notebooks/Training-with-RoBERTa.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "papermill": {
     "duration": 0.010695,
     "end_time": "2020-08-17T00:54:10.187660",
     "exception": false,
     "start_time": "2020-08-17T00:54:10.176965",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# Text Classification with RoBERTa\n",
    "\n",
    "The RoBERTa model was proposed in RoBERTa: A Robustly Optimized BERT Pretraining Approach by Yinhan Liu et al. It is based on Google’s BERT model released in 2018: it modifies key hyperparameters, removing the next-sentence pretraining objective and training with much larger mini-batches and learning rates.\n",
    "\n",
    "This notebook is also available in Kaggle: https://www.kaggle.com/dimasmunoz/text-classification-with-roberta-and-tpus\n",
    "\n",
    "See:\n",
    "- RoBERTa's paper: https://arxiv.org/pdf/1907.11692.pdf\n",
    "- BERT's paper: https://arxiv.org/pdf/1810.04805.pdf"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-17T00:54:10.216279Z",
     "iopub.status.busy": "2020-08-17T00:54:10.215473Z",
     "iopub.status.idle": "2020-08-17T00:54:20.403926Z",
     "shell.execute_reply": "2020-08-17T00:54:20.403111Z"
    },
    "papermill": {
     "duration": 10.207617,
     "end_time": "2020-08-17T00:54:20.404114",
     "exception": false,
     "start_time": "2020-08-17T00:54:10.196497",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\u001b[34m\u001b[1mwandb\u001b[0m: \u001b[33mWARNING\u001b[0m W&B installed but not logged in.  Run `wandb login` or set the WANDB_API_KEY env variable.\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "import regex as re\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "import statistics\n",
    "import math\n",
    "import os\n",
    "\n",
    "from sklearn.model_selection import StratifiedKFold\n",
    "from sklearn.metrics import accuracy_score\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "import tensorflow as tf\n",
    "import tensorflow.keras.backend as K\n",
    "import tokenizers\n",
    "from transformers import RobertaTokenizer, TFRobertaModel\n",
    "\n",
    "from collections import Counter\n",
    "\n",
    "import warnings\n",
    "warnings.filterwarnings(\"ignore\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-17T00:54:20.534134Z",
     "iopub.status.busy": "2020-08-17T00:54:20.531975Z",
     "iopub.status.idle": "2020-08-17T00:54:24.802475Z",
     "shell.execute_reply": "2020-08-17T00:54:24.805600Z"
    },
    "papermill": {
     "duration": 4.392084,
     "end_time": "2020-08-17T00:54:24.805888",
     "exception": false,
     "start_time": "2020-08-17T00:54:20.413804",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Running on TPU  grpc://10.0.0.2:8470\n",
      "Number of replicas: 8\n"
     ]
    }
   ],
   "source": [
    "# Detect hardware, return appropriate distribution strategy (you can see that it is pretty easy to set up).\n",
    "try:\n",
    "    # TPU detection. No parameters necessary if TPU_NAME environment variable is set (always set in Kaggle)\n",
    "    tpu = tf.distribute.cluster_resolver.TPUClusterResolver()\n",
    "    tf.config.experimental_connect_to_cluster(tpu)\n",
    "    tf.tpu.experimental.initialize_tpu_system(tpu)\n",
    "    strategy = tf.distribute.experimental.TPUStrategy(tpu)\n",
    "    print('Running on TPU ', tpu.master())\n",
    "except ValueError:\n",
    "    # Default distribution strategy in Tensorflow. Works on CPU and single GPU.\n",
    "    strategy = tf.distribute.get_strategy()\n",
    "\n",
    "print('Number of replicas:', strategy.num_replicas_in_sync)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-17T00:54:24.837595Z",
     "iopub.status.busy": "2020-08-17T00:54:24.836634Z",
     "iopub.status.idle": "2020-08-17T00:54:24.840372Z",
     "shell.execute_reply": "2020-08-17T00:54:24.839749Z"
    },
    "papermill": {
     "duration": 0.021917,
     "end_time": "2020-08-17T00:54:24.840489",
     "exception": false,
     "start_time": "2020-08-17T00:54:24.818572",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "MODEL_NAME = 'roberta-base'\n",
    "MAX_LEN = 256\n",
    "ARTIFACTS_PATH = '../artifacts/'\n",
    "\n",
    "BATCH_SIZE = 8 * strategy.num_replicas_in_sync\n",
    "EPOCHS = 3\n",
    "\n",
    "if not os.path.exists(ARTIFACTS_PATH):\n",
    "    os.makedirs(ARTIFACTS_PATH)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "papermill": {
     "duration": 0.008929,
     "end_time": "2020-08-17T00:54:24.858434",
     "exception": false,
     "start_time": "2020-08-17T00:54:24.849505",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "## Prepare dataset\n",
    "\n",
    "Note that I have already cleaned the dataset, so the training should be better:\n",
    "- Remove punctuation symbols and double white spaces.\n",
    "- Lemmatization.\n",
    "- Remove stop words (see `spacy.lang.en.stop_words.STOP_WORDS`).\n",
    "- TD-IDF\n",
    "\n",
    "The reason of this preprocess is that I have used this dataset with other models, such as LSTM. However, I believe (I didn't test it) that RoBERTa could deal with it."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-17T00:54:24.888446Z",
     "iopub.status.busy": "2020-08-17T00:54:24.887424Z",
     "iopub.status.idle": "2020-08-17T00:54:25.038571Z",
     "shell.execute_reply": "2020-08-17T00:54:25.037856Z"
    },
    "papermill": {
     "duration": 0.170554,
     "end_time": "2020-08-17T00:54:25.038715",
