ag_news_category_prediction.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "id": "initial_id",
   "metadata": {
    "collapsed": true,
    "ExecuteTime": {
     "end_time": "2024-05-15T17:12:47.998929Z",
     "start_time": "2024-05-15T17:12:45.167863Z"
    }
   },
   "source": [
    "import os\n",
    "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2'\n",
    "import numpy as np\n",
    "# Load the ag_news dataset from tensorflow_datasets\n",
    "import tensorflow_datasets as tfds\n",
    "import tensorflow as tf\n",
    "\n",
    "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2'\n",
    "\n",
    "# Load the ag_news dataset\n",
    "dataset, info = tfds.load('ag_news_subset', with_info=True, as_supervised=False)\n",
    "train_dataset, test_dataset = dataset['train'], dataset['test']\n"
   ],
   "outputs": [],
   "execution_count": 1
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-05-15T17:12:48.002501Z",
     "start_time": "2024-05-15T17:12:47.999995Z"
    }
   },
   "cell_type": "code",
   "source": [
    "# Display the dataset info\n",
    "print(info)\n"
   ],
   "id": "e1ad77d1a998f960",
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tfds.core.DatasetInfo(\n",
      "    name='ag_news_subset',\n",
      "    full_name='ag_news_subset/1.0.0',\n",
      "    description=\"\"\"\n",
      "    AG is a collection of more than 1 million news articles. News articles have been\n",
      "    gathered from more than 2000 news sources by ComeToMyHead in more than 1 year of\n",
      "    activity. ComeToMyHead is an academic news search engine which has been running\n",
      "    since July, 2004. The dataset is provided by the academic comunity for research\n",
      "    purposes in data mining (clustering, classification, etc), information retrieval\n",
      "    (ranking, search, etc), xml, data compression, data streaming, and any other\n",
      "    non-commercial activity. For more information, please refer to the link\n",
      "    http://www.di.unipi.it/~gulli/AG_corpus_of_news_articles.html .\n",
      "    \n",
      "    The AG's news topic classification dataset is constructed by Xiang Zhang\n",
      "    (xiang.zhang@nyu.edu) from the dataset above. It is used as a text\n",
      "    classification benchmark in the following paper: Xiang Zhang, Junbo Zhao, Yann\n",
      "    LeCun. Character-level Convolutional Networks for Text Classification. Advances\n",
      "    in Neural Information Processing Systems 28 (NIPS 2015).\n",
      "    \n",
      "    The AG's news topic classification dataset is constructed by choosing 4 largest\n",
      "    classes from the original corpus. Each class contains 30,000 training samples\n",
      "    and 1,900 testing samples. The total number of training samples is 120,000 and\n",
      "    testing 7,600.\n",
      "    \"\"\",\n",
      "    homepage='https://arxiv.org/abs/1509.01626',\n",
      "    data_path='/Users/acalma/tensorflow_datasets/ag_news_subset/1.0.0',\n",
      "    file_format=tfrecord,\n",
      "    download_size=11.24 MiB,\n",
      "    dataset_size=35.79 MiB,\n",
      "    features=FeaturesDict({\n",
      "        'description': Text(shape=(), dtype=string),\n",
      "        'label': ClassLabel(shape=(), dtype=int64, num_classes=4),\n",
      "        'title': Text(shape=(), dtype=string),\n",
      "    }),\n",
      "    supervised_keys=('description', 'label'),\n",
      "    disable_shuffling=False,\n",
      "    splits={\n",
      "        'test': <SplitInfo num_examples=7600, num_shards=1>,\n",
      "        'train': <SplitInfo num_examples=120000, num_shards=1>,\n",
      "    },\n",
      "    citation=\"\"\"@misc{zhang2015characterlevel,\n",
      "        title={Character-level Convolutional Networks for Text Classification},\n",
      "        author={Xiang Zhang and Junbo Zhao and Yann LeCun},\n",
      "        year={2015},\n",
      "        eprint={1509.01626},\n",
      "        archivePrefix={arXiv},\n",
      "        primaryClass={cs.LG}\n",
      "    }\"\"\",\n",
      ")\n"
     ]
    }
   ],
   "execution_count": 2
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-05-15T17:12:48.813324Z",
     "start_time": "2024-05-15T17:12:48.808455Z"
    }
   },
   "cell_type": "code",
   "source": [
    "import numpy as np\n",
    "def split_features_labels(dataset):\n",
    "    # Extract the features and labels\n",
    "    features = []\n",
    "    labels = []\n",
    "    for example in dataset:\n",
    "        features.append(example['title'].numpy())\n",
    "        labels.append(example['label'].numpy())\n",
    "    features = np.array([x.decode('utf-8') for x in features])\n",
    "    labels = np.array([float(x) for x in labels])\n",
    "    return features, labels"
   ],
   "id": "6dfc648f69ba510b",
   "outputs": [],
   "execution_count": 3
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-05-15T17:12:56.104827Z",
     "start_time": "2024-05-15T17:12:49.708989Z"
    }
   },
   "cell_type": "code",
   "source": [
    "train_titles, train_labels = split_features_labels(train_dataset)\n",
    "test_titles, test_labels = split_features_labels(test_dataset)"
   ],
   "id": "203da1a9c0c8ef17",
   "outputs": [],
   "execution_count": 4
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-05-15T17:12:56.156763Z",
     "start_time": "2024-05-15T17:12:56.105855Z"
    }
   },
   "cell_type": "code",
   "source": [
    "title_lengths = [len(feature) for feature in train_titles]\n",
    "min(title_lengths), max(title_lengths)"
   ],
   "id": "8f9a4d8e9a82522f",
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(6, 115)"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "execution_count": 5
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-05-15T17:12:57.645377Z",
     "start_time": "2024-05-15T17:12:57.001479Z"
    }
   },
   "cell_type": "code",
   "source": [
    "# plot the distribution of the title lengths\n",
    "import matplotlib.pyplot as plt\n",
    "plt.hist(title_lengths, bins=10)"
   ],
   "id": "7fca1f4aa1daf585",
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(array([1.6310e+03, 1.2089e+04, 3.5483e+04, 4.2403e+04, 1.8702e+04,\n",
       "        5.9000e+03, 1.1830e+03, 2.4420e+03, 1.5800e+02, 9.0000e+00]),\n",
       " array([  6. ,  16.9,  27.8,  38.7,  49.6,  60.5,  71.4,  82.3,  93.2,\n",
       "        104.1, 115. ]),\n",
       " <BarContainer object of 10 artists>)"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ],
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAkQAAAGdCAYAAADzOWwgAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjUuMywgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/NK7nSAAAACXBIWXMAAA9hAAAPYQGoP6dpAAAzjElEQVR4nO3df1RU953/8Rc/ZPDXDFEDyBeMNLZRKmoExWnSbKzUSULa2JhdNa4hhiSri26AVpHWYpK2izUnjaYaaTa7IXtWGrUn2gYqlmDETSX+wFB/JNIkJcUUB2gMjBIFhfv9Yw+3TiWJKAbx83ycc0+c+3nfO+/7OafldS73fgiwLMsSAACAwQJ7uwEAAIDeRiACAADGIxABAADjEYgAAIDxCEQAAMB4BCIAAGA8AhEAADAegQgAABgvuLcb6E0dHR2qq6vT4MGDFRAQ0NvtAACAi2BZlk6ePKmoqCgFBvbMvR2jA1FdXZ1iYmJ6uw0AAHAJjh07pujo6B45l9GBaPDgwZL+b0KdTmcvdwMAAC6Gz+dTTEyM/XO8JxgdiDp/TeZ0OglEAAD0MT35uAsPVQMAAOMRiAAAgPEIRAAAwHgEIgAAYDwCEQAAMB6BCAAAGI9ABAAAjEcgAgAAxiMQAQAA4xGIAACA8QhEAADAeAQiAABgPAIRAAAwHoEIAAAYL7i3GwBMM3JZcW+3cEk+WJnS2y0AwBXDHSIAAGA8AhEAADAegQgAABiPQAQAAIxHIAIAAMYjEAEAAOMRiAAAgPEIRAAAwHgEIgAAYDwCEQAAMB6BCAAAGI9ABAAAjEcgAgAAxiMQAQAA4xGIAACA8S4rEK1cuVIBAQHKyMiw9505c0bp6ekaOnSoBg0apJkzZ6q+vt7vuNraWqWkpGjAgAEKDw/XkiVLdO7cOb+anTt3auLEiXI4HBo1apQKCgou+P5169Zp5MiRCg0NVVJSkvbu3Xs5lwMAAAx1yYFo3759+sUvfqFx48b57c/MzNSrr76qzZs3q7y8XHV1dbr33nvt8fb2dqWkpKitrU27d+/WSy+9pIKCAuXm5to1NTU1SklJ0dSpU1VVVaWMjAw9/PDD2r59u12zceNGZWVlacWKFTpw4IDGjx8vj8ejhoaGS70kAABgqADLsqzuHnTq1ClNnDhRzz33nH784x9rwoQJWr16tZqbm3X99dersLBQ9913nyTp6NGjGjNmjCoqKjRlyhRt27ZNd999t+rq6hQRESFJys/PV3Z2thobGxUSEqLs7GwVFxfr8OHD9nfOnj1bTU1NKikpkSQlJSVp0qRJWrt2rSSpo6NDMTExWrx4sZYtW3ZR1+Hz+eRyudTc3Cyn09ndaQAuychlxb3dwiX5YGVKb7cAAJKuzM/vS7pDlJ6erpSUFCUnJ/vtr6ys1NmzZ/32jx49WiNGjFBFRYUkqaKiQvHx8XYYkiSPxyOfz6cjR47YNX9/bo/HY5+jra1NlZWVfjWBgYFKTk62a7rS2toqn8/ntwEAAAR394CXX35ZBw4c0L59+y4Y83q9CgkJUVhYmN/+iIgIeb1eu+b8MNQ53jn2WTU+n0+nT5/Wxx9/rPb29i5rjh49+qm95+Xl6Yknnri4CwUAAMbo1h2iY8eO6bHHHtOGDRsUGhp6pXq6YnJyctTc3Gxvx44d6+2WAADAVaBbgaiyslINDQ2aOHGigoODFRwcrPLycj377LMKDg5WRESE2tra1NTU5HdcfX29IiMjJUmRkZEXvHXW+fnzapxOp/r3769hw4YpKCioy5rOc3TF4XDI6XT6bQAAAN0KRNOmTdOhQ4dUVVVlb4mJiZo7d6797379+qmsrMw+prq6WrW1tXK73ZIkt9utQ4cO+b0NVlpaKqfTqbi4OLvm/HN01nSeIyQkRAkJCX41HR0dKisrs2sAAAAuVreeIRo8eLDGjh3rt2/gwIEaOnSovT8tLU1ZWVkaMmSInE6nFi9eLLfbrSlTpkiSpk+frri4OM2bN0+rVq2S1+vV8uXLlZ6eLofDIUlasGCB1q5dq6VLl+qhhx7Sjh07tGnTJhUX/+3tnKysLKWmpioxMVGTJ0/W6tWr1dLSovnz51/WhAAAAPN0+6Hqz/PMM88oMDBQM2fOVGtrqzwej5577jl7PCgoSEVFRVq4cKHcbrcGDhyo1NRUPfnkk3ZNbGysiouLlZmZqTVr1ig6OlovvPCCPB6PXTNr1iw1NjYqNzdXXq9XEyZMUElJyQUPWgMAAHyeS1qH6FrBOkToDaxDBACX56pZhwgAAOBaQiACAADGIxABAADjEYgAAIDxCEQAAMB4BCIAAGA8AhEAADAegQgAABiPQAQAAIxHIAIAAMYjEAEAAOMRiAAAgPEIRAAAwHgEIgAAYDwCEQAAMB6BCAAAGI9ABAAAjEcgAgAAxiMQAQAA4xGIAACA8QhEAADAeAQiAABgPAIRAAAwHoEIAAAYj0AