4d25b7a083b6cd7b373c": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "ProgressStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "ProgressStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "bar_color": null,
      "description_width": "initial"
     }
    },
    "065a222b4ac14adb8a332e8b6f2cf2a8": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "DescriptionStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "DescriptionStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "description_width": ""
     }
    },
    "070ba289643e4bdb8873cc225aad2326": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "07236e145d2844fd9b129dd3ee7e1718": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "078378bbad8a41f697ef44eacfb88037": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "07e5725f24544b1580a3ae1699973779": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "09bba54f402943c38fae70019553586c": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HBoxModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HBoxModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HBoxView",
      "box_style": "",
      "children": [
       "IPY_MODEL_4df465b390af4c0a84fc7841d6e35c5d",
       "IPY_MODEL_f87b02922496412183d3ff676fa0318c"
      ],
      "layout": "IPY_MODEL_23e91246f4e04ad0bdd125bcdd098c3f"
     }
    },
    "0a8d1371d7744db18edf19fa41f85555": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "0bb5a03c228449f497f57d9fc02f4bbc": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "FloatProgressModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "FloatProgressModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "ProgressView",
      "bar_style": "success",
      "description": "Downloading: 100%",
      "description_tooltip": null,
      "layout": "IPY_MODEL_177475e15cc14018b2352132d426b4e4",
      "max": 536063208,
      "min": 0,
      "orientation": "horizontal",
      "style": "IPY_MODEL_48f5e96933a249198d4a7cea366447e6",
      "value": 536063208
     }
    },
    "0c43c5aab6084565a6e6c607c686dcc0": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "1318c81749d64b1ab4f58c712b4d9bb1": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HBoxModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HBoxModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HBoxView",
      "box_style": "",
      "children": [
       "IPY_MODEL_c1a100c2dc334553a97f7fe99a66d5de",
       "IPY_MODEL_4c4675d5429745f7a45bf52b79d136d0"
      ],
      "layout": "IPY_MODEL_d0cad404720740338f3472016dc53e2b"
     }
    },
    "1412b01aadf24634a7bbb50ecfc1fac9": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "14453fe91077495fa7b767c2885f3d5a": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HTMLModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HTMLModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HTMLView",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_aed12c255e93476f900aa1e3cbdad25e",
      "placeholder": "​",
      "style": "IPY_MODEL_9f8c8e97c53c4238a65dc69672dbad81",
      "value": " 23/23 [00:30&lt;00:00,  1.34s/it]"
     }
    },
    "1555f6aac6074ba09dcf569cf9d612f8": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "DescriptionStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "DescriptionStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "description_width": ""
     }
    },
    "15b4a5793e464151844a3ac0c6cacad5": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "ProgressStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "ProgressStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "bar_color": null,
      "description_width": "initial"
     }
    },
    "173231b0a0d34ab29897e44b2ad461df": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HBoxModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HBoxModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HBoxView",
      "box_style": "",
      "children": [
       "IPY_MODEL_749b7e4362964aa09aea33315ba6f14b",
       "IPY_MODEL_f5b71821098948598e73e61b025db362"
      ],
      "layout": "IPY_MODEL_cb45c7c08787447bb99ee370fd8ba716"
     }
    },
    "177475e15cc14018b2352132d426b4e4": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "181bc27df5024994a34545b243d21b58": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "1aaecf0b3b68444889b82106c452ed6f": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HTMLModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HTMLModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HTMLView",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_9e38d7ade99d4423b028f44161d045b0",
      "placeholder": "​",
      "style": "IPY_MODEL_efa1895edc5e4e619e4eb588b9127cae",
      "value": " 168/168 [02:16&lt;00:00,  1.23it/s]"
     }
    },
    "1b4dbe64fa2b4bb39b2304110e8903d3": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "ProgressStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "ProgressStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "bar_color": null,
      "description_width": "initial"
     }
    },
    "1d6bd0fb338e49e1acbbf3e51d2f1d7b": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HBoxModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HBoxModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HBoxView",
      "box_style": "",
      "children": [
       "IPY_MODEL_cb49ecbea18648928c9a12827ef3b1f8",
       "IPY_MODEL_d96e164e55214e1ebd3c9a7b7842f76e"
      ],
      "layout": "IPY_MODEL_76f3b9de46da468b824e3f49d39b83f2"
     }
    },
    "1dadb5f7f4bb4e71b0864c472d7712e4": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "1f8550f8dc224d3fb2723b6df6bfc6f3": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "ProgressStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "ProgressStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "bar_color": null,
      "description_width": "initial"
     }
    },
    "21a3a032149a49528625e7d0441d9585": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "DescriptionStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "DescriptionStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "description_width": ""
     }
    },
    "22c888cfe57a43f396145fadc60c40e0": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "DescriptionStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "DescriptionStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "description_width": ""
     }
    },
    "22cd761781ac42388dae310574759cd7": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "23e91246f4e04ad0bdd125bcdd098c3f": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "2521fd1894dd4da2a0a638ce7aed888a": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "FloatProgressModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "FloatProgressModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "ProgressView",
      "bar_style": "success",
      "description": "100%",
      "description_tooltip": null,
      "layout": "IPY_MODEL_d835b46e4dba47599effe36cc0150ce9",
      "max": 23,
      "min": 0,
      "orientation": "horizontal",
      "style": "IPY_MODEL_d1b0f94f3e6f469684ff1010635af586",
      "value": 23
     }
    },
    "26d2be5c8a654680a86a7272cf3b023b": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HTMLModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HTMLModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HTMLView",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_bf49bac4c7494e39b5d742a6098446c4",
      "placeholder": "​",
      "style": "IPY_MODEL_c101ca918f094a3cb940e384f89c558e",
      "value": " 168/168 [04:16&lt;00:00,  1.52s/it]"
     }
    },
    "28696c687449462e9566ad837a105bfc": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HBoxModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HBoxModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HBoxView",
      "box_style": "",
      "children": [
       "IPY_MODEL_a32279f5d1204557ae0e51c239d3f4c8",
       "IPY_MODEL_ca4adeb16771401d8bbd25352e4fb126"
      ],
      "layout": "IPY_MODEL_91fd6282e75e4e71b388a1ff67d450d6"
     }
    },
    "290b1788f1ac40d9b70348b5d0d5a1b0": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "2b8a84b54afa476db34de197efe98b86": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "2bb40cb966f144ba87c4595d508ede35": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HBoxModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HBoxModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HBoxView",
      "box_style": "",
      "children": [
       "IPY_MODEL_e02611fc3370422eab8db7952872060f",
       "IPY_MODEL_4b53887915d340438b79eb0d501539b2"
      ],
      "layout": "IPY_MODEL_7d5cc91153244093b8571aca86edae46"
     }
    },
    "2c3f2b79a0ee4d9993c070b679692a47": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "2cb7dcc84d30404789680ea54e8265d9": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "2d594e25d5e84813a13228cfde6fa74c": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HBoxModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HBoxModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HBoxView",
      "box_style": "",
      "children": [
       "IPY_MODEL_d61b2a1d66d541c38f3c47fba3f3e00e",
       "IPY_MODEL_a2c7c9107f244cc5a0bb04ef18324e8f"
      ],
      "layout": "IPY_MODEL_e4e3afa6697b44c1b46ce4723d57b2e0"
     }
    },
    "2da60bb609d04930bcc574d975625c22": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "2dfb108a53f54071afe40a4f66b80833": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "DescriptionStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "DescriptionStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "description_width": ""
     }
    },
    "2f56058ab7e84240889cc2b2ca8899ee": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "2fe1e199b53240318444af597fde0cdb": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "DescriptionStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "DescriptionStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "description_width": ""
     }
    },
    "30201ac2d0174912b4c30bd374ca878f": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HTMLModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HTMLModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HTMLView",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_a8f61cd4acac4b8583314fef0faceab4",
      "placeholder": "​",
      "style": "IPY_MODEL_22c888cfe57a43f396145fadc60c40e0",
      "value": " 565M/565M [00:07&lt;00:00, 77.1MB/s]"
     }
    },
    "3184ca332a4c410ab0a8abf172d6a247": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HBoxModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HBoxModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HBoxView",
      "box_style": "",
      "children": [
       "IPY_MODEL_0bb5a03c228449f497f57d9fc02f4bbc",
       "IPY_MODEL_999357f02a2446e2815e71a3e6dc979a"
      ],
      "layout": "IPY_MODEL_3b5d20bf24834e0bb9b8856a152e3b1e"
     }
    },
    "3214e2dce9314362b7b71a88791f04b9": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HTMLModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HTMLModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HTMLView",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_742e8976c14f4ab2b50c54545e359f0a",
      "placeholder": "​",
      "style": "IPY_MODEL_5dde40b97bbf497981d9f4962ed48873",
      "value": " 481/481 [00:00&lt;00:00, 780B/s]"
     }
    },
    "328beb7281604e378b867f29be0221fe": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "33b74b9e2a094c0a8dc4fed58a8a0efa": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "34027535904b4b4d9c4bbd19c58306a3": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HTMLModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HTMLModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HTMLView",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_4be40858751347649165a1f1b054321c",
      "placeholder": "​",
      "style": "IPY_MODEL_f196815db4574391a3c27e0bf5e9da8b",
      "value": " 23/23 [02:36&lt;00:00,  6.78s/it]"
     }
    },
    "34c916e95c114420bcf488d97dcb1c1b": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "37342f5fe94448989775ccfd2ead99ae": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "ProgressStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "ProgressStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "bar_color": null,
      "description_width": "initial"
     }
    },
    "3757d8ddd2e04e11a44fdb802a4af458": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "ProgressStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "ProgressStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "bar_color": null,
      "description_width": "initial"
     }
    },
    "38937e83918147fa8e4f246439550ec8": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "3a4fa61568e542a581a987d6b03c544c": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "DescriptionStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "DescriptionStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "description_width": ""
     }
    },
    "3a789e90935647a3860665cc8618dd5c": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "FloatProgressModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "FloatProgressModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "ProgressView",
      "bar_style": "success",
      "description": "Downloading: 100%",
      "description_tooltip": null,
      "layout": "IPY_MODEL_078378bbad8a41f697ef44eacfb88037",
      "max": 565485600,
      "min": 0,
      "orientation": "horizontal",
      "style": "IPY_MODEL_7a8ae20bce954983a509c835d6b796f2",
      "value": 565485600
     }
    },
    "3a87b11b1846450588eeb287c14a9cd0": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "FloatProgressModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "FloatProgressModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "ProgressView",
      "bar_style": "success",
      "description": "Downloading: 100%",
      "description_tooltip": null,
      "layout": "IPY_MODEL_2da60bb609d04930bcc574d975625c22",
      "max": 433,
      "min": 0,
      "orientation": "horizontal",
      "style": "IPY_MODEL_e98d667af83c4abb9e2caa7b13648770",
      "value": 433
     }
    },
    "3b5d20bf24834e0bb9b8856a152e3b1e": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "3f4b0e5ab79a4ffc81adc3c6966d7308": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "FloatProgressModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "FloatProgressModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "ProgressView",
      "bar_style": "success",
      "description": "100%",
      "description_tooltip": null,