     "exception": false,
     "start_time": "2020-08-17T00:54:24.868161",
     "status": "completed"
    },
    "tags": []
   },
   "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>text</th>\n",
       "      <th>category</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>musicians tackle red tape musicians groups tac...</td>\n",
       "      <td>entertainment</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>u2 desire number u2 won prestigious grammy awa...</td>\n",
       "      <td>entertainment</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>rocker doherty stage fight rock singer pete do...</td>\n",
       "      <td>entertainment</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>snicket tops box office chart film adaptation ...</td>\n",
       "      <td>entertainment</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>ocean raids box office ocean crime caper seque...</td>\n",
       "      <td>entertainment</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                text       category\n",
       "0  musicians tackle red tape musicians groups tac...  entertainment\n",
       "1  u2 desire number u2 won prestigious grammy awa...  entertainment\n",
       "2  rocker doherty stage fight rock singer pete do...  entertainment\n",
       "3  snicket tops box office chart film adaptation ...  entertainment\n",
       "4  ocean raids box office ocean crime caper seque...  entertainment"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = pd.read_csv('/kaggle/input/bbc-articles-cleaned/tfidf_dataset.csv')\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-17T00:54:25.071606Z",
     "iopub.status.busy": "2020-08-17T00:54:25.070557Z",
     "iopub.status.idle": "2020-08-17T00:54:25.074060Z",
     "shell.execute_reply": "2020-08-17T00:54:25.073309Z"
    },
    "papermill": {
     "duration": 0.02603,
     "end_time": "2020-08-17T00:54:25.074191",
     "exception": false,
     "start_time": "2020-08-17T00:54:25.048161",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "X_data = df[['text']].to_numpy().reshape(-1)\n",
    "y_data = df[['category']].to_numpy().reshape(-1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "papermill": {
     "duration": 0.009042,
     "end_time": "2020-08-17T00:54:25.092420",
     "exception": false,
     "start_time": "2020-08-17T00:54:25.083378",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# Dataset analysis\n",
    "\n",
    "As you can see in the following plot, the samples **are not balanced**. This could cause problems during the training but, since they are not highly unbalanced, I have left it as it is.\n",
    "\n",
    "In other cases, such as fraud detecting where the positive classes are very few compared to the negative ones, we must apply techniques to balance it. For example, we could undersample the biggest category."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-17T00:54:25.123293Z",
     "iopub.status.busy": "2020-08-17T00:54:25.122556Z",
     "iopub.status.idle": "2020-08-17T00:54:25.327630Z",
     "shell.execute_reply": "2020-08-17T00:54:25.328399Z"
    },
    "papermill": {
     "duration": 0.226971,
     "end_time": "2020-08-17T00:54:25.328644",
     "exception": false,
     "start_time": "2020-08-17T00:54:25.101673",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Counter({'sport': 511, 'business': 510, 'politics': 417, 'tech': 401, 'entertainment': 386})\n"
     ]
    }
   ],
   "source": [
    "categories = df[['category']].values.reshape(-1)\n",
    "\n",
    "counter_categories = Counter(categories)\n",
    "category_names = counter_categories.keys()\n",
    "category_values = counter_categories.values()\n",
    "\n",
    "y_pos = np.arange(len(category_names))\n",
    "\n",
    "plt.figure(1, figsize=(10, 5))\n",
    "plt.bar(y_pos, category_values, align='center', alpha=0.5)\n",
    "plt.xticks(y_pos, category_names)\n",
    "plt.ylabel('Number of texts')\n",
    "plt.title('Distribution of texts per category')\n",
    "plt.gca().yaxis.grid(True)\n",
    "plt.show()\n",
    "\n",
    "print(counter_categories)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-17T00:54:25.372178Z",
     "iopub.status.busy": "2020-08-17T00:54:25.371115Z",
     "iopub.status.idle": "2020-08-17T00:54:25.374365Z",
     "shell.execute_reply": "2020-08-17T00:54:25.373806Z"
    },
    "papermill": {
     "duration": 0.035052,
     "end_time": "2020-08-17T00:54:25.374494",
     "exception": false,
     "start_time": "2020-08-17T00:54:25.339442",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "def calculate_stats(df, split_char=' '):\n",
    "    categories = df['category'].unique()\n",
    "    \n",
    "    all_lengths = []\n",
    "    per_category = {\n",
    "        'lengths': {c:[] for c in categories},\n",
    "        'mean': {c:0 for c in categories},\n",
    "        'stdev': {c:0 for c in categories}\n",
    "    }\n",
    "\n",
    "    for index, row in df.iterrows():\n",
    "        text = row['text']\n",
    "        text = re.sub(r\"\\s+\", ' ', text) # Normalize\n",
    "        text = text.split(split_char)\n",
    "        l = len(text)\n",
    "        \n",
    "        category = row['category']\n",
    "        \n",
    "        all_lengths.append(l)\n",
    "        per_category['lengths'][category].append(l)\n",
    "    \n",
    "    for c in categories:\n",
    "        per_category['mean'][c] = statistics.mean(per_category['lengths'][c])\n",
