class_imbalance.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "949ea94b-fe31-4e36-a450-2d9d22a247d5",
   "metadata": {},
   "source": [
    "# Sentence encoders: Class imbalance"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "33ee072e-9578-4af2-8b13-4fc56a1a5474",
   "metadata": {},
   "source": [
    "This notebook investigates some of the approaches that can be taken to class imbalances when using tensorflow sentence encoders on a [Kaggle essay scores](https://www.kaggle.com/competitions/learning-agency-lab-automated-essay-scoring-2) dataset of ~17,000 essays. The strategies explored here include perfectly balanced classes (but with less data), each class having a maximum of x2, x4, and x8 that of the most underepresented class, and including all the data. The model is fit and evaluated on the embeddings for each approach 4 times to obtain more stable accuracy averages. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "5a9104aa-d3ee-4030-b29f-a3b63c59df01",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>essay_id</th>\n",
       "      <th>full_text</th>\n",
       "      <th>score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>000d118</td>\n",
       "      <td>Many people have car where they live. The thin...</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>000fe60</td>\n",
       "      <td>I am a scientist at NASA that is discussing th...</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>001ab80</td>\n",
       "      <td>People always wish they had the same technolog...</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>001bdc0</td>\n",
       "      <td>We all heard about Venus, the planet without a...</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>002ba53</td>\n",
       "      <td>Dear, State Senator\\n\\nThis is a letter to arg...</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  essay_id                                          full_text  score\n",
       "0  000d118  Many people have car where they live. The thin...      3\n",
       "1  000fe60  I am a scientist at NASA that is discussing th...      3\n",
       "2  001ab80  People always wish they had the same technolog...      4\n",
       "3  001bdc0  We all heard about Venus, the planet without a...      4\n",
       "4  002ba53  Dear, State Senator\\n\\nThis is a letter to arg...      3"
      ]
     },
     "execution_count": 1,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "\n",
    "train = pd.read_csv('data/train.csv')\n",
    "train.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c80bbd42-77ea-4d23-a052-69c482c412a2",
   "metadata": {},
   "source": [
    "## Preprocess text"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "30950f63-ef50-4cc8-9422-55715bb36b1b",
   "metadata": {},
   "source": [
    "Common preprocessing steps are applied to simplify the texts and reduce potential noise."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "c506b32b-643d-4fcd-8dff-da1af1c16889",
   "metadata": {},
   "outputs": [],
   "source": [
    "import re\n",
    "import string\n",
    "import nltk\n",
    "from nltk.tokenize import word_tokenize\n",
    "from nltk.corpus import stopwords\n",
    "from nltk.stem import WordNetLemmatizer\n",
    "\n",
    "def preprocess_text(text):\n",
    "\n",
    "    # removes unprintable characters\n",
    "    unprintable = r\"[^\\w\\s]\"\n",
    "    text = re.sub(unprintable, '', text)\n",
    "\n",
    "    # removes non-breaking space characters\n",
    "    non_breaking = '\\xa0'\n",
    "    text = re.sub(non_breaking, ' ', text)\n",
    "\n",
    "    # removes newline characters\n",
    "    newline = '\\n'\n",
    "    text = re.sub(newline, ' ', text)\n",
    "\n",
    "    # removes punctuation\n",
    "    text = text.translate(str.maketrans('', '', string.punctuation))\n",
    "    \n",
    "    # lowercase the text\n",
    "    text = text.lower()\n",
    "    \n",
    "    # tokenization\n",
    "    tokens = word_tokenize(text)\n",
    "    \n",
    "    # remove stopwords\n",
    "    stop_words = set(stopwords.words('english'))\n",
    "    tokens = [token for token in tokens if token not in stop_words]\n",
    "    \n",
    "    # lemmatization\n",
    "    lemmatizer = WordNetLemmatizer()\n",
    "    tokens = [lemmatizer.lemmatize(token) for token in tokens]\n",
    "    \n",
    "    # join tokens back into a string\n",
    "    preprocessed_text = ' '.join(tokens)\n",
    "    \n",
    "    return preprocessed_text"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c3ef8230-75d4-45c4-9275-0ff1204e0bcf",
   "metadata": {},
   "source": [
    "## Misspelling function"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5fcbe45d-0f28-4000-a91b-1443d5859beb",
   "metadata": {},
   "source": [
    "A misspelling function splits the essay into words and either drops or corrects unknown words."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "e712edc1-6cb0-43b2-a1d5-0ac19999fc32",
   "metadata": {},
   "outputs": [],
   "source": [
    "from spellchecker import SpellChecker\n",
    "\n",
    "# creates a spellckecker object\n",
    "spell = SpellChecker()\n",
    "\n",
    "# returns a count of misspelled words\n",
    "def misspellings(essay, drop=True):\n",
    "\n",
    "    # the essay is split into words\n",
    "    words = essay.split()\n",
    "\n",
    "    # either drops or corrects unknown words\n",
    "    final_words = []\n",
    "    if drop == True:\n",
    "        # loops through each word\n",
    "        for word in words:\n",
    "            \n",
    "            # checks if the word is known\n",
    "            if spell.known([word]):\n",
    "                \n",
    "                # appends to a list of words to return\n",
    "                final_words.append(word)\n",
    "    \n",
    "    else:\n",
    "        # loops through each word\n",
    "        for word in words:\n",
    "            \n",
    "            # checks if the word is known\n",
    "            if spell.known([word]):\n",
    "                \n",
    "                # appends to a list of words to return\n",
    "                final_words.append(word)\n",
    "\n",
    "            else:\n",
    "                # attempts to correct the word\n",
    "                correction = spell.correction(word)\n",
    "\n",
    "                # appends the correction if one was found\n",
    "                if correction != None:\n",
    "                    # appends to a list of words to return\n",
    "                    final_words.append(correction)\n",
    "\n",
    "    # joins the words back together with a space \n",
    "    return ' '.join(final_words)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "74f912ef-9844-4697-8bdd-3979685a378a",
   "metadata": {},
   "source": [
    "## Model functions"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "008ce65e-50bf-4235-aa41-e0e3adf0577c",
   "metadata": {},
   "source": [
    "The sentence encoder model is first loaded-in. A number of functions are then defined to split the essay texts in into a variable number of words, obtain the embeddings, fit a model and then record accuracy results."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "340d34ba-8b24-4d10-a950-984d2e7eeb11",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From C:\\Users\\Jack-\\AppData\\Local\\Programs\\Python\\Python311\\Lib\\site-packages\\tf_keras\\src\\losses.py:2976: The name tf.losses.sparse_softmax_cross_entropy is deprecated. Please use tf.compat.v1.losses.sparse_softmax_cross_entropy instead.\n",
      "\n",
      "WARNING:tensorflow:From C:\\Users\\Jack-\\AppData\\Local\\Programs\\Python\\Python311\\Lib\\site-packages\\tensorflow_hub\\resolver.py:120: The name tf.gfile.MakeDirs is deprecated. Please use tf.io.gfile.makedirs instead.\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From C:\\Users\\Jack-\\AppData\\Local\\Programs\\Python\\Python311\\Lib\\site-packages\\tensorflow_hub\\resolver.py:120: The name tf.gfile.MakeDirs is deprecated. Please use tf.io.gfile.makedirs instead.\n",
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From C:\\Users\\Jack-\\AppData\\Local\\Programs\\Python\\Python311\\Lib\\site-packages\\tensorflow_hub\\module_v2.py:126: The name tf.saved_model.load_v2 is deprecated. Please use tf.compat.v2.saved_model.load instead.\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From C:\\Users\\Jack-\\AppData\\Local\\Programs\\Python\\Python311\\Lib\\site-packages\\tensorflow_hub\\module_v2.py:126: The name tf.saved_model.load_v2 is deprecated. Please use tf.compat.v2.saved_model.load instead.\n",
      "\n"
     ]
    }
   ],
   "source": [
    "import tensorflow as tf\n",
    "import tensorflow_hub as hub\n",
    "from sklearn.model_selection import train_test_split\n",
    "from tensorflow.keras.utils import to_categorical\n",
    "from tensorflow.keras.models import Sequential\n",
    "from tensorflow.keras.layers import Dense, Input, Embedding\n",
    "from sklearn.metrics import cohen_kappa_score\n",
    "\n",
    "# downloads the model to obtain embeddings \n",
    "model_url = \"https://tfhub.dev/google/universal-sentence-encoder/4\"\n",
    "encoder_model = hub.load(model_url)\n",
    "\n",
    "# returns the essay embeddings\n",
    "def encode_sentences(essays):\n",
    "    embeddings = encoder_model(essays)\n",
    "    return embeddings.numpy()\n",
    "\n",
    "# creates the model\n",
    "def create_model(embedding_dim):\n",
    "    model = Sequential([\n",
    "          Input(shape=(embedding_dim,)),\n",
    "          Dense(64, activation='relu'),\n",
    "          Dense(32, activation='relu'),\n",
    "          Dense(7, activation='softmax')\n",
    "      ])\n",
    "    model.compile(optimizer='adam', loss='categorical_crossentropy', metrics=['accuracy'])\n",
    "    return model\n",
    "\n",
    "# fits the model on the training data and evaluates accuracy\n",
    "def fit_model(embeddings, scores, accuracy_df, class_balance):\n",
    "\n",
    "    # defines how many times the model should be fit and evaluated to obain a more stable accuracy average\n",
    "    turns = 4\n",
    "\n",
    "    # stores each accuracy result in order to obtain an average\n",
    "    accuracy_scores = []\n",
    "\n",
    "    # loops through each turn\n",
    "    for i in range(turns):\n",
    "\n",
    "        # splits the data into train and validation sets\n",
    "        x_train, x_val, y_train, y_val = train_test_split(embeddings, scores, test_size=0.3)\n",
    "    \n",
    "        # creates and embedding dimension and uses it to create a model\n",
    "        embedding_dim = embeddings[0].shape[0]\n",
    "        model = create_model(embedding_dim)\n",
    "    \n",
    "        # converts essay scores to one hot encodings\n",
    "        y_train_one_hot = to_categorical(y_train)\n",
    "        y_val_one_hot = to_categorical(y_val)\n",
    "    \n",
    "        # fits the model\n",
    "        model.fit(x_train, y_train_one_hot, epochs=12, batch_size=32, validation_data=(x_val, y_val_one_hot)) \n",
    "    \n",
    "        # retrieves predictions \n",
    "        predictions = model.predict(x_val)\n",
    "    \n",
    "        # converts raw predictions (decimals) into final predictions (1 - 6 integers)\n",
    "        converted_preds = [np.argmax(pred) for pred in predictions]\n",
    "        converted_y_val = [np.argmax(y) for y in y_val_one_hot]\n",
    "    \n",
    "        # appends rounded Cohen Kappa scores \n",
    "        accuracy_scores.append(round(cohen_kappa_score(converted_preds, converted_y_val, weights='quadratic'), 2))\n",
    "    \n",
    "    # adds the average accuracy score to the accuracy dataframe\n",
    "    accuracy_df.loc[0, class_balance] = round(np.mean(accuracy_scores), 2)\n",
    "    \n",
    "    return accuracy_df"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "12be6220-d74f-4ea2-9edf-a83a7cecd2ce",
   "metadata": {},
   "source": [
    "## Class imbalance"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "da1b9efd-4075-44f5-9fe3-b62992d638e6",
   "metadata": {},
   "source": [
    "Here the class imalances are addressed. First we list out the value counts for scores. Then we create a series of spliced dataframes, including perfect class balance (with the same number of each scores, 156), reduced class imbalance (with a maximum of x2, x4, and x8 the smallest value count for each class) and all the data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "73b04481-4f72-49d0-bce6-670cce4cf125",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "score\n",
       "3    6280\n",
       "2    4723\n",
       "4    3926\n",
       "1    1252\n",
       "5     970\n",
       "6     156\n",
       "Name: count, dtype: int64"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# lists value counts\n",
    "train.score.value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "43d306b1-fecd-4a75-88d7-cb4a9f8189c5",
   "metadata": {},
   "outputs": [],
   "source": [
    "import random\n",
    "\n",
    "# preprocesses the texts\n",
    "train['full_text'] = [preprocess_text(x) for x in train['full_text']]\n",
    "\n",
    "# limits each essay to the first 300 words\n",
    "texts = [' '.join(x.split()[0:300]) for x in train['full_text']]\n",
    "\n",
    "# names of different class balance dataframes\n",
    "class_balances = ['balanced', 'x2', 'x4', 'x8', 'all']\n",
    "\n",
    "# the different maximum number for each class in the different dataframes\n",
    "target_nums = [156, 156*2, 156*4, 156*8, np.floor(len(train)/6)]\n",
    "\n",
    "# selects a specified maxium number of random essays from each score distribution\n",
    "def splice_df(target_num):\n",
    "    random_index = []\n",
    "    for i in range(6):\n",
    "        num_available = len(train.loc[train.score==i+1])\n",
    "        if num_available <= target_num:\n",
    "            target_num = num_available\n",
    "        random_index.extend( random.sample(list(train.loc[train.score==i+1].index), target_num) )\n",
    "    return train.loc[random_index].reset_index(drop=True)\n",
    "\n",
    "# creates a series of dataframes \n",
    "lists = {}\n",
    "for i, class_balance in enumerate(class_balances):\n",
    "    lists[class_balance] = splice_df(target_nums[i])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "145bb180-fab5-4d70-8cb8-ea6f3f907309",
   "metadata": {},
   "source": [
    "## Class imbalance approaches"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "26cb8762-7256-40f9-b4ef-e206c41d27c6",
   "metadata": {},
   "source": [
    "The relavant class imbalance dataframe is obtained, misspellings dropped, embeddings obtained, model fit and results evaluated."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "174dbfd7-7cf5-4756-b3fd-1b2443337e66",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Dropping misspellings: balanced\n",
      "Preparing embeddings: balanced\n",
      "Fitting model: balanced\n",
      "Epoch 1/12\n",
      "\u001b[1m21/21\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m2s\u001b[0m 15ms/step - accuracy: 0.2319 - loss: 1.9120 - val_accuracy: 0.3345 - val_loss: 1.7923\n",