      "layout": "IPY_MODEL_7325a03b04684879a54094ca3b246f90",
      "max": 23,
      "min": 0,
      "orientation": "horizontal",
      "style": "IPY_MODEL_88794d3fd37c43ffab4a6e275ee6e5db",
      "value": 23
     }
    },
    "3f77953ca9df4f6aa35efabd78c538c5": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HBoxModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HBoxModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HBoxView",
      "box_style": "",
      "children": [
       "IPY_MODEL_969d400ea2484a99b54a1e14de00103c",
       "IPY_MODEL_94f91259526f4c539da1f88701a8d244"
      ],
      "layout": "IPY_MODEL_33b74b9e2a094c0a8dc4fed58a8a0efa"
     }
    },
    "400f42347a444c6486c7380b00dc54e1": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HTMLModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HTMLModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HTMLView",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_70586fb1124f418f99f76fda730d401d",
      "placeholder": "​",
      "style": "IPY_MODEL_dbd9635795b848f5b654109f4a34ec3a",
      "value": " 23/23 [00:43&lt;00:00,  1.90s/it]"
     }
    },
    "40d0cdf589f64c0b8cc73059b7be6bad": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "415069863abb454686fcd00a0971e555": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HBoxModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HBoxModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HBoxView",
      "box_style": "",
      "children": [
       "IPY_MODEL_3f4b0e5ab79a4ffc81adc3c6966d7308",
       "IPY_MODEL_67aafe24778f433486f1bf9345f81c6b"
      ],
      "layout": "IPY_MODEL_290b1788f1ac40d9b70348b5d0d5a1b0"
     }
    },
    "41d6d529dc604841907652d578b392bb": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "430481b1567548bf995b1a21b8f1296b": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "DescriptionStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "DescriptionStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "description_width": ""
     }
    },
    "4353b94a332043f088936ab5c19fa9a5": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "44c33f30b8264061a402d84632661071": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "DescriptionStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "DescriptionStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "description_width": ""
     }
    },
    "454e610c4cb745eba9607bc3c95574d1": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "DescriptionStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "DescriptionStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "description_width": ""
     }
    },
    "48a9d939ab494c18b2b744c1f8d668fb": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "DescriptionStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "DescriptionStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "description_width": ""
     }
    },
    "48f5e96933a249198d4a7cea366447e6": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "ProgressStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "ProgressStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "bar_color": null,
      "description_width": "initial"
     }
    },
    "4ade78eb35de46be8e9d4c5f0dde3c98": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "ProgressStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "ProgressStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "bar_color": null,
      "description_width": "initial"
     }
    },
    "4b28705d9d7c4369b98013ae1363dea3": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HBoxModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HBoxModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HBoxView",
      "box_style": "",
      "children": [
       "IPY_MODEL_81abdaa83ea747928118b9b8f47d0693",
       "IPY_MODEL_26d2be5c8a654680a86a7272cf3b023b"
      ],
      "layout": "IPY_MODEL_d732539b2ed94c09b2e38ba0727c222d"
     }
    },
    "4b53887915d340438b79eb0d501539b2": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HTMLModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HTMLModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HTMLView",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_5f4768eb854641aea6f3f26714b285f2",
      "placeholder": "​",
      "style": "IPY_MODEL_78a4ab65d80f4744addc3b54a5f8bdb7",
      "value": " 657M/657M [00:08&lt;00:00, 74.5MB/s]"
     }
    },
    "4be40858751347649165a1f1b054321c": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "4c4675d5429745f7a45bf52b79d136d0": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HTMLModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HTMLModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HTMLView",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_07236e145d2844fd9b129dd3ee7e1718",
      "placeholder": "​",
      "style": "IPY_MODEL_9a11576090ab4ff592e2a10c2ee4b903",
      "value": " 168/168 [06:14&lt;00:00,  2.23s/it]"
     }
    },
    "4df465b390af4c0a84fc7841d6e35c5d": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "FloatProgressModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "FloatProgressModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "ProgressView",
      "bar_style": "success",
      "description": "Downloading: 100%",
      "description_tooltip": null,
      "layout": "IPY_MODEL_defe087e9c494f10b2c00c39bfd449d6",
      "max": 456318,
      "min": 0,
      "orientation": "horizontal",
      "style": "IPY_MODEL_4ade78eb35de46be8e9d4c5f0dde3c98",
      "value": 456318
     }
    },
    "52723b099ee8460091e9ff5db914d0e3": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "55b4169d866b426aa34752d3707126d0": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "FloatProgressModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "FloatProgressModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "ProgressView",
      "bar_style": "success",
      "description": "Downloading: 100%",
      "description_tooltip": null,
      "layout": "IPY_MODEL_711e4a4223d744feadd33967ad2a77ac",
      "max": 481,
      "min": 0,
      "orientation": "horizontal",
      "style": "IPY_MODEL_f58a2d64686a4750b1a3190067b26f6b",
      "value": 481
     }
    },
    "5892924a220043ada70ecc5435672697": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HBoxModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HBoxModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HBoxView",
      "box_style": "",
      "children": [
       "IPY_MODEL_d469efbe30b1486a9c61c67d97167e49",
       "IPY_MODEL_87ac7ee2f67d447c9c2549fcb3b5ed43"
      ],
      "layout": "IPY_MODEL_34c916e95c114420bcf488d97dcb1c1b"
     }
    },
    "58a0ed8938df49608ed05e277596eaad": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "5ad382ea213f4eee8d8d35811975ffdc": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "5dde40b97bbf497981d9f4962ed48873": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "DescriptionStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "DescriptionStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "description_width": ""
     }
    },
    "5e3f496954a94d2c97ca68a2358e4843": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "5e4a1fb1b41240428062b9106cb8f6d4": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "5e886f6e73de42b5ab762216070e9993": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "FloatProgressModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "FloatProgressModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "ProgressView",
      "bar_style": "success",
      "description": "100%",
      "description_tooltip": null,
      "layout": "IPY_MODEL_715c3419ef9e4197bfc89fff7aee0e09",
      "max": 12,
      "min": 0,
      "orientation": "horizontal",
      "style": "IPY_MODEL_86d084ebdaaa4d94a0e23a6752448647",
      "value": 12
     }
    },
    "5f4768eb854641aea6f3f26714b285f2": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "62d36481dba74df180179d104988c71b": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HBoxModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HBoxModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HBoxView",
      "box_style": "",
      "children": [
       "IPY_MODEL_64501f87ef5e475497291fae69e5d0bc",
       "IPY_MODEL_34027535904b4b4d9c4bbd19c58306a3"
      ],
      "layout": "IPY_MODEL_cfa56e32bb79478393524302d4a931bc"
     }
    },
    "6307ebc36945404fb153514c702e0fa6": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "63ed3558c4a14df085e09d74deb889b8": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HTMLModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HTMLModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HTMLView",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_9c52a6ea14b04a22a349847aa33f08df",
      "placeholder": "​",
      "style": "IPY_MODEL_48a9d939ab494c18b2b744c1f8d668fb",
      "value": " 23/23 [01:06&lt;00:00,  2.91s/it]"
     }
    },
    "64501f87ef5e475497291fae69e5d0bc": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "FloatProgressModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "FloatProgressModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "ProgressView",
      "bar_style": "success",
      "description": "100%",
      "description_tooltip": null,
      "layout": "IPY_MODEL_52723b099ee8460091e9ff5db914d0e3",
      "max": 23,
      "min": 0,
      "orientation": "horizontal",
      "style": "IPY_MODEL_37342f5fe94448989775ccfd2ead99ae",
      "value": 23
     }
    },
    "655a1b4b626f4bdd93a993f7147fd858": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "67aafe24778f433486f1bf9345f81c6b": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HTMLModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HTMLModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HTMLView",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_dc7bc3a76c7e40d990eccdddd2927a7a",
      "placeholder": "​",
      "style": "IPY_MODEL_033a5e21fde447b592f6a655270e9704",
      "value": " 23/23 [00:16&lt;00:00,  1.36it/s]"
     }
    },
    "69bdefc56b30498f943a24e96e99175d": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "6e5bf6c7f6aa458ab32b47159153e4fa": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HBoxModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HBoxModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HBoxView",
      "box_style": "",
      "children": [
       "IPY_MODEL_5e886f6e73de42b5ab762216070e9993",
       "IPY_MODEL_9ff4b21640d548a2b4e54af0793a0a92"
      ],
      "layout": "IPY_MODEL_c0769515e0da4f53b88a3e472a7045e5"
     }
    },
    "70586fb1124f418f99f76fda730d401d": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "711e4a4223d744feadd33967ad2a77ac": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "715c3419ef9e4197bfc89fff7aee0e09": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "7325a03b04684879a54094ca3b246f90": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "742e8976c14f4ab2b50c54545e359f0a": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "749b7e4362964aa09aea33315ba6f14b": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "FloatProgressModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "FloatProgressModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "ProgressView",
      "bar_style": "success",
      "description": "100%",
      "description_tooltip": null,
      "layout": "IPY_MODEL_03371146acc44a65981ee3fcbc91377e",
      "max": 168,
      "min": 0,
      "orientation": "horizontal",
      "style": "IPY_MODEL_baa37fa43b5a4f2987f09277249659df",
      "value": 168
     }
    },
    "75212dbf2fbb4e04b7a728b438f339ce": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "DescriptionStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "DescriptionStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "description_width": ""
     }
    },
    "76f3b9de46da468b824e3f49d39b83f2": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "77b09c5619064d58a2796023448389cd": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "FloatProgressModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "FloatProgressModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "ProgressView",
      "bar_style": "success",
      "description": "Downloading: 100%",
      "description_tooltip": null,
      "layout": "IPY_MODEL_be720b8098fb4accaca6710416a15451",
      "max": 798011,
      "min": 0,
      "orientation": "horizontal",
      "style": "IPY_MODEL_88ee5b27999c40f399bb8ca4a756c069",
      "value": 798011
     }
    },
    "78a4ab65d80f4744addc3b54a5f8bdb7": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "DescriptionStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "DescriptionStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "description_width": ""
     }
    },
    "7a8ae20bce954983a509c835d6b796f2": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "ProgressStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "ProgressStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "bar_color": null,
      "description_width": "initial"
     }
    },
    "7a93bc995be5417998fe1a05a792cd9c": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "ProgressStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "ProgressStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "bar_color": null,
      "description_width": "initial"
     }
    },
    "7d5cc91153244093b8571aca86edae46": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "7f812edd5c5e4451b043e4e85f9da4c1": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "ProgressStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "ProgressStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "bar_color": null,
      "description_width": "initial"
     }
    },
    "81abdaa83ea747928118b9b8f47d0693": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "FloatProgressModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "FloatProgressModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "ProgressView",
      "bar_style": "success",
      "description": "100%",
      "description_tooltip": null,
      "layout": "IPY_MODEL_58a0ed8938df49608ed05e277596eaad",
      "max": 168,
      "min": 0,
      "orientation": "horizontal",
      "style": "IPY_MODEL_7a93bc995be5417998fe1a05a792cd9c",
      "value": 168
     }
    },
    "82519ce7196643a2b49840e00b183358": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "8551901507524337b5b6828efbddd5e9": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "ProgressStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "ProgressStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "bar_color": null,
      "description_width": "initial"
     }
    },
    "86074fd920aa4ebab4a9c338192b4166": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "86d084ebdaaa4d94a0e23a6752448647": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "ProgressStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "ProgressStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "bar_color": null,
      "description_width": "initial"
     }
    },
    "87419ed4f82e4f9589c5563cd2b8f7b9": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "8788d63631334938aac76a16a84e5f62": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HBoxModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HBoxModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HBoxView",
      "box_style": "",
      "children": [
       "IPY_MODEL_2521fd1894dd4da2a0a638ce7aed888a",
       "IPY_MODEL_400f42347a444c6486c7380b00dc54e1"
      ],
      "layout": "IPY_MODEL_5e4a1fb1b41240428062b9106cb8f6d4"
     }
    },
    "87ac7ee2f67d447c9c2549fcb3b5ed43": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HTMLModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HTMLModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HTMLView",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_2f56058ab7e84240889cc2b2ca8899ee",
      "placeholder": "​",
      "style": "IPY_MODEL_2fe1e199b53240318444af597fde0cdb",
      "value": " 899k/899k [00:01&lt;00:00, 730kB/s]"
     }
    },
    "88794d3fd37c43ffab4a6e275ee6e5db": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "ProgressStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "ProgressStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "bar_color": null,
      "description_width": "initial"
     }
    },
    "88938176fe364f9ab389f5764dd791c1": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "88ee5b27999c40f399bb8ca4a756c069": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "ProgressStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "ProgressStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "bar_color": null,
      "description_width": "initial"
     }
    },
    "89d12b0a31c04a76bdcaf88fee1fd658": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "8bee015c4f52479b8d7fba6898b643c0": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "DescriptionStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "DescriptionStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "description_width": ""
     }
    },
    "8dc39021daaa447393d7bdd41736088a": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "91fd6282e75e4e71b388a1ff67d450d6": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "9474a20ee953478890830ce6d1d7f797": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "FloatProgressModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "FloatProgressModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "ProgressView",
      "bar_style": "success",
      "description": "100%",
      "description_tooltip": null,
      "layout": "IPY_MODEL_2cb7dcc84d30404789680ea54e8265d9",
      "max": 23,
      "min": 0,
      "orientation": "horizontal",
      "style": "IPY_MODEL_b0d45f5420774ba59f0fea05ea9788f3",
      "value": 23
     }
    },
    "94bd21b6c89d47488ef43dd045ff282a": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HBoxModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HBoxModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HBoxView",
      "box_style": "",
      "children": [
       "IPY_MODEL_55b4169d866b426aa34752d3707126d0",
       "IPY_MODEL_3214e2dce9314362b7b71a88791f04b9"
      ],
      "layout": "IPY_MODEL_4353b94a332043f088936ab5c19fa9a5"
     }
    },
    "94f91259526f4c539da1f88701a8d244": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HTMLModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HTMLModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HTMLView",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_41d6d529dc604841907652d578b392bb",
      "placeholder": "​",
      "style": "IPY_MODEL_430481b1567548bf995b1a21b8f1296b",
      "value": " 760/760 [00:01&lt;00:00, 512B/s]"
     }
    },
    "969d400ea2484a99b54a1e14de00103c": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "FloatProgressModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "FloatProgressModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "ProgressView",
      "bar_style": "success",
      "description": "Downloading: 100%",
      "description_tooltip": null,
      "layout": "IPY_MODEL_2c3f2b79a0ee4d9993c070b679692a47",
      "max": 760,
      "min": 0,
      "orientation": "horizontal",
      "style": "IPY_MODEL_1f8550f8dc224d3fb2723b6df6bfc6f3",
      "value": 760
     }
    },
    "96a674ad9acc45f58713274d3ffe9cef": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "999357f02a2446e2815e71a3e6dc979a": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HTMLModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HTMLModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HTMLView",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_070ba289643e4bdb8873cc225aad2326",
      "placeholder": "​",
      "style": "IPY_MODEL_ccca2cd356de49f1954702b20136af3a",
      "value": " 536M/536M [00:25&lt;00:00, 21.3MB/s]"
     }
    },
    "9a11576090ab4ff592e2a10c2ee4b903": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "DescriptionStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "DescriptionStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "description_width": ""
     }
    },
    "9a3a8a8e335c4c32983f541dd97fa462": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HBoxModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HBoxModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HBoxView",
      "box_style": "",
      "children": [
       "IPY_MODEL_d6ab42d1436e49e68fc84273f8942ec7",
       "IPY_MODEL_ac743d8ca32b4b1289df3969c99b4861"
      ],
      "layout": "IPY_MODEL_cd8e3aae7e98400a8244c1cd8e67a6f5"
     }
    },
    "9b5b1fc1cb85429c86655f717a9b1bb9": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "9c52a6ea14b04a22a349847aa33f08df": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "9e38d7ade99d4423b028f44161d045b0": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "9e6ae79f17194fdfb08d517fa70110dd": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "ProgressStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "ProgressStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "bar_color": null,
      "description_width": "initial"
     }
    },
    "9f80ed5cf07e4c07a1a79e1e1af476ac": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HBoxModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HBoxModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HBoxView",
      "box_style": "",
      "children": [
       "IPY_MODEL_9474a20ee953478890830ce6d1d7f797",
       "IPY_MODEL_fa6af85380124e24b5a4cf8421f168a3"
      ],
      "layout": "IPY_MODEL_69bdefc56b30498f943a24e96e99175d"
     }
    },
    "9f8c8e97c53c4238a65dc69672dbad81": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "DescriptionStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "DescriptionStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "description_width": ""
     }
    },
    "9ff4b21640d548a2b4e54af0793a0a92": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HTMLModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HTMLModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HTMLView",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_1412b01aadf24634a7bbb50ecfc1fac9",