    "        per_category['stdev'][c] = statistics.stdev(per_category['lengths'][c])\n",
    "    \n",
    "    global_stats = {\n",
    "        'mean': statistics.mean(all_lengths),\n",
    "        'stdev': statistics.stdev(all_lengths),\n",
    "        'lengths': all_lengths\n",
    "    }\n",
    "    \n",
    "    return {\n",
    "        'global': global_stats,\n",
    "        'per_category': pd.DataFrame(per_category)\n",
    "    }\n",
    "\n",
    "\n",
    "def display_lengths_histograms(df_stats, n_cols=3):\n",
    "    categories = df['category'].unique()\n",
    "    n_rows = math.ceil(len(categories) / n_cols)\n",
    "    \n",
    "    plt.figure(figsize=(15, 8))\n",
    "    plt.suptitle('Distribution of lengths')\n",
    "    \n",
    "    # Subplot of all lengths\n",
    "    plt.subplot(n_rows, n_cols, 1)\n",
    "    plt.title('All categories')\n",
    "    lengths = df_stats['global']['lengths']\n",
    "    plt.hist(lengths, color='r')\n",
    "\n",
    "    # Subplot of each category\n",
    "    index_subplot = 2\n",
    "    for c in categories:\n",
    "        plt.subplot(n_rows, n_cols, index_subplot)\n",
    "        plt.title('Category: %s' % c)\n",
    "        \n",
    "        lengths = df_stats['per_category']['lengths'][c]\n",
    "        plt.hist(lengths, color='b')\n",
    "\n",
    "        index_subplot += 1\n",
    "\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "papermill": {
     "duration": 0.009317,
     "end_time": "2020-08-17T00:54:25.393724",
     "exception": false,
     "start_time": "2020-08-17T00:54:25.384407",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "In these histograms and stats, we can see that almost all texts contain $500$ or less words. Also, we can see that the average length is very different depending on the category.\n",
    "\n",
    "Please, note that the tokenization process may split words into several parts, so lengths could increase (or decrease too). This is only an orientative result."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-17T00:54:25.427388Z",
     "iopub.status.busy": "2020-08-17T00:54:25.424927Z",
     "iopub.status.idle": "2020-08-17T00:54:26.224376Z",
     "shell.execute_reply": "2020-08-17T00:54:26.223588Z"
    },
    "papermill": {
     "duration": 0.821006,
     "end_time": "2020-08-17T00:54:26.224510",
     "exception": false,
     "start_time": "2020-08-17T00:54:25.403504",
     "status": "completed"
    },
    "tags": []
   },
   "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>lengths</th>\n",
       "      <th>mean</th>\n",
       "      <th>stdev</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>entertainment</th>\n",
       "      <td>[165, 375, 174, 108, 167, 189, 130, 230, 115, ...</td>\n",
       "      <td>168.585492</td>\n",
       "      <td>118.648402</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sport</th>\n",
       "      <td>[86, 86, 92, 72, 71, 68, 703, 257, 226, 71, 71...</td>\n",
       "      <td>159.681018</td>\n",
       "      <td>92.709783</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>business</th>\n",
       "      <td>[90, 82, 105, 169, 229, 321, 100, 85, 82, 127,...</td>\n",
       "      <td>172.876471</td>\n",
       "      <td>65.119283</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>politics</th>\n",
       "      <td>[115, 297, 289, 116, 208, 235, 231, 183, 113, ...</td>\n",
       "      <td>215.347722</td>\n",
       "      <td>128.812873</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>tech</th>\n",
       "      <td>[194, 117, 274, 135, 223, 170, 222, 165, 168, ...</td>\n",
       "      <td>241.835411</td>\n",
       "      <td>103.570666</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                         lengths        mean  \\\n",
       "entertainment  [165, 375, 174, 108, 167, 189, 130, 230, 115, ...  168.585492   \n",
       "sport          [86, 86, 92, 72, 71, 68, 703, 257, 226, 71, 71...  159.681018   \n",
       "business       [90, 82, 105, 169, 229, 321, 100, 85, 82, 127,...  172.876471   \n",
       "politics       [115, 297, 289, 116, 208, 235, 231, 183, 113, ...  215.347722   \n",
       "tech           [194, 117, 274, 135, 223, 170, 222, 165, 168, ...  241.835411   \n",
       "\n",
       "                    stdev  \n",
       "entertainment  118.648402  \n",
       "sport           92.709783  \n",
       "business        65.119283  \n",
       "politics       128.812873  \n",
       "tech           103.570666  "
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_stats = calculate_stats(df)\n",
    "df_stats['per_category']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-17T00:54:26.259851Z",
     "iopub.status.busy": "2020-08-17T00:54:26.259115Z",
     "iopub.status.idle": "2020-08-17T00:54:27.562332Z",
     "shell.execute_reply": "2020-08-17T00:54:27.562973Z"
    },
    "papermill": {
     "duration": 1.327443,
     "end_time": "2020-08-17T00:54:27.563154",
     "exception": false,
     "start_time": "2020-08-17T00:54:26.235711",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x576 with 6 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "display_lengths_histograms(df_stats)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-17T00:54:27.594084Z",
     "iopub.status.busy": "2020-08-17T00:54:27.592984Z",
     "iopub.status.idle": "2020-08-17T00:54:27.598171Z",