      "Epoch 2/12\n",
      "\u001b[1m21/21\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 4ms/step - accuracy: 0.3748 - loss: 1.7193 - val_accuracy: 0.3772 - val_loss: 1.5823\n",
      "Epoch 3/12\n",
      "\u001b[1m21/21\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 4ms/step - accuracy: 0.3986 - loss: 1.4950 - val_accuracy: 0.4021 - val_loss: 1.4205\n",
      "Epoch 4/12\n",
      "\u001b[1m21/21\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 4ms/step - accuracy: 0.4819 - loss: 1.3535 - val_accuracy: 0.4413 - val_loss: 1.3316\n",
      "Epoch 5/12\n",
      "\u001b[1m21/21\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 5ms/step - accuracy: 0.5020 - loss: 1.2622 - val_accuracy: 0.4769 - val_loss: 1.2766\n",
      "Epoch 6/12\n",
      "\u001b[1m21/21\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 5ms/step - accuracy: 0.5278 - loss: 1.2003 - val_accuracy: 0.4413 - val_loss: 1.2476\n",
      "Epoch 7/12\n",
      "\u001b[1m21/21\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 5ms/step - accuracy: 0.5648 - loss: 1.1238 - val_accuracy: 0.5160 - val_loss: 1.1987\n",
      "Epoch 8/12\n",
      "\u001b[1m21/21\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.6241 - loss: 1.0710 - val_accuracy: 0.4840 - val_loss: 1.1842\n",
      "Epoch 9/12\n",
      "\u001b[1m21/21\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 4ms/step - accuracy: 0.5865 - loss: 1.0447 - val_accuracy: 0.4947 - val_loss: 1.1685\n",
      "Epoch 10/12\n",
      "\u001b[1m21/21\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 4ms/step - accuracy: 0.6301 - loss: 1.0225 - val_accuracy: 0.5018 - val_loss: 1.1360\n",
      "Epoch 11/12\n",
      "\u001b[1m21/21\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 5ms/step - accuracy: 0.6489 - loss: 0.9403 - val_accuracy: 0.5160 - val_loss: 1.1222\n",
      "Epoch 12/12\n",
      "\u001b[1m21/21\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 4ms/step - accuracy: 0.6084 - loss: 0.9322 - val_accuracy: 0.5267 - val_loss: 1.1176\n",
      "\u001b[1m9/9\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 7ms/step\n",
      "Epoch 1/12\n",
      "\u001b[1m21/21\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m2s\u001b[0m 14ms/step - accuracy: 0.1918 - loss: 1.9295 - val_accuracy: 0.2206 - val_loss: 1.8229\n",
      "Epoch 2/12\n",
      "\u001b[1m21/21\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 4ms/step - accuracy: 0.2549 - loss: 1.7746 - val_accuracy: 0.3238 - val_loss: 1.6246\n",
      "Epoch 3/12\n",
      "\u001b[1m21/21\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 4ms/step - accuracy: 0.3987 - loss: 1.5472 - val_accuracy: 0.3808 - val_loss: 1.4214\n",
      "Epoch 4/12\n",
      "\u001b[1m21/21\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 4ms/step - accuracy: 0.4709 - loss: 1.3652 - val_accuracy: 0.4164 - val_loss: 1.3188\n",
      "Epoch 5/12\n",
      "\u001b[1m21/21\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 4ms/step - accuracy: 0.5629 - loss: 1.1928 - val_accuracy: 0.4626 - val_loss: 1.2691\n",
      "Epoch 6/12\n",
      "\u001b[1m21/21\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 4ms/step - accuracy: 0.5493 - loss: 1.1827 - val_accuracy: 0.4555 - val_loss: 1.2378\n",
      "Epoch 7/12\n",
      "\u001b[1m21/21\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 4ms/step - accuracy: 0.5853 - loss: 1.1234 - val_accuracy: 0.4698 - val_loss: 1.2145\n",
      "Epoch 8/12\n",
      "\u001b[1m21/21\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 9ms/step - accuracy: 0.6216 - loss: 1.0438 - val_accuracy: 0.4840 - val_loss: 1.1845\n",
      "Epoch 9/12\n",
      "\u001b[1m21/21\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 5ms/step - accuracy: 0.6159 - loss: 1.0302 - val_accuracy: 0.4804 - val_loss: 1.1772\n",
      "Epoch 10/12\n",
      "\u001b[1m21/21\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 5ms/step - accuracy: 0.6140 - loss: 1.0168 - val_accuracy: 0.5018 - val_loss: 1.1631\n",
      "Epoch 11/12\n",
      "\u001b[1m21/21\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 5ms/step - accuracy: 0.6704 - loss: 0.9157 - val_accuracy: 0.5018 - val_loss: 1.1615\n",
      "Epoch 12/12\n",
      "\u001b[1m21/21\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 4ms/step - accuracy: 0.6589 - loss: 0.9063 - val_accuracy: 0.4911 - val_loss: 1.1380\n",
      "\u001b[1m9/9\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 8ms/step\n",
      "Epoch 1/12\n",
      "\u001b[1m21/21\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m2s\u001b[0m 14ms/step - accuracy: 0.2440 - loss: 1.9210 - val_accuracy: 0.2847 - val_loss: 1.8318\n",
      "Epoch 2/12\n",
      "\u001b[1m21/21\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.3422 - loss: 1.7731 - val_accuracy: 0.3559 - val_loss: 1.6283\n",
      "Epoch 3/12\n",
      "\u001b[1m21/21\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 4ms/step - accuracy: 0.3788 - loss: 1.5542 - val_accuracy: 0.4306 - val_loss: 1.4349\n",
      "Epoch 4/12\n",
      "\u001b[1m21/21\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.4612 - loss: 1.3466 - val_accuracy: 0.4377 - val_loss: 1.3281\n",
      "Epoch 5/12\n",
      "\u001b[1m21/21\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.5513 - loss: 1.2121 - val_accuracy: 0.4342 - val_loss: 1.2842\n",
      "Epoch 6/12\n",
      "\u001b[1m21/21\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 4ms/step - accuracy: 0.5312 - loss: 1.1401 - val_accuracy: 0.4484 - val_loss: 1.2520\n",
      "Epoch 7/12\n",
      "\u001b[1m21/21\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.6204 - loss: 1.0580 - val_accuracy: 0.4484 - val_loss: 1.2406\n",
      "Epoch 8/12\n",
      "\u001b[1m21/21\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.5914 - loss: 1.0842 - val_accuracy: 0.4769 - val_loss: 1.2279\n",
      "Epoch 9/12\n",
      "\u001b[1m21/21\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 5ms/step - accuracy: 0.6205 - loss: 1.0137 - val_accuracy: 0.4733 - val_loss: 1.2220\n",
      "Epoch 10/12\n",
      "\u001b[1m21/21\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.6774 - loss: 0.9717 - val_accuracy: 0.4804 - val_loss: 1.2002\n",
      "Epoch 11/12\n",
      "\u001b[1m21/21\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 4ms/step - accuracy: 0.6808 - loss: 0.9056 - val_accuracy: 0.5053 - val_loss: 1.1827\n",
      "Epoch 12/12\n",
      "\u001b[1m21/21\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.6791 - loss: 0.8811 - val_accuracy: 0.4840 - val_loss: 1.1812\n",
      "\u001b[1m9/9\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 5ms/step \n",
      "Epoch 1/12\n",
      "\u001b[1m21/21\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 13ms/step - accuracy: 0.2132 - loss: 1.9112 - val_accuracy: 0.2491 - val_loss: 1.8238\n",
      "Epoch 2/12\n",
      "\u001b[1m21/21\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 4ms/step - accuracy: 0.3553 - loss: 1.7561 - val_accuracy: 0.3808 - val_loss: 1.6604\n",
      "Epoch 3/12\n",
      "\u001b[1m21/21\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.4500 - loss: 1.5736 - val_accuracy: 0.4591 - val_loss: 1.4988\n",
      "Epoch 4/12\n",
      "\u001b[1m21/21\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.5082 - loss: 1.3971 - val_accuracy: 0.4626 - val_loss: 1.3785\n",
      "Epoch 5/12\n",
      "\u001b[1m21/21\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.4921 - loss: 1.3095 - val_accuracy: 0.5089 - val_loss: 1.2796\n",
      "Epoch 6/12\n",
      "\u001b[1m21/21\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 4ms/step - accuracy: 0.5354 - loss: 1.2107 - val_accuracy: 0.5089 - val_loss: 1.2237\n",
      "Epoch 7/12\n",
      "\u001b[1m21/21\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 4ms/step - accuracy: 0.5636 - loss: 1.1714 - val_accuracy: 0.5160 - val_loss: 1.1797\n",
      "Epoch 8/12\n",
      "\u001b[1m21/21\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.5508 - loss: 1.0903 - val_accuracy: 0.5125 - val_loss: 1.1536\n",
      "Epoch 9/12\n",
      "\u001b[1m21/21\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 4ms/step - accuracy: 0.5935 - loss: 1.0846 - val_accuracy: 0.5409 - val_loss: 1.1335\n",
      "Epoch 10/12\n",
      "\u001b[1m21/21\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.6099 - loss: 0.9822 - val_accuracy: 0.5231 - val_loss: 1.1659\n",
      "Epoch 11/12\n",
      "\u001b[1m21/21\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 4ms/step - accuracy: 0.6154 - loss: 1.0113 - val_accuracy: 0.5623 - val_loss: 1.0999\n",
      "Epoch 12/12\n",
      "\u001b[1m21/21\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.6488 - loss: 0.9509 - val_accuracy: 0.5623 - val_loss: 1.0821\n",
      "\u001b[1m9/9\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 6ms/step \n",
      "Dropping misspellings: x2\n",
      "Preparing embeddings: x2\n",
      "Fitting model: x2\n",
      "Epoch 1/12\n",
      "\u001b[1m38/38\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m3s\u001b[0m 17ms/step - accuracy: 0.2447 - loss: 1.8963 - val_accuracy: 0.2641 - val_loss: 1.6517\n",
      "Epoch 2/12\n",
      "\u001b[1m38/38\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 5ms/step - accuracy: 0.3085 - loss: 1.5916 - val_accuracy: 0.4466 - val_loss: 1.3552\n",
      "Epoch 3/12\n",
      "\u001b[1m38/38\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 6ms/step - accuracy: 0.4678 - loss: 1.3268 - val_accuracy: 0.4369 - val_loss: 1.2641\n",
      "Epoch 4/12\n",
      "\u001b[1m38/38\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 5ms/step - accuracy: 0.4674 - loss: 1.2334 - val_accuracy: 0.4893 - val_loss: 1.2215\n",
      "Epoch 5/12\n",
      "\u001b[1m38/38\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 5ms/step - accuracy: 0.5004 - loss: 1.1825 - val_accuracy: 0.4757 - val_loss: 1.1964\n",
      "Epoch 6/12\n",
      "\u001b[1m38/38\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 5ms/step - accuracy: 0.5169 - loss: 1.1245 - val_accuracy: 0.5340 - val_loss: 1.1423\n",
      "Epoch 7/12\n",
      "\u001b[1m38/38\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 4ms/step - accuracy: 0.5608 - loss: 1.0625 - val_accuracy: 0.4932 - val_loss: 1.1458\n",
      "Epoch 8/12\n",
      "\u001b[1m38/38\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 5ms/step - accuracy: 0.5840 - loss: 1.0303 - val_accuracy: 0.5223 - val_loss: 1.1156\n",
      "Epoch 9/12\n",
      "\u001b[1m38/38\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 5ms/step - accuracy: 0.6331 - loss: 0.9631 - val_accuracy: 0.5165 - val_loss: 1.1164\n",
      "Epoch 10/12\n",
      "\u001b[1m38/38\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 5ms/step - accuracy: 0.6135 - loss: 0.9478 - val_accuracy: 0.4874 - val_loss: 1.1264\n",
      "Epoch 11/12\n",
      "\u001b[1m38/38\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 5ms/step - accuracy: 0.6510 - loss: 0.9096 - val_accuracy: 0.5282 - val_loss: 1.0884\n",
      "Epoch 12/12\n",
      "\u001b[1m38/38\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 8ms/step - accuracy: 0.6391 - loss: 0.8711 - val_accuracy: 0.5282 - val_loss: 1.1047\n",
      "\u001b[1m17/17\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 6ms/step\n",
      "Epoch 1/12\n",
      "\u001b[1m38/38\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m2s\u001b[0m 11ms/step - accuracy: 0.2767 - loss: 1.8885 - val_accuracy: 0.3728 - val_loss: 1.6332\n",
      "Epoch 2/12\n",
      "\u001b[1m38/38\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.3565 - loss: 1.5372 - val_accuracy: 0.4019 - val_loss: 1.3567\n",
      "Epoch 3/12\n",
      "\u001b[1m38/38\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 4ms/step - accuracy: 0.4288 - loss: 1.3071 - val_accuracy: 0.4155 - val_loss: 1.2804\n",
      "Epoch 4/12\n",
      "\u001b[1m38/38\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 4ms/step - accuracy: 0.4595 - loss: 1.2488 - val_accuracy: 0.4388 - val_loss: 1.2551\n",
      "Epoch 5/12\n",
      "\u001b[1m38/38\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.5104 - loss: 1.1634 - val_accuracy: 0.4388 - val_loss: 1.2264\n",
      "Epoch 6/12\n",
      "\u001b[1m38/38\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 4ms/step - accuracy: 0.5364 - loss: 1.1145 - val_accuracy: 0.4835 - val_loss: 1.2058\n",
      "Epoch 7/12\n",
      "\u001b[1m38/38\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 4ms/step - accuracy: 0.5727 - loss: 1.0774 - val_accuracy: 0.4874 - val_loss: 1.1938\n",
      "Epoch 8/12\n",
      "\u001b[1m38/38\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 4ms/step - accuracy: 0.5848 - loss: 1.0265 - val_accuracy: 0.5049 - val_loss: 1.1790\n",
      "Epoch 9/12\n",
      "\u001b[1m38/38\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 4ms/step - accuracy: 0.6085 - loss: 0.9711 - val_accuracy: 0.5126 - val_loss: 1.1641\n",
      "Epoch 10/12\n",
      "\u001b[1m38/38\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 4ms/step - accuracy: 0.6274 - loss: 0.9497 - val_accuracy: 0.5126 - val_loss: 1.1775\n",
      "Epoch 11/12\n",
      "\u001b[1m38/38\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 4ms/step - accuracy: 0.6571 - loss: 0.9002 - val_accuracy: 0.5184 - val_loss: 1.1488\n",
      "Epoch 12/12\n",
      "\u001b[1m38/38\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.6433 - loss: 0.8846 - val_accuracy: 0.5223 - val_loss: 1.1635\n",
      "\u001b[1m17/17\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step\n",
      "Epoch 1/12\n",
      "\u001b[1m38/38\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 7ms/step - accuracy: 0.2609 - loss: 1.8910 - val_accuracy: 0.2466 - val_loss: 1.7140\n",
      "Epoch 2/12\n",
      "\u001b[1m38/38\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.3052 - loss: 1.6237 - val_accuracy: 0.4136 - val_loss: 1.4346\n",
      "Epoch 3/12\n",
      "\u001b[1m38/38\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.4548 - loss: 1.3780 - val_accuracy: 0.4563 - val_loss: 1.2946\n",
      "Epoch 4/12\n",
      "\u001b[1m38/38\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.5025 - loss: 1.2468 - val_accuracy: 0.4641 - val_loss: 1.2352\n",
      "Epoch 5/12\n",
      "\u001b[1m38/38\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 4ms/step - accuracy: 0.5016 - loss: 1.1666 - val_accuracy: 0.4854 - val_loss: 1.1962\n",
      "Epoch 6/12\n",
      "\u001b[1m38/38\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.5250 - loss: 1.1366 - val_accuracy: 0.4990 - val_loss: 1.1793\n",
      "Epoch 7/12\n",
      "\u001b[1m38/38\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.5714 - loss: 1.0717 - val_accuracy: 0.4680 - val_loss: 1.1744\n",
      "Epoch 8/12\n",
      "\u001b[1m38/38\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.5635 - loss: 1.0369 - val_accuracy: 0.5243 - val_loss: 1.1243\n",
      "Epoch 9/12\n",
      "\u001b[1m38/38\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.5851 - loss: 0.9905 - val_accuracy: 0.5126 - val_loss: 1.1108\n",
      "Epoch 10/12\n",
      "\u001b[1m38/38\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.6176 - loss: 0.9317 - val_accuracy: 0.5243 - val_loss: 1.0998\n",