      "placeholder": "​",
      "style": "IPY_MODEL_75212dbf2fbb4e04b7a728b438f339ce",
      "value": " 12/12 [00:57&lt;00:00,  4.76s/it]"
     }
    },
    "a2c7c9107f244cc5a0bb04ef18324e8f": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HTMLModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HTMLModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HTMLView",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_22cd761781ac42388dae310574759cd7",
      "placeholder": "​",
      "style": "IPY_MODEL_065a222b4ac14adb8a332e8b6f2cf2a8",
      "value": " 168/168 [08:25&lt;00:00,  3.01s/it]"
     }
    },
    "a32279f5d1204557ae0e51c239d3f4c8": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "FloatProgressModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "FloatProgressModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "ProgressView",
      "bar_style": "success",
      "description": "Downloading: 100%",
      "description_tooltip": null,
      "layout": "IPY_MODEL_d995f5dcd7d14093bacca3750e50e818",
      "max": 231508,
      "min": 0,
      "orientation": "horizontal",
      "style": "IPY_MODEL_3757d8ddd2e04e11a44fdb802a4af458",
      "value": 231508
     }
    },
    "a7d35e5de40f4b979c35f9e9d9b71b57": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "ProgressStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "ProgressStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "bar_color": null,
      "description_width": "initial"
     }
    },
    "a84e42c2cf6f499cb78824fa2da60dd5": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "FloatProgressModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "FloatProgressModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "ProgressView",
      "bar_style": "success",
      "description": "100%",
      "description_tooltip": null,
      "layout": "IPY_MODEL_328beb7281604e378b867f29be0221fe",
      "max": 168,
      "min": 0,
      "orientation": "horizontal",
      "style": "IPY_MODEL_f88504163dd24c64843f76e5952b070a",
      "value": 168
     }
    },
    "a852197af681419d85139b39a428c5a2": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HBoxModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HBoxModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HBoxView",
      "box_style": "",
      "children": [
       "IPY_MODEL_01c5462b34a741deb508c4a6aa9c1308",
       "IPY_MODEL_14453fe91077495fa7b767c2885f3d5a"
      ],
      "layout": "IPY_MODEL_82519ce7196643a2b49840e00b183358"
     }
    },
    "a8f61cd4acac4b8583314fef0faceab4": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "ac743d8ca32b4b1289df3969c99b4861": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HTMLModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HTMLModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HTMLView",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_40d0cdf589f64c0b8cc73059b7be6bad",
      "placeholder": "​",
      "style": "IPY_MODEL_b13d72305b1c47789c2c2a05228d2ea1",
      "value": " 168/168 [00:50&lt;00:00,  3.31it/s]"
     }
    },
    "aed12c255e93476f900aa1e3cbdad25e": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "af2fd6f1a2f74a11b86ddaafa3ca7419": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HBoxModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HBoxModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HBoxView",
      "box_style": "",
      "children": [
       "IPY_MODEL_a84e42c2cf6f499cb78824fa2da60dd5",
       "IPY_MODEL_1aaecf0b3b68444889b82106c452ed6f"
      ],
      "layout": "IPY_MODEL_dfb0d68084a04b549f1c06abe1edd6e7"
     }
    },
    "b0d45f5420774ba59f0fea05ea9788f3": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "ProgressStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "ProgressStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "bar_color": null,
      "description_width": "initial"
     }
    },
    "b13d72305b1c47789c2c2a05228d2ea1": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "DescriptionStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "DescriptionStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "description_width": ""
     }
    },
    "b8caaa40985b4dcbb41fe671817c389a": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HBoxModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HBoxModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HBoxView",
      "box_style": "",
      "children": [
       "IPY_MODEL_d4cf8d837172408ab631f98779eae95c",
       "IPY_MODEL_63ed3558c4a14df085e09d74deb889b8"
      ],
      "layout": "IPY_MODEL_9b5b1fc1cb85429c86655f717a9b1bb9"
     }
    },
    "b941cbd32450435fb453f1c88ade22d6": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HBoxModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HBoxModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HBoxView",
      "box_style": "",
      "children": [
       "IPY_MODEL_3a87b11b1846450588eeb287c14a9cd0",
       "IPY_MODEL_d7664b5f00614731b660cb1f093490c3"
      ],
      "layout": "IPY_MODEL_8dc39021daaa447393d7bdd41736088a"
     }
    },
    "baa37fa43b5a4f2987f09277249659df": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "ProgressStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "ProgressStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "bar_color": null,
      "description_width": "initial"
     }
    },
    "be3d89e9334c44d1a6acded33e08cee3": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HBoxModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HBoxModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HBoxView",
      "box_style": "",
      "children": [
       "IPY_MODEL_77b09c5619064d58a2796023448389cd",
       "IPY_MODEL_be836e272611471c9dae7b5770f48793"
      ],
      "layout": "IPY_MODEL_e635e2918eee490fad0cf98561a3397e"
     }
    },
    "be720b8098fb4accaca6710416a15451": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "be836e272611471c9dae7b5770f48793": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HTMLModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HTMLModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HTMLView",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_0c43c5aab6084565a6e6c607c686dcc0",
      "placeholder": "​",
      "style": "IPY_MODEL_21a3a032149a49528625e7d0441d9585",
      "value": " 798k/798k [00:00&lt;00:00, 883kB/s]"
     }
    },
    "bf49bac4c7494e39b5d742a6098446c4": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "c0769515e0da4f53b88a3e472a7045e5": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "c0c6772a00d942e08aa6ae47de8981ce": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "DescriptionStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "DescriptionStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "description_width": ""
     }
    },
    "c101ca918f094a3cb940e384f89c558e": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "DescriptionStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "DescriptionStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "description_width": ""
     }
    },
    "c1a100c2dc334553a97f7fe99a66d5de": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "FloatProgressModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "FloatProgressModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "ProgressView",