     "shell.execute_reply": "2020-08-17T00:54:27.597344Z"
    },
    "papermill": {
     "duration": 0.023118,
     "end_time": "2020-08-17T00:54:27.598331",
     "exception": false,
     "start_time": "2020-08-17T00:54:27.575213",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Texts in dataset: 2225\n",
      "Number of categories: 5\n",
      "Done!\n"
     ]
    }
   ],
   "source": [
    "n_texts = len(X_data)\n",
    "print('Texts in dataset: %d' % n_texts)\n",
    "\n",
    "categories = df['category'].unique()\n",
    "n_categories = len(categories)\n",
    "print('Number of categories: %d' % n_categories)\n",
    "\n",
    "print('Done!')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "papermill": {
     "duration": 0.010189,
     "end_time": "2020-08-17T00:54:27.619679",
     "exception": false,
     "start_time": "2020-08-17T00:54:27.609490",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# Tokenize & encode"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "papermill": {
     "duration": 0.0101,
     "end_time": "2020-08-17T00:54:27.640395",
     "exception": false,
     "start_time": "2020-08-17T00:54:27.630295",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "I am going to import the RoBERTa model from HuggingFace. Note that I must import not only the model but also the tokenizer (since I must use the same vocabulary as the model was trained with).\n",
    "\n",
    "We should take into account that RoBERTa's input accepts up-to 512 tokens, thus **we must truncate the tokenized texts**. In my case, I truncate to 256 tokens, but you can put a higher value in the variable `MAX_LEN`.\n",
    "\n",
    "See https://huggingface.co/roberta-base"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-17T00:54:27.673875Z",
     "iopub.status.busy": "2020-08-17T00:54:27.672948Z",
     "iopub.status.idle": "2020-08-17T00:54:27.675814Z",
     "shell.execute_reply": "2020-08-17T00:54:27.675097Z"
    },
    "papermill": {
     "duration": 0.025277,
     "end_time": "2020-08-17T00:54:27.675931",
     "exception": false,
     "start_time": "2020-08-17T00:54:27.650654",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "def roberta_encode(texts, tokenizer):\n",
    "    ct = len(texts)\n",
    "    input_ids = np.ones((ct, MAX_LEN), dtype='int32')\n",
    "    attention_mask = np.zeros((ct, MAX_LEN), dtype='int32')\n",
    "    token_type_ids = np.zeros((ct, MAX_LEN), dtype='int32') # Not used in text classification\n",
    "\n",
    "    for k, text in enumerate(texts):\n",
    "        # Tokenize\n",
    "        tok_text = tokenizer.tokenize(text)\n",
    "        \n",
    "        # Truncate and convert tokens to numerical IDs\n",
    "        enc_text = tokenizer.convert_tokens_to_ids(tok_text[:(MAX_LEN-2)])\n",
    "        \n",
    "        input_length = len(enc_text) + 2\n",
    "        input_length = input_length if input_length < MAX_LEN else MAX_LEN\n",
    "        \n",
    "        # Add tokens [CLS] and [SEP] at the beginning and the end\n",
    "        input_ids[k,:input_length] = np.asarray([0] + enc_text + [2], dtype='int32')\n",
    "        \n",
    "        # Set to 1s in the attention input\n",
    "        attention_mask[k,:input_length] = 1\n",
    "\n",
    "    return {\n",
    "        'input_word_ids': input_ids,\n",
    "        'input_mask': attention_mask,\n",
    "        'input_type_ids': token_type_ids\n",
    "    }"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-17T00:54:27.707846Z",
     "iopub.status.busy": "2020-08-17T00:54:27.707063Z",
     "iopub.status.idle": "2020-08-17T00:54:27.712023Z",
     "shell.execute_reply": "2020-08-17T00:54:27.711264Z"
    },
    "papermill": {
     "duration": 0.025508,
     "end_time": "2020-08-17T00:54:27.712149",
     "exception": false,
     "start_time": "2020-08-17T00:54:27.686641",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{0: 'entertainment', 1: 'sport', 2: 'business', 3: 'politics', 4: 'tech'}"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Transform categories into numbers\n",
    "category_to_id = {}\n",
    "category_to_name = {}\n",
    "\n",
    "for index, c in enumerate(y_data):\n",
    "    if c in category_to_id:\n",
    "        category_id = category_to_id[c]\n",
    "    else:\n",
    "        category_id = len(category_to_id)\n",
    "        category_to_id[c] = category_id\n",
    "        category_to_name[category_id] = c\n",
    "    \n",
    "    y_data[index] = category_id\n",
    "\n",
    "# Display dictionary\n",
    "category_to_name"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-17T00:54:27.740811Z",
     "iopub.status.busy": "2020-08-17T00:54:27.740049Z",
     "iopub.status.idle": "2020-08-17T00:54:27.743329Z",
     "shell.execute_reply": "2020-08-17T00:54:27.742740Z"
    },
    "papermill": {
     "duration": 0.020535,
     "end_time": "2020-08-17T00:54:27.743506",
     "exception": false,
     "start_time": "2020-08-17T00:54:27.722971",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "# Split into train and test datasets\n",
    "X_train, X_test, y_train, y_test = train_test_split(X_data, y_data, test_size=0.3, random_state=777) # random_state to reproduce results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-17T00:54:27.770323Z",
     "iopub.status.busy": "2020-08-17T00:54:27.769588Z",
     "iopub.status.idle": "2020-08-17T00:54:30.653159Z",
     "shell.execute_reply": "2020-08-17T00:54:30.652457Z"
    },