      "Epoch 11/12\n",
      "\u001b[1m38/38\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.6366 - loss: 0.8890 - val_accuracy: 0.5165 - val_loss: 1.1000\n",
      "Epoch 12/12\n",
      "\u001b[1m38/38\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.6373 - loss: 0.9082 - val_accuracy: 0.5223 - val_loss: 1.0999\n",
      "\u001b[1m17/17\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 4ms/step\n",
      "Epoch 1/12\n",
      "\u001b[1m38/38\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 7ms/step - accuracy: 0.1815 - loss: 1.9122 - val_accuracy: 0.1883 - val_loss: 1.7797\n",
      "Epoch 2/12\n",
      "\u001b[1m38/38\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.3268 - loss: 1.6836 - val_accuracy: 0.4039 - val_loss: 1.4904\n",
      "Epoch 3/12\n",
      "\u001b[1m38/38\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.4326 - loss: 1.3966 - val_accuracy: 0.4369 - val_loss: 1.3475\n",
      "Epoch 4/12\n",
      "\u001b[1m38/38\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.4647 - loss: 1.2594 - val_accuracy: 0.4155 - val_loss: 1.3074\n",
      "Epoch 5/12\n",
      "\u001b[1m38/38\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.4957 - loss: 1.1992 - val_accuracy: 0.4485 - val_loss: 1.2646\n",
      "Epoch 6/12\n",
      "\u001b[1m38/38\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.5501 - loss: 1.1332 - val_accuracy: 0.4563 - val_loss: 1.2468\n",
      "Epoch 7/12\n",
      "\u001b[1m38/38\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.5480 - loss: 1.0722 - val_accuracy: 0.4699 - val_loss: 1.2245\n",
      "Epoch 8/12\n",
      "\u001b[1m38/38\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.5919 - loss: 1.0312 - val_accuracy: 0.4699 - val_loss: 1.2110\n",
      "Epoch 9/12\n",
      "\u001b[1m38/38\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.5930 - loss: 1.0174 - val_accuracy: 0.4893 - val_loss: 1.1937\n",
      "Epoch 10/12\n",
      "\u001b[1m38/38\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.6231 - loss: 0.9520 - val_accuracy: 0.4816 - val_loss: 1.1907\n",
      "Epoch 11/12\n",
      "\u001b[1m38/38\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.6092 - loss: 0.9484 - val_accuracy: 0.4835 - val_loss: 1.1781\n",
      "Epoch 12/12\n",
      "\u001b[1m38/38\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.6668 - loss: 0.8957 - val_accuracy: 0.4835 - val_loss: 1.1810\n",
      "\u001b[1m17/17\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 4ms/step\n",
      "Dropping misspellings: x4\n",
      "Preparing embeddings: x4\n",
      "Fitting model: x4\n",
      "Epoch 1/12\n",
      "\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m3s\u001b[0m 8ms/step - accuracy: 0.2793 - loss: 1.8344 - val_accuracy: 0.4008 - val_loss: 1.4278\n",
      "Epoch 2/12\n",
      "\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 5ms/step - accuracy: 0.4642 - loss: 1.3305 - val_accuracy: 0.4588 - val_loss: 1.2498\n",
      "Epoch 3/12\n",
      "\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 4ms/step - accuracy: 0.4989 - loss: 1.2028 - val_accuracy: 0.4659 - val_loss: 1.2054\n",
      "Epoch 4/12\n",
      "\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 5ms/step - accuracy: 0.5155 - loss: 1.1345 - val_accuracy: 0.5005 - val_loss: 1.1599\n",
      "Epoch 5/12\n",
      "\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 6ms/step - accuracy: 0.5537 - loss: 1.0692 - val_accuracy: 0.4995 - val_loss: 1.1436\n",
      "Epoch 6/12\n",
      "\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.5796 - loss: 1.0286 - val_accuracy: 0.4995 - val_loss: 1.1270\n",
      "Epoch 7/12\n",
      "\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 5ms/step - accuracy: 0.5920 - loss: 0.9811 - val_accuracy: 0.5097 - val_loss: 1.1276\n",
      "Epoch 8/12\n",
      "\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.6042 - loss: 0.9645 - val_accuracy: 0.5127 - val_loss: 1.1046\n",
      "Epoch 9/12\n",
      "\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.6044 - loss: 0.9465 - val_accuracy: 0.5178 - val_loss: 1.1329\n",
      "Epoch 10/12\n",
      "\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 5ms/step - accuracy: 0.6240 - loss: 0.9085 - val_accuracy: 0.5148 - val_loss: 1.1125\n",
      "Epoch 11/12\n",
      "\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 4ms/step - accuracy: 0.6395 - loss: 0.8689 - val_accuracy: 0.5025 - val_loss: 1.1246\n",
      "Epoch 12/12\n",
      "\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 5ms/step - accuracy: 0.6480 - loss: 0.8672 - val_accuracy: 0.5209 - val_loss: 1.1471\n",
      "\u001b[1m31/31\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 6ms/step\n",
      "Epoch 1/12\n",
      "\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m2s\u001b[0m 6ms/step - accuracy: 0.3006 - loss: 1.8370 - val_accuracy: 0.3937 - val_loss: 1.4464\n",
      "Epoch 2/12\n",
      "\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.4366 - loss: 1.3646 - val_accuracy: 0.4354 - val_loss: 1.2840\n",
      "Epoch 3/12\n",
      "\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.4674 - loss: 1.2413 - val_accuracy: 0.4730 - val_loss: 1.2196\n",
      "Epoch 4/12\n",
      "\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.4954 - loss: 1.1575 - val_accuracy: 0.5097 - val_loss: 1.1609\n",
      "Epoch 5/12\n",
      "\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.5356 - loss: 1.0957 - val_accuracy: 0.5005 - val_loss: 1.1387\n",
      "Epoch 6/12\n",
      "\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.5715 - loss: 1.0224 - val_accuracy: 0.5229 - val_loss: 1.1126\n",
      "Epoch 7/12\n",
      "\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.5770 - loss: 1.0121 - val_accuracy: 0.5270 - val_loss: 1.0970\n",
      "Epoch 8/12\n",
      "\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.6126 - loss: 0.9604 - val_accuracy: 0.5239 - val_loss: 1.0958\n",
      "Epoch 9/12\n",
      "\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 4ms/step - accuracy: 0.6147 - loss: 0.9129 - val_accuracy: 0.5300 - val_loss: 1.1081\n",
      "Epoch 10/12\n",
      "\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.6134 - loss: 0.8993 - val_accuracy: 0.5249 - val_loss: 1.0867\n",
      "Epoch 11/12\n",
      "\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 4ms/step - accuracy: 0.6434 - loss: 0.8744 - val_accuracy: 0.5219 - val_loss: 1.0995\n",
      "Epoch 12/12\n",
      "\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.6615 - loss: 0.8307 - val_accuracy: 0.5361 - val_loss: 1.1071\n",
      "\u001b[1m31/31\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step\n",
      "Epoch 1/12\n",
      "\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m2s\u001b[0m 6ms/step - accuracy: 0.3308 - loss: 1.7861 - val_accuracy: 0.4151 - val_loss: 1.4348\n",
      "Epoch 2/12\n",
      "\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.4620 - loss: 1.3133 - val_accuracy: 0.4425 - val_loss: 1.2959\n",
      "Epoch 3/12\n",
      "\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.4852 - loss: 1.1996 - val_accuracy: 0.4629 - val_loss: 1.2282\n",
      "Epoch 4/12\n",
      "\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.5272 - loss: 1.1000 - val_accuracy: 0.4842 - val_loss: 1.1889\n",
      "Epoch 5/12\n",
      "\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.5851 - loss: 1.0447 - val_accuracy: 0.4975 - val_loss: 1.1692\n",
      "Epoch 6/12\n",
      "\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.5877 - loss: 0.9991 - val_accuracy: 0.4934 - val_loss: 1.1482\n",