      "bar_style": "success",
      "description": "100%",
      "description_tooltip": null,
      "layout": "IPY_MODEL_d9a0e85bd1274f988f7950abd23882b2",
      "max": 168,
      "min": 0,
      "orientation": "horizontal",
      "style": "IPY_MODEL_9e6ae79f17194fdfb08d517fa70110dd",
      "value": 168
     }
    },
    "c30dcfac72a140d5b9317e382d9c7edc": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "c347122704754d009fd6aee346165275": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "ProgressStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "ProgressStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "bar_color": null,
      "description_width": "initial"
     }
    },
    "ca4adeb16771401d8bbd25352e4fb126": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HTMLModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HTMLModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HTMLView",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_d9f6b0bf75eb49588614011e4c453a14",
      "placeholder": "​",
      "style": "IPY_MODEL_2dfb108a53f54071afe40a4f66b80833",
      "value": " 232k/232k [00:00&lt;00:00, 786kB/s]"
     }
    },
    "cb45c7c08787447bb99ee370fd8ba716": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "cb49ecbea18648928c9a12827ef3b1f8": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "FloatProgressModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "FloatProgressModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "ProgressView",
      "bar_style": "success",
      "description": "100%",
      "description_tooltip": null,
      "layout": "IPY_MODEL_0375dbb6cfd14ac59a93b531a2a974f9",
      "max": 168,
      "min": 0,
      "orientation": "horizontal",
      "style": "IPY_MODEL_a7d35e5de40f4b979c35f9e9d9b71b57",
      "value": 168
     }
    },
    "ccca2cd356de49f1954702b20136af3a": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "DescriptionStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "DescriptionStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "description_width": ""
     }
    },
    "cd8e3aae7e98400a8244c1cd8e67a6f5": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "ce1cafb40f5845f1a622427efb9c91a6": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HBoxModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HBoxModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HBoxView",
      "box_style": "",
      "children": [
       "IPY_MODEL_d70dfff45a584559919f412e84d1e23d",
       "IPY_MODEL_ff85249aaf8e428bbf3a51cd2d5dd87e"
      ],
      "layout": "IPY_MODEL_89d12b0a31c04a76bdcaf88fee1fd658"
     }
    },
    "cfa56e32bb79478393524302d4a931bc": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "d0cad404720740338f3472016dc53e2b": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "d1b0f94f3e6f469684ff1010635af586": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "ProgressStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "ProgressStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "bar_color": null,
      "description_width": "initial"
     }
    },
    "d469efbe30b1486a9c61c67d97167e49": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "FloatProgressModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "FloatProgressModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "ProgressView",
      "bar_style": "success",
      "description": "Downloading: 100%",
      "description_tooltip": null,
      "layout": "IPY_MODEL_87419ed4f82e4f9589c5563cd2b8f7b9",
      "max": 898823,
      "min": 0,
      "orientation": "horizontal",
      "style": "IPY_MODEL_055725b181bb4d25b7a083b6cd7b373c",
      "value": 898823
     }
    },
    "d4cf8d837172408ab631f98779eae95c": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "FloatProgressModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "FloatProgressModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "ProgressView",
      "bar_style": "success",
      "description": "100%",
      "description_tooltip": null,
      "layout": "IPY_MODEL_5e3f496954a94d2c97ca68a2358e4843",
      "max": 23,
      "min": 0,
      "orientation": "horizontal",
      "style": "IPY_MODEL_1b4dbe64fa2b4bb39b2304110e8903d3",
      "value": 23
     }
    },
    "d61b2a1d66d541c38f3c47fba3f3e00e": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "FloatProgressModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "FloatProgressModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "ProgressView",
      "bar_style": "success",
      "description": "100%",
      "description_tooltip": null,
      "layout": "IPY_MODEL_181bc27df5024994a34545b243d21b58",
      "max": 168,
      "min": 0,
      "orientation": "horizontal",
      "style": "IPY_MODEL_c347122704754d009fd6aee346165275",
      "value": 168
     }
    },
    "d6ab42d1436e49e68fc84273f8942ec7": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "FloatProgressModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "FloatProgressModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "ProgressView",
      "bar_style": "success",
      "description": "100%",
      "description_tooltip": null,
      "layout": "IPY_MODEL_655a1b4b626f4bdd93a993f7147fd858",
      "max": 168,
      "min": 0,
      "orientation": "horizontal",
      "style": "IPY_MODEL_fd7687612b704c0f82548318d32146ef",
      "value": 168
     }
    },
    "d70dfff45a584559919f412e84d1e23d": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "FloatProgressModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "FloatProgressModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "ProgressView",
      "bar_style": "success",
      "description": "100%",
      "description_tooltip": null,
      "layout": "IPY_MODEL_f72eef90b33a42edb2a61185e6150881",
      "max": 23,
      "min": 0,
      "orientation": "horizontal",
      "style": "IPY_MODEL_8551901507524337b5b6828efbddd5e9",
      "value": 23
     }
    },
    "d732539b2ed94c09b2e38ba0727c222d": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "d7664b5f00614731b660cb1f093490c3": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HTMLModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HTMLModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HTMLView",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_88938176fe364f9ab389f5764dd791c1",
      "placeholder": "​",
      "style": "IPY_MODEL_1555f6aac6074ba09dcf569cf9d612f8",
      "value": " 433/433 [00:00&lt;00:00, 477B/s]"
     }
    },
    "d835b46e4dba47599effe36cc0150ce9": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "d96e164e55214e1ebd3c9a7b7842f76e": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HTMLModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HTMLModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HTMLView",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_38937e83918147fa8e4f246439550ec8",
      "placeholder": "​",
      "style": "IPY_MODEL_44c33f30b8264061a402d84632661071",
      "value": " 168/168 [15:22&lt;00:00,  5.49s/it]"
     }
    },