    "papermill": {
     "duration": 2.898736,
     "end_time": "2020-08-17T00:54:30.653305",
     "exception": false,
     "start_time": "2020-08-17T00:54:27.754569",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "8ba5b5d48551478aa619a303aca887c2",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, description='Downloading', max=898823.0, style=ProgressStyle(descripti…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "e2486b910c4142e3affeae7663e1a987",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, description='Downloading', max=456318.0, style=ProgressStyle(descripti…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "# Import tokenizer from HuggingFace\n",
    "tokenizer = RobertaTokenizer.from_pretrained(MODEL_NAME)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-17T00:54:30.682849Z",
     "iopub.status.busy": "2020-08-17T00:54:30.682057Z",
     "iopub.status.idle": "2020-08-17T00:54:36.211374Z",
     "shell.execute_reply": "2020-08-17T00:54:36.211977Z"
    },
    "papermill": {
     "duration": 5.54731,
     "end_time": "2020-08-17T00:54:36.212155",
     "exception": false,
     "start_time": "2020-08-17T00:54:30.664845",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "X_train = roberta_encode(X_train, tokenizer)\n",
    "X_test = roberta_encode(X_test, tokenizer)\n",
    "\n",
    "y_train = np.asarray(y_train, dtype='int32')\n",
    "y_test = np.asarray(y_test, dtype='int32')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "papermill": {
     "duration": 0.011666,
     "end_time": "2020-08-17T00:54:36.235726",
     "exception": false,
     "start_time": "2020-08-17T00:54:36.224060",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "## Create RoBERTa model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-17T00:54:36.276650Z",
     "iopub.status.busy": "2020-08-17T00:54:36.275686Z",
     "iopub.status.idle": "2020-08-17T00:54:36.279045Z",
     "shell.execute_reply": "2020-08-17T00:54:36.278290Z"
    },
    "papermill": {
     "duration": 0.031839,
     "end_time": "2020-08-17T00:54:36.279183",
     "exception": false,
     "start_time": "2020-08-17T00:54:36.247344",
     "status": "completed"
    },
    "scrolled": true,
    "tags": []
   },
   "outputs": [],
   "source": [
    "def build_model(n_categories):\n",
    "    with strategy.scope():\n",
    "        input_word_ids = tf.keras.Input(shape=(MAX_LEN,), dtype=tf.int32, name='input_word_ids')\n",
    "        input_mask = tf.keras.Input(shape=(MAX_LEN,), dtype=tf.int32, name='input_mask')\n",
    "        input_type_ids = tf.keras.Input(shape=(MAX_LEN,), dtype=tf.int32, name='input_type_ids')\n",
    "\n",
    "        # Import RoBERTa model from HuggingFace\n",
    "        roberta_model = TFRobertaModel.from_pretrained(MODEL_NAME)\n",
    "        x = roberta_model(input_word_ids, attention_mask=input_mask, token_type_ids=input_type_ids)\n",
    "\n",
    "        # Huggingface transformers have multiple outputs, embeddings are the first one,\n",
    "        # so let's slice out the first position\n",
    "        x = x[0]\n",
    "\n",
    "        x = tf.keras.layers.Dropout(0.1)(x)\n",
    "        x = tf.keras.layers.Flatten()(x)\n",
    "        x = tf.keras.layers.Dense(256, activation='relu')(x)\n",
    "        x = tf.keras.layers.Dense(n_categories, activation='softmax')(x)\n",
    "\n",
    "        model = tf.keras.Model(inputs=[input_word_ids, input_mask, input_type_ids], outputs=x)\n",
    "        model.compile(\n",
    "            optimizer=tf.keras.optimizers.Adam(lr=1e-5),\n",
    "            loss='sparse_categorical_crossentropy',\n",
    "            metrics=['accuracy'])\n",
    "\n",
    "        return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-17T00:54:36.310225Z",
     "iopub.status.busy": "2020-08-17T00:54:36.309483Z",
     "iopub.status.idle": "2020-08-17T00:55:30.295155Z",
     "shell.execute_reply": "2020-08-17T00:55:30.294135Z"
    },
    "papermill": {
     "duration": 54.004597,
     "end_time": "2020-08-17T00:55:30.295343",
     "exception": false,
     "start_time": "2020-08-17T00:54:36.290746",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "df64321df6a84519a33a6b371cbe8f05",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, description='Downloading', max=481.0, style=ProgressStyle(description_…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "2905d02e54134106b6569b15080816a4",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, description='Downloading', max=657434796.0, style=ProgressStyle(descri…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Model: \"model\"\n",
      "__________________________________________________________________________________________________\n",
      "Layer (type)                    Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      "input_word_ids (InputLayer)     [(None, 256)]        0                                            \n",
      "__________________________________________________________________________________________________\n",
      "input_mask (InputLayer)         [(None, 256)]        0                                            \n",
      "__________________________________________________________________________________________________\n",
      "input_type_ids (InputLayer)     [(None, 256)]        0                                            \n",
      "__________________________________________________________________________________________________\n",
      "tf_roberta_model (TFRobertaMode ((None, 256, 768), ( 124645632   input_word_ids[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "dropout_38 (Dropout)            (None, 256, 768)     0           tf_roberta_model[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "flatten (Flatten)               (None, 196608)       0           dropout_38[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "dense (Dense)                   (None, 256)          50331904    flatten[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "dense_1 (Dense)                 (None, 5)            1285        dense[0][0]                      \n",
      "==================================================================================================\n",
      "Total params: 174,978,821\n",
      "Trainable params: 174,978,821\n",
      "Non-trainable params: 0\n",
      "__________________________________________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "with strategy.scope():\n",
    "    model = build_model(n_categories)\n",
    "    model.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "papermill": {
     "duration": 0.012359,
     "end_time": "2020-08-17T00:55:30.320709",
     "exception": false,
     "start_time": "2020-08-17T00:55:30.308350",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "## Train model\n",
    "\n",
    "This is basic training of RoBERTa but, if your dataset is larger, you may use K-Folds in this section. In this notebook, I use K-Folds (use it as inspiration): https://www.kaggle.com/dimasmunoz/clean-english-data-roberta ;)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-17T00:55:30.365925Z",
     "iopub.status.busy": "2020-08-17T00:55:30.364306Z",
     "iopub.status.idle": "2020-08-17T00:57:29.469057Z",
     "shell.execute_reply": "2020-08-17T00:57:29.468347Z"
    },
    "papermill": {
     "duration": 119.136108,
     "end_time": "2020-08-17T00:57:29.469189",
     "exception": false,
     "start_time": "2020-08-17T00:55:30.333081",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training...\n",
      "Epoch 1/3\n",
      "25/25 [==============================] - 40s 2s/step - loss: 0.6481 - accuracy: 0.7720 - val_loss: 0.1106 - val_accuracy: 0.9671\n",
      "Epoch 2/3\n",
      "25/25 [==============================] - 4s 152ms/step - loss: 0.0792 - accuracy: 0.9743 - val_loss: 0.0749 - val_accuracy: 0.9716\n",
      "Epoch 3/3\n",
      "25/25 [==============================] - 4s 149ms/step - loss: 0.0267 - accuracy: 0.9917 - val_loss: 0.0586 - val_accuracy: 0.9775\n"
     ]
    }
   ],
   "source": [
    "with strategy.scope():\n",
    "    print('Training...')\n",
    "    history = model.fit(X_train,\n",
    "                        y_train,\n",
    "                        epochs=EPOCHS,\n",
    "                        batch_size=BATCH_SIZE,\n",
    "                        verbose=1,\n",
    "                        validation_data=(X_test, y_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-17T00:57:29.521310Z",
     "iopub.status.busy": "2020-08-17T00:57:29.519864Z",
     "iopub.status.idle": "2020-08-17T00:57:29.749720Z",
     "shell.execute_reply": "2020-08-17T00:57:29.748973Z"
    },
    "papermill": {
     "duration": 0.263106,
     "end_time": "2020-08-17T00:57:29.749851",
     "exception": false,
     "start_time": "2020-08-17T00:57:29.486745",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x7f16219bc610>"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# This plot will look much better if we train models with more epochs, but anyway here is\n",
    "plt.figure(figsize=(10, 10))\n",
    "plt.title('Accuracy')\n",
    "\n",
    "xaxis = np.arange(len(history.history['accuracy']))\n",
    "plt.plot(xaxis, history.history['accuracy'], label='Train set')\n",
    "plt.plot(xaxis, history.history['val_accuracy'], label='Validation set')\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "papermill": {
     "duration": 0.017458,
     "end_time": "2020-08-17T00:57:29.785325",
     "exception": false,
     "start_time": "2020-08-17T00:57:29.767867",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "## Evaluation\n",
    "\n",
    "In a confusion matrix, we can see how many categories are classified c"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-17T00:57:29.833626Z",
     "iopub.status.busy": "2020-08-17T00:57:29.832514Z",
     "iopub.status.idle": "2020-08-17T00:57:29.835307Z",
     "shell.execute_reply": "2020-08-17T00:57:29.835889Z"
    },
    "papermill": {
     "duration": 0.032937,
     "end_time": "2020-08-17T00:57:29.836044",
     "exception": false,
     "start_time": "2020-08-17T00:57:29.803107",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "def plot_confusion_matrix(X_test, y_test, model):\n",
    "    y_pred = model.predict(X_test)\n",
    "    y_pred = [np.argmax(i) for i in model.predict(X_test)]\n",
    "\n",
    "    con_mat = tf.math.confusion_matrix(labels=y_test, predictions=y_pred).numpy()\n",
    "\n",
    "    con_mat_norm = np.around(con_mat.astype('float') / con_mat.sum(axis=1)[:, np.newaxis], decimals=2)\n",
    "    label_names = list(range(len(con_mat_norm)))\n",
    "\n",
    "    con_mat_df = pd.DataFrame(con_mat_norm,\n",
    "                              index=label_names, \n",
    "                              columns=label_names)\n",
    "\n",
    "    figure = plt.figure(figsize=(10, 10))\n",
    "    sns.heatmap(con_mat_df, cmap=plt.cm.Blues, annot=True)\n",
    "    plt.ylabel('True label')\n",
    "    plt.xlabel('Predicted label')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-17T00:57:29.882110Z",
     "iopub.status.busy": "2020-08-17T00:57:29.880987Z",
     "iopub.status.idle": "2020-08-17T00:57:30.709523Z",
     "shell.execute_reply": "2020-08-17T00:57:30.708449Z"
    },
    "papermill": {
     "duration": 0.855641,
     "end_time": "2020-08-17T00:57:30.709750",
     "exception": false,
     "start_time": "2020-08-17T00:57:29.854109",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy: 97.75%\n"
     ]
    }
   ],
   "source": [
    "scores = model.evaluate(X_test, y_test, verbose=0)\n",
    "print(\"Accuracy: %.2f%%\" % (scores[1] * 100))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-17T00:57:30.755398Z",
     "iopub.status.busy": "2020-08-17T00:57:30.754569Z",
     "iopub.status.idle": "2020-08-17T00:57:43.024221Z",
     "shell.execute_reply": "2020-08-17T00:57:43.024851Z"
    },
    "papermill": {
     "duration": 12.296648,
     "end_time": "2020-08-17T00:57:43.025030",
     "exception": false,
     "start_time": "2020-08-17T00:57:30.728382",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_confusion_matrix(X_test, y_test, model)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 0.018597,
     "end_time": "2020-08-17T00:57:43.063954",
     "exception": false,
     "start_time": "2020-08-17T00:57:43.045357",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "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": "3.7.6"
  },
  "papermill": {
   "duration": 219.095827,
   "end_time": "2020-08-17T00:57:44.149707",
   "environment_variables": {},
   "exception": null,
   "input_path": "__notebook__.ipynb",
   "output_path": "__notebook__.ipynb",
   "parameters": {},
   "start_time": "2020-08-17T00:54:05.053880",
   "version": "2.1.0"
  },
  "widgets": {
   "application/vnd.jupyter.widget-state+json": {
    "state": {
     "049790b7d0e24bb192f1de25e486d2f6": {
      "model_module": "@jupyter-widgets/controls",
      "model_module_version": "1.5.0",
      "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": ""
      }
     },
     "2905d02e54134106b6569b15080816a4": {
      "model_module": "@jupyter-widgets/controls",
      "model_module_version": "1.5.0",
      "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_7cc5c0ba9595454187e23220f9b0cde2",
        "IPY_MODEL_e6ea23ca65aa4987aed8da5aa13770bc"
       ],
       "layout": "IPY_MODEL_ebf25ce5f1904beb98ad2543cde0ea8d"
      }
     },
     "29c87b0192f5439c94944e2360824ff0": {
      "model_module": "@jupyter-widgets/base",
      "model_module_version": "1.2.0",
      "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
      }
     },
     "2ccbab59f73c4d4195842017147bfa60": {
      "model_module": "@jupyter-widgets/controls",
      "model_module_version": "1.5.0",
      "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_636ec27e93c34ab4aa89eaec5473a169",
       "placeholder": "​",
       "style": "IPY_MODEL_304c63e00e2747d0b25944717ce82af7",
       "value": " 456k/456k [00:00&lt;00:00, 1.04MB/s]"
      }
     },
     "304c63e00e2747d0b25944717ce82af7": {
      "model_module": "@jupyter-widgets/controls",
      "model_module_version": "1.5.0",
      "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": ""
      }
     },
     "308d0978b54c4c3e85de312a734cbbec": {
      "model_module": "@jupyter-widgets/base",
      "model_module_version": "1.2.0",
      "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
      }
     },
     "325c77d0e0f744ef9da9600eb2c82789": {
      "model_module": "@jupyter-widgets/controls",
      "model_module_version": "1.5.0",
      "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"
      }
     },
     "35943d38b5e64218b5fc6c39da912696": {
      "model_module": "@jupyter-widgets/controls",
      "model_module_version": "1.5.0",
      "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_92423912efdb4854b4277de353d5cb0d",
       "max": 898823,
       "min": 0,
       "orientation": "horizontal",
       "style": "IPY_MODEL_fb4fb115fff7489dad26a832289599c9",
       "value": 898823
      }
     },
     "3a3be04faafb4d5d998def1e268bf293": {
      "model_module": "@jupyter-widgets/controls",
      "model_module_version": "1.5.0",
      "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": ""
      }
     },
     "41106e7945a740bfa1a5a774000d157b": {
      "model_module": "@jupyter-widgets/base",
      "model_module_version": "1.2.0",
      "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
      }
     },
     "56aaf7600b50474084971338588bbc29": {
      "model_module": "@jupyter-widgets/base",
      "model_module_version": "1.2.0",
      "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
      }
     },
     "5c0be40881da4dc8a58205dfef39c97d": {
      "model_module": "@jupyter-widgets/controls",
      "model_module_version": "1.5.0",
      "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": ""
      }
     },
     "5cde7b456d674588b9cad774e6a65d28": {
      "model_module": "@jupyter-widgets/controls",
      "model_module_version": "1.5.0",
      "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_b06ed1efeb0a48a8a86d99ef43345b5b",
       "max": 456318,
       "min": 0,
       "orientation": "horizontal",
       "style": "IPY_MODEL_e173b9a1ff844e2491f91aaa2be39c71",
       "value": 456318
      }
     },