      "Epoch 7/12\n",
      "\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.6110 - loss: 0.9716 - val_accuracy: 0.4954 - val_loss: 1.1368\n",
      "Epoch 8/12\n",
      "\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 4ms/step - accuracy: 0.6154 - loss: 0.9169 - val_accuracy: 0.5097 - val_loss: 1.1337\n",
      "Epoch 9/12\n",
      "\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.6279 - loss: 0.8914 - val_accuracy: 0.4954 - val_loss: 1.1381\n",
      "Epoch 10/12\n",
      "\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.6510 - loss: 0.8645 - val_accuracy: 0.5168 - val_loss: 1.1340\n",
      "Epoch 11/12\n",
      "\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.6300 - loss: 0.8813 - val_accuracy: 0.5086 - val_loss: 1.1446\n",
      "Epoch 12/12\n",
      "\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 4ms/step - accuracy: 0.6648 - loss: 0.8131 - val_accuracy: 0.5036 - val_loss: 1.1502\n",
      "\u001b[1m31/31\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step\n",
      "Epoch 1/12\n",
      "\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m2s\u001b[0m 5ms/step - accuracy: 0.2861 - loss: 1.8333 - val_accuracy: 0.4201 - val_loss: 1.4401\n",
      "Epoch 2/12\n",
      "\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.4290 - loss: 1.3862 - val_accuracy: 0.4680 - val_loss: 1.2512\n",
      "Epoch 3/12\n",
      "\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 5ms/step - accuracy: 0.4605 - loss: 1.2533 - val_accuracy: 0.4812 - val_loss: 1.1933\n",
      "Epoch 4/12\n",
      "\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 3ms/step - accuracy: 0.5073 - loss: 1.1554 - val_accuracy: 0.4934 - val_loss: 1.1653\n",
      "Epoch 5/12\n",
      "\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 4ms/step - accuracy: 0.5048 - loss: 1.1119 - val_accuracy: 0.5117 - val_loss: 1.1233\n",
      "Epoch 6/12\n",
      "\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.5476 - loss: 1.0579 - val_accuracy: 0.5188 - val_loss: 1.0982\n",
      "Epoch 7/12\n",
      "\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.5699 - loss: 1.0063 - val_accuracy: 0.5351 - val_loss: 1.0904\n",
      "Epoch 8/12\n",
      "\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 5ms/step - accuracy: 0.5817 - loss: 0.9795 - val_accuracy: 0.5198 - val_loss: 1.0871\n",
      "Epoch 9/12\n",
      "\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.6060 - loss: 0.9566 - val_accuracy: 0.5198 - val_loss: 1.0803\n",
      "Epoch 10/12\n",
      "\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.6286 - loss: 0.9299 - val_accuracy: 0.5219 - val_loss: 1.0769\n",
      "Epoch 11/12\n",
      "\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.6397 - loss: 0.8852 - val_accuracy: 0.5086 - val_loss: 1.0851\n",
      "Epoch 12/12\n",
      "\u001b[1m72/72\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.6382 - loss: 0.8850 - val_accuracy: 0.5209 - val_loss: 1.0740\n",
      "\u001b[1m31/31\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m2s\u001b[0m 61ms/step\n",
      "Dropping misspellings: x8\n",
      "Preparing embeddings: x8\n",
      "Fitting model: x8\n",
      "Epoch 1/12\n",
      "\u001b[1m134/134\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m3s\u001b[0m 5ms/step - accuracy: 0.3210 - loss: 1.6990 - val_accuracy: 0.4401 - val_loss: 1.2911\n",
      "Epoch 2/12\n",
      "\u001b[1m134/134\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.4988 - loss: 1.2172 - val_accuracy: 0.4771 - val_loss: 1.1994\n",
      "Epoch 3/12\n",
      "\u001b[1m134/134\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.5202 - loss: 1.1154 - val_accuracy: 0.5163 - val_loss: 1.1307\n",
      "Epoch 4/12\n",
      "\u001b[1m134/134\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.5374 - loss: 1.0618 - val_accuracy: 0.5370 - val_loss: 1.0905\n",
      "Epoch 5/12\n",
      "\u001b[1m134/134\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.5611 - loss: 1.0137 - val_accuracy: 0.5556 - val_loss: 1.0652\n",
      "Epoch 6/12\n",
      "\u001b[1m134/134\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.5792 - loss: 0.9946 - val_accuracy: 0.5550 - val_loss: 1.0588\n",
      "Epoch 7/12\n",
      "\u001b[1m134/134\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.5927 - loss: 0.9599 - val_accuracy: 0.5398 - val_loss: 1.0479\n",
      "Epoch 8/12\n",
      "\u001b[1m134/134\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.6016 - loss: 0.9459 - val_accuracy: 0.5626 - val_loss: 1.0386\n",
      "Epoch 9/12\n",
      "\u001b[1m134/134\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.6219 - loss: 0.9129 - val_accuracy: 0.5517 - val_loss: 1.0442\n",
      "Epoch 10/12\n",
      "\u001b[1m134/134\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.6344 - loss: 0.8928 - val_accuracy: 0.5566 - val_loss: 1.0490\n",
      "Epoch 11/12\n",
      "\u001b[1m134/134\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.6495 - loss: 0.8715 - val_accuracy: 0.5599 - val_loss: 1.0424\n",
      "Epoch 12/12\n",
      "\u001b[1m134/134\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.6441 - loss: 0.8738 - val_accuracy: 0.5479 - val_loss: 1.0474\n",
      "\u001b[1m58/58\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step\n",
      "Epoch 1/12\n",
      "\u001b[1m134/134\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m2s\u001b[0m 4ms/step - accuracy: 0.2964 - loss: 1.6904 - val_accuracy: 0.4581 - val_loss: 1.2781\n",
      "Epoch 2/12\n",
      "\u001b[1m134/134\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.5015 - loss: 1.2062 - val_accuracy: 0.5114 - val_loss: 1.1651\n",
      "Epoch 3/12\n",
      "\u001b[1m134/134\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.5402 - loss: 1.1134 - val_accuracy: 0.5202 - val_loss: 1.1279\n",
      "Epoch 4/12\n",
      "\u001b[1m134/134\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.5454 - loss: 1.0638 - val_accuracy: 0.5343 - val_loss: 1.0879\n",
      "Epoch 5/12\n",
      "\u001b[1m134/134\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.5620 - loss: 1.0154 - val_accuracy: 0.5485 - val_loss: 1.0631\n",
      "Epoch 6/12\n",
      "\u001b[1m134/134\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.5781 - loss: 0.9882 - val_accuracy: 0.5300 - val_loss: 1.0896\n",
      "Epoch 7/12\n",
      "\u001b[1m134/134\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.5957 - loss: 0.9617 - val_accuracy: 0.5594 - val_loss: 1.0449\n",
      "Epoch 8/12\n",
      "\u001b[1m134/134\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.6070 - loss: 0.9338 - val_accuracy: 0.5637 - val_loss: 1.0297\n",
      "Epoch 9/12\n",
      "\u001b[1m134/134\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.6172 - loss: 0.9119 - val_accuracy: 0.5692 - val_loss: 1.0240\n",
      "Epoch 10/12\n",
      "\u001b[1m134/134\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.6202 - loss: 0.9043 - val_accuracy: 0.5528 - val_loss: 1.0451\n",
      "Epoch 11/12\n",
      "\u001b[1m134/134\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.6211 - loss: 0.8693 - val_accuracy: 0.5621 - val_loss: 1.0536\n",
      "Epoch 12/12\n",
      "\u001b[1m134/134\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.6082 - loss: 0.9081 - val_accuracy: 0.5605 - val_loss: 1.0396\n",
      "\u001b[1m58/58\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step\n",
      "Epoch 1/12\n",
      "\u001b[1m134/134\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m2s\u001b[0m 4ms/step - accuracy: 0.3142 - loss: 1.7368 - val_accuracy: 0.4450 - val_loss: 1.2616\n",