    "d995f5dcd7d14093bacca3750e50e818": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "d9a0e85bd1274f988f7950abd23882b2": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "d9f6b0bf75eb49588614011e4c453a14": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "dbd9635795b848f5b654109f4a34ec3a": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "DescriptionStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "DescriptionStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "description_width": ""
     }
    },
    "dc2f041f43144db6a2c90ed08dca051b": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "FloatProgressModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "FloatProgressModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "ProgressView",
      "bar_style": "success",
      "description": "100%",
      "description_tooltip": null,
      "layout": "IPY_MODEL_2b8a84b54afa476db34de197efe98b86",
      "max": 168,
      "min": 0,
      "orientation": "horizontal",
      "style": "IPY_MODEL_ffdf5fb1173b4c18ae7327756ec28887",
      "value": 168
     }
    },
    "dc7bc3a76c7e40d990eccdddd2927a7a": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "defe087e9c494f10b2c00c39bfd449d6": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "dfb0d68084a04b549f1c06abe1edd6e7": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "e02611fc3370422eab8db7952872060f": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "FloatProgressModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "FloatProgressModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "ProgressView",
      "bar_style": "success",
      "description": "Downloading: 100%",
      "description_tooltip": null,
      "layout": "IPY_MODEL_1dadb5f7f4bb4e71b0864c472d7712e4",
      "max": 657434796,
      "min": 0,
      "orientation": "horizontal",
      "style": "IPY_MODEL_7f812edd5c5e4451b043e4e85f9da4c1",
      "value": 657434796
     }
    },
    "e3e8f8a326df488e97d4c7d5fd398890": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HTMLModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HTMLModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HTMLView",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_86074fd920aa4ebab4a9c338192b4166",
      "placeholder": "​",
      "style": "IPY_MODEL_8bee015c4f52479b8d7fba6898b643c0",
      "value": " 168/168 [02:18&lt;00:00,  1.22it/s]"
     }
    },
    "e4e3afa6697b44c1b46ce4723d57b2e0": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "e635e2918eee490fad0cf98561a3397e": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "e98d667af83c4abb9e2caa7b13648770": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "ProgressStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "ProgressStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "bar_color": null,
      "description_width": "initial"
     }
    },
    "efa1895edc5e4e619e4eb588b9127cae": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "DescriptionStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "DescriptionStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "description_width": ""
     }
    },
    "f196815db4574391a3c27e0bf5e9da8b": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "DescriptionStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "DescriptionStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "description_width": ""
     }
    },
    "f58a2d64686a4750b1a3190067b26f6b": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "ProgressStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "ProgressStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "bar_color": null,
      "description_width": "initial"
     }
    },
    "f5b71821098948598e73e61b025db362": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HTMLModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HTMLModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HTMLView",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_0a8d1371d7744db18edf19fa41f85555",
      "placeholder": "​",
      "style": "IPY_MODEL_3a4fa61568e542a581a987d6b03c544c",
      "value": " 168/168 [06:10&lt;00:00,  2.20s/it]"
     }
    },
    "f72eef90b33a42edb2a61185e6150881": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "f87b02922496412183d3ff676fa0318c": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HTMLModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HTMLModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HTMLView",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_c30dcfac72a140d5b9317e382d9c7edc",
      "placeholder": "​",
      "style": "IPY_MODEL_04b1246c315743e885e28770d34fd83f",
      "value": " 456k/456k [00:00&lt;00:00, 1.42MB/s]"
     }
    },
    "f88504163dd24c64843f76e5952b070a": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "ProgressStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "ProgressStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "bar_color": null,
      "description_width": "initial"
     }
    },
    "fa6af85380124e24b5a4cf8421f168a3": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HTMLModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HTMLModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HTMLView",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_07e5725f24544b1580a3ae1699973779",
      "placeholder": "​",
      "style": "IPY_MODEL_c0c6772a00d942e08aa6ae47de8981ce",
      "value": " 23/23 [00:31&lt;00:00,  1.36s/it]"
     }
    },
    "fc3ee3f19d244a48bd1d145efd9d9264": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "fd7687612b704c0f82548318d32146ef": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "ProgressStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "ProgressStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "bar_color": null,
      "description_width": "initial"
     }
    },
    "fd8fc0fbf3bf4c749d2754026dec3024": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HBoxModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HBoxModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HBoxView",
      "box_style": "",
      "children": [
       "IPY_MODEL_dc2f041f43144db6a2c90ed08dca051b",
       "IPY_MODEL_e3e8f8a326df488e97d4c7d5fd398890"
      ],
      "layout": "IPY_MODEL_96a674ad9acc45f58713274d3ffe9cef"
     }
    },
    "fdec9f8d91464284a3ef937691258d43": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HBoxModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HBoxModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HBoxView",
      "box_style": "",
      "children": [
       "IPY_MODEL_3a789e90935647a3860665cc8618dd5c",
       "IPY_MODEL_30201ac2d0174912b4c30bd374ca878f"
      ],
      "layout": "IPY_MODEL_5ad382ea213f4eee8d8d35811975ffdc"
     }
    },
    "ff85249aaf8e428bbf3a51cd2d5dd87e": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HTMLModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HTMLModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HTMLView",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_6307ebc36945404fb153514c702e0fa6",
      "placeholder": "​",
      "style": "IPY_MODEL_454e610c4cb745eba9607bc3c95574d1",
      "value": " 23/23 [00:30&lt;00:00,  1.33s/it]"
     }
    },
    "ffdf5fb1173b4c18ae7327756ec28887": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "ProgressStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "ProgressStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "bar_color": null,
      "description_width": "initial"
     }
    }
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}