     "636ec27e93c34ab4aa89eaec5473a169": {
      "model_module": "@jupyter-widgets/base",
      "model_module_version": "1.2.0",
      "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
      }
     },
     "71a1db0fca734d168e7ef2a7d992d511": {
      "model_module": "@jupyter-widgets/controls",
      "model_module_version": "1.5.0",
      "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"
      }
     },
     "7cc5c0ba9595454187e23220f9b0cde2": {
      "model_module": "@jupyter-widgets/controls",
      "model_module_version": "1.5.0",
      "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_8ca80dd822c44e68872af1062c1eaa9c",
       "max": 657434796,
       "min": 0,
       "orientation": "horizontal",
       "style": "IPY_MODEL_71a1db0fca734d168e7ef2a7d992d511",
       "value": 657434796
      }
     },
     "8ba5b5d48551478aa619a303aca887c2": {
      "model_module": "@jupyter-widgets/controls",
      "model_module_version": "1.5.0",
      "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_35943d38b5e64218b5fc6c39da912696",
        "IPY_MODEL_ec377fa71ab34c35a1e83c00ee224a8a"
       ],
       "layout": "IPY_MODEL_308d0978b54c4c3e85de312a734cbbec"
      }
     },
     "8ca80dd822c44e68872af1062c1eaa9c": {
      "model_module": "@jupyter-widgets/base",
      "model_module_version": "1.2.0",
      "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
      }
     },
     "92423912efdb4854b4277de353d5cb0d": {
      "model_module": "@jupyter-widgets/base",
      "model_module_version": "1.2.0",
      "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
      }
     },
     "9ba96aaa3afc4a538dd5804724aea174": {
      "model_module": "@jupyter-widgets/base",
      "model_module_version": "1.2.0",
      "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
      }
     },
     "b06ed1efeb0a48a8a86d99ef43345b5b": {
      "model_module": "@jupyter-widgets/base",
      "model_module_version": "1.2.0",
      "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
      }
     },
     "b071dc5a38c14f5584fd81e7d3076635": {
      "model_module": "@jupyter-widgets/base",
      "model_module_version": "1.2.0",
      "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
      }
     },
     "b13ad60327fc47ed85d26339648c1175": {
      "model_module": "@jupyter-widgets/controls",
      "model_module_version": "1.5.0",
      "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_9ba96aaa3afc4a538dd5804724aea174",
       "placeholder": "​",
       "style": "IPY_MODEL_3a3be04faafb4d5d998def1e268bf293",
       "value": " 481/481 [00:33&lt;00:00, 14.2B/s]"
      }
     },
     "df64321df6a84519a33a6b371cbe8f05": {
      "model_module": "@jupyter-widgets/controls",
      "model_module_version": "1.5.0",
      "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_ea264cbf2b1e439c939fb92137bdd2c4",
        "IPY_MODEL_b13ad60327fc47ed85d26339648c1175"
       ],
       "layout": "IPY_MODEL_f605c3e5fc434b139762f599a611ea01"
      }
     },
     "e173b9a1ff844e2491f91aaa2be39c71": {
      "model_module": "@jupyter-widgets/controls",
      "model_module_version": "1.5.0",
      "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"
      }
     },
     "e2486b910c4142e3affeae7663e1a987": {
      "model_module": "@jupyter-widgets/controls",
      "model_module_version": "1.5.0",
      "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_5cde7b456d674588b9cad774e6a65d28",
        "IPY_MODEL_2ccbab59f73c4d4195842017147bfa60"
       ],
       "layout": "IPY_MODEL_41106e7945a740bfa1a5a774000d157b"
      }
     },
     "e6ea23ca65aa4987aed8da5aa13770bc": {
      "model_module": "@jupyter-widgets/controls",
      "model_module_version": "1.5.0",
      "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_56aaf7600b50474084971338588bbc29",
       "placeholder": "​",
       "style": "IPY_MODEL_5c0be40881da4dc8a58205dfef39c97d",
       "value": " 657M/657M [00:32&lt;00:00, 20.3MB/s]"
      }
     },
     "ea264cbf2b1e439c939fb92137bdd2c4": {
      "model_module": "@jupyter-widgets/controls",
      "model_module_version": "1.5.0",
      "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_29c87b0192f5439c94944e2360824ff0",
       "max": 481,
       "min": 0,
       "orientation": "horizontal",
       "style": "IPY_MODEL_325c77d0e0f744ef9da9600eb2c82789",
       "value": 481
      }
     },
     "ebf25ce5f1904beb98ad2543cde0ea8d": {
      "model_module": "@jupyter-widgets/base",
      "model_module_version": "1.2.0",
      "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
      }
     },
     "ec377fa71ab34c35a1e83c00ee224a8a": {
      "model_module": "@jupyter-widgets/controls",
      "model_module_version": "1.5.0",
      "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_b071dc5a38c14f5584fd81e7d3076635",
       "placeholder": "​",
       "style": "IPY_MODEL_049790b7d0e24bb192f1de25e486d2f6",
       "value": " 899k/899k [00:01&lt;00:00, 486kB/s]"
      }
     },
     "f605c3e5fc434b139762f599a611ea01": {
      "model_module": "@jupyter-widgets/base",
      "model_module_version": "1.2.0",
      "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
      }
     },
     "fb4fb115fff7489dad26a832289599c9": {
      "model_module": "@jupyter-widgets/controls",
      "model_module_version": "1.5.0",
      "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"
      }
     }
    },
    "version_major": 2,
    "version_minor": 0
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}