      "Epoch 2/12\n",
      "\u001b[1m134/134\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.4914 - loss: 1.1975 - val_accuracy: 0.5185 - val_loss: 1.1484\n",
      "Epoch 3/12\n",
      "\u001b[1m134/134\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.5399 - loss: 1.0868 - val_accuracy: 0.5136 - val_loss: 1.1163\n",
      "Epoch 4/12\n",
      "\u001b[1m134/134\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.5572 - loss: 1.0516 - val_accuracy: 0.5338 - val_loss: 1.0825\n",
      "Epoch 5/12\n",
      "\u001b[1m134/134\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.5733 - loss: 1.0070 - val_accuracy: 0.5463 - val_loss: 1.0595\n",
      "Epoch 6/12\n",
      "\u001b[1m134/134\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.5984 - loss: 0.9649 - val_accuracy: 0.5534 - val_loss: 1.0634\n",
      "Epoch 7/12\n",
      "\u001b[1m134/134\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.6013 - loss: 0.9504 - val_accuracy: 0.5664 - val_loss: 1.0453\n",
      "Epoch 8/12\n",
      "\u001b[1m134/134\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 3ms/step - accuracy: 0.6107 - loss: 0.9301 - val_accuracy: 0.5588 - val_loss: 1.0360\n",
      "Epoch 9/12\n",
      "\u001b[1m134/134\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.6054 - loss: 0.9104 - val_accuracy: 0.5626 - val_loss: 1.0382\n",
      "Epoch 10/12\n",
      "\u001b[1m134/134\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.6372 - loss: 0.8668 - val_accuracy: 0.5621 - val_loss: 1.0471\n",
      "Epoch 11/12\n",
      "\u001b[1m134/134\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.6463 - loss: 0.8412 - val_accuracy: 0.5599 - val_loss: 1.0558\n",
      "Epoch 12/12\n",
      "\u001b[1m134/134\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.6348 - loss: 0.8531 - val_accuracy: 0.5566 - val_loss: 1.0635\n",
      "\u001b[1m58/58\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step\n",
      "Epoch 1/12\n",
      "\u001b[1m134/134\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m2s\u001b[0m 4ms/step - accuracy: 0.3194 - loss: 1.7235 - val_accuracy: 0.4700 - val_loss: 1.2367\n",
      "Epoch 2/12\n",
      "\u001b[1m134/134\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.4961 - loss: 1.1859 - val_accuracy: 0.5087 - val_loss: 1.1498\n",
      "Epoch 3/12\n",
      "\u001b[1m134/134\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.5374 - loss: 1.1101 - val_accuracy: 0.5044 - val_loss: 1.1165\n",
      "Epoch 4/12\n",
      "\u001b[1m134/134\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.5462 - loss: 1.0599 - val_accuracy: 0.5316 - val_loss: 1.0799\n",
      "Epoch 5/12\n",
      "\u001b[1m134/134\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.5533 - loss: 1.0249 - val_accuracy: 0.5425 - val_loss: 1.0652\n",
      "Epoch 6/12\n",
      "\u001b[1m134/134\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.5912 - loss: 0.9711 - val_accuracy: 0.5534 - val_loss: 1.0494\n",
      "Epoch 7/12\n",
      "\u001b[1m134/134\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.5998 - loss: 0.9634 - val_accuracy: 0.5392 - val_loss: 1.0483\n",
      "Epoch 8/12\n",
      "\u001b[1m134/134\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.6011 - loss: 0.9289 - val_accuracy: 0.5490 - val_loss: 1.0327\n",
      "Epoch 9/12\n",
      "\u001b[1m134/134\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.6221 - loss: 0.9038 - val_accuracy: 0.5403 - val_loss: 1.0512\n",
      "Epoch 10/12\n",
      "\u001b[1m134/134\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.6443 - loss: 0.8781 - val_accuracy: 0.5479 - val_loss: 1.0411\n",
      "Epoch 11/12\n",
      "\u001b[1m134/134\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.6282 - loss: 0.8884 - val_accuracy: 0.5436 - val_loss: 1.0375\n",
      "Epoch 12/12\n",
      "\u001b[1m134/134\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.6459 - loss: 0.8578 - val_accuracy: 0.5566 - val_loss: 1.0461\n",
      "\u001b[1m58/58\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step\n",
      "Dropping misspellings: all\n",
      "Preparing embeddings: all\n",
      "Fitting model: all\n",
      "Epoch 1/12\n",
      "\u001b[1m135/135\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m5s\u001b[0m 9ms/step - accuracy: 0.3210 - loss: 1.7273 - val_accuracy: 0.4514 - val_loss: 1.2624\n",
      "Epoch 2/12\n",
      "\u001b[1m135/135\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 5ms/step - accuracy: 0.4920 - loss: 1.1995 - val_accuracy: 0.4861 - val_loss: 1.1782\n",
      "Epoch 3/12\n",
      "\u001b[1m135/135\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 6ms/step - accuracy: 0.5197 - loss: 1.1074 - val_accuracy: 0.5372 - val_loss: 1.1114\n",
      "Epoch 4/12\n",
      "\u001b[1m135/135\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.5533 - loss: 1.0450 - val_accuracy: 0.5443 - val_loss: 1.0694\n",
      "Epoch 5/12\n",
      "\u001b[1m135/135\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 4ms/step - accuracy: 0.5718 - loss: 0.9980 - val_accuracy: 0.5557 - val_loss: 1.0617\n",
      "Epoch 6/12\n",
      "\u001b[1m135/135\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 7ms/step - accuracy: 0.5893 - loss: 0.9715 - val_accuracy: 0.5606 - val_loss: 1.0519\n",
      "Epoch 7/12\n",
      "\u001b[1m135/135\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m2s\u001b[0m 9ms/step - accuracy: 0.5965 - loss: 0.9571 - val_accuracy: 0.5671 - val_loss: 1.0432\n",
      "Epoch 8/12\n",
      "\u001b[1m135/135\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 6ms/step - accuracy: 0.6221 - loss: 0.9081 - val_accuracy: 0.5535 - val_loss: 1.0378\n",
      "Epoch 9/12\n",
      "\u001b[1m135/135\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 5ms/step - accuracy: 0.6416 - loss: 0.8878 - val_accuracy: 0.5578 - val_loss: 1.0479\n",
      "Epoch 10/12\n",
      "\u001b[1m135/135\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 7ms/step - accuracy: 0.6115 - loss: 0.8930 - val_accuracy: 0.5481 - val_loss: 1.0528\n",
      "Epoch 11/12\n",
      "\u001b[1m135/135\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m2s\u001b[0m 5ms/step - accuracy: 0.6413 - loss: 0.8593 - val_accuracy: 0.5524 - val_loss: 1.0425\n",
      "Epoch 12/12\n",
      "\u001b[1m135/135\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 4ms/step - accuracy: 0.6397 - loss: 0.8565 - val_accuracy: 0.5573 - val_loss: 1.0478\n",
      "\u001b[1m58/58\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step\n",
      "Epoch 1/12\n",
      "\u001b[1m135/135\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m2s\u001b[0m 5ms/step - accuracy: 0.3295 - loss: 1.6924 - val_accuracy: 0.4682 - val_loss: 1.2546\n",
      "Epoch 2/12\n",
      "\u001b[1m135/135\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 4ms/step - accuracy: 0.4825 - loss: 1.2201 - val_accuracy: 0.5106 - val_loss: 1.1516\n",
      "Epoch 3/12\n",
      "\u001b[1m135/135\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.5172 - loss: 1.1110 - val_accuracy: 0.5301 - val_loss: 1.0870\n",
      "Epoch 4/12\n",
      "\u001b[1m135/135\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.5555 - loss: 1.0513 - val_accuracy: 0.5437 - val_loss: 1.0582\n",
      "Epoch 5/12\n",
      "\u001b[1m135/135\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 5ms/step - accuracy: 0.5805 - loss: 1.0042 - val_accuracy: 0.5443 - val_loss: 1.0412\n",
      "Epoch 6/12\n",
      "\u001b[1m135/135\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 3ms/step - accuracy: 0.5763 - loss: 0.9938 - val_accuracy: 0.5557 - val_loss: 1.0287\n",
      "Epoch 7/12\n",
      "\u001b[1m135/135\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.5832 - loss: 0.9670 - val_accuracy: 0.5589 - val_loss: 1.0175\n",
      "Epoch 8/12\n",
      "\u001b[1m135/135\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.5918 - loss: 0.9571 - val_accuracy: 0.5562 - val_loss: 1.0190\n",
      "Epoch 9/12\n",
      "\u001b[1m135/135\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 3ms/step - accuracy: 0.6169 - loss: 0.9013 - val_accuracy: 0.5432 - val_loss: 1.0321\n",
      "Epoch 10/12\n",
      "\u001b[1m135/135\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.6187 - loss: 0.8989 - val_accuracy: 0.5492 - val_loss: 1.0128\n",
      "Epoch 11/12\n",
      "\u001b[1m135/135\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.6275 - loss: 0.8801 - val_accuracy: 0.5481 - val_loss: 1.0369\n",
      "Epoch 12/12\n",
      "\u001b[1m135/135\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.6258 - loss: 0.8740 - val_accuracy: 0.5606 - val_loss: 1.0209\n",
      "\u001b[1m58/58\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step\n",
      "Epoch 1/12\n",
      "\u001b[1m135/135\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m3s\u001b[0m 6ms/step - accuracy: 0.3397 - loss: 1.6732 - val_accuracy: 0.4677 - val_loss: 1.2288\n",
      "Epoch 2/12\n",
      "\u001b[1m135/135\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.5015 - loss: 1.1680 - val_accuracy: 0.5024 - val_loss: 1.1376\n",
      "Epoch 3/12\n",
      "\u001b[1m135/135\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.5482 - loss: 1.0755 - val_accuracy: 0.5301 - val_loss: 1.0864\n",
      "Epoch 4/12\n",
      "\u001b[1m135/135\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.5663 - loss: 1.0305 - val_accuracy: 0.5383 - val_loss: 1.0658\n",
      "Epoch 5/12\n",
      "\u001b[1m135/135\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.5619 - loss: 0.9976 - val_accuracy: 0.5421 - val_loss: 1.0477\n",
      "Epoch 6/12\n",
      "\u001b[1m135/135\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.6119 - loss: 0.9454 - val_accuracy: 0.5356 - val_loss: 1.0409\n",
      "Epoch 7/12\n",
      "\u001b[1m135/135\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.5921 - loss: 0.9649 - val_accuracy: 0.5372 - val_loss: 1.0343\n",
      "Epoch 8/12\n",
      "\u001b[1m135/135\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.6104 - loss: 0.9200 - val_accuracy: 0.5399 - val_loss: 1.0502\n",
      "Epoch 9/12\n",
      "\u001b[1m135/135\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.6012 - loss: 0.9288 - val_accuracy: 0.5399 - val_loss: 1.0429\n",
      "Epoch 10/12\n",
      "\u001b[1m135/135\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.6297 - loss: 0.9131 - val_accuracy: 0.5454 - val_loss: 1.0369\n",
      "Epoch 11/12\n",
      "\u001b[1m135/135\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.6387 - loss: 0.8703 - val_accuracy: 0.5508 - val_loss: 1.0278\n",
      "Epoch 12/12\n",
      "\u001b[1m135/135\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.6450 - loss: 0.8671 - val_accuracy: 0.5502 - val_loss: 1.0334\n",
      "\u001b[1m58/58\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step\n",
      "Epoch 1/12\n",
      "\u001b[1m135/135\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m2s\u001b[0m 5ms/step - accuracy: 0.2808 - loss: 1.7120 - val_accuracy: 0.4639 - val_loss: 1.2654\n",
      "Epoch 2/12\n",
      "\u001b[1m135/135\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.4789 - loss: 1.2390 - val_accuracy: 0.5133 - val_loss: 1.1465\n",
      "Epoch 3/12\n",
      "\u001b[1m135/135\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.5286 - loss: 1.1111 - val_accuracy: 0.5307 - val_loss: 1.0948\n",
      "Epoch 4/12\n",
      "\u001b[1m135/135\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.5584 - loss: 1.0509 - val_accuracy: 0.5334 - val_loss: 1.0632\n",
      "Epoch 5/12\n",
      "\u001b[1m135/135\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.5726 - loss: 1.0083 - val_accuracy: 0.5655 - val_loss: 1.0389\n",
      "Epoch 6/12\n",
      "\u001b[1m135/135\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.5732 - loss: 0.9938 - val_accuracy: 0.5448 - val_loss: 1.0482\n",
      "Epoch 7/12\n",
      "\u001b[1m135/135\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.6104 - loss: 0.9328 - val_accuracy: 0.5584 - val_loss: 1.0491\n",
      "Epoch 8/12\n",
      "\u001b[1m135/135\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.5980 - loss: 0.9305 - val_accuracy: 0.5573 - val_loss: 1.0397\n",
      "Epoch 9/12\n",
      "\u001b[1m135/135\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.6284 - loss: 0.8980 - val_accuracy: 0.5578 - val_loss: 1.0241\n",
      "Epoch 10/12\n",
      "\u001b[1m135/135\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 3ms/step - accuracy: 0.6196 - loss: 0.8762 - val_accuracy: 0.5578 - val_loss: 1.0363\n",
      "Epoch 11/12\n",
      "\u001b[1m135/135\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.6351 - loss: 0.8539 - val_accuracy: 0.5660 - val_loss: 1.0266\n",
      "Epoch 12/12\n",
      "\u001b[1m135/135\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.6594 - loss: 0.8422 - val_accuracy: 0.5464 - val_loss: 1.0601\n",
      "\u001b[1m58/58\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step\n"
     ]
    }
   ],
   "source": [
    "# creates an accuracy and time dataframe to store results\n",
    "accuracy_df = pd.DataFrame()\n",
    "\n",
    "# loops through each class balance\n",
    "for class_balance in class_balances:\n",
    "\n",
    "    # defines the relavant dataframe\n",
    "    texts = lists[class_balance]\n",
    "\n",
    "    # converts scores to a numpy array for use in the model\n",
    "    scores = np.array(texts.score)\n",
    "\n",
    "    # corrects misspellings\n",
    "    print('Dropping misspellings: ' + class_balance)\n",
    "    dropped_misspellings = [misspellings(x) for x in texts.full_text]\n",
    "\n",
    "    # obtains the embeddings\n",
    "    print('Preparing embeddings: ' + class_balance)\n",
    "    embeddings = encode_sentences(dropped_misspellings)\n",
    "\n",
    "    # fits the model 3 times and adds the average accuracy to the accuracy_df\n",
    "    print('Fitting model: ' + class_balance)\n",
    "    accuracy_df = fit_model(embeddings, scores, accuracy_df, class_balance)\n",
    "\n",
    "    # reassigns the embeddings object to nother to conserve memory\n",
    "    embeddings = ''"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7789901b-f06a-41fa-919b-a26abfa5dcab",
   "metadata": {},
   "source": [
    "## Visualization"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "34b68593-2dd6-45ab-8f8b-98193a79119e",
   "metadata": {},
   "source": [
    "The results seem to show that including more balanced classes generally results in greater accuracy."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "02bcc28b-e04b-469d-89c0-4acd59d397a4",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "\n",
    "# transposes the dataframe to get it in the right shape for plotting\n",
    "accuracy_df = accuracy_df.transpose()\n",
    "\n",
    "# plots the results as a lineplot in searborn\n",
    "fig = plt.figure(figsize=(10, 6));\n",
    "sns.lineplot(x=accuracy_df.index, y=accuracy_df.iloc[:,0], data=accuracy_df);\n",
    "plt.title('Class Imbalances', size=30);\n",
    "plt.ylabel('Cohen Kappa Score');\n",
    "plt.xlabel('Class imbalances');"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.11.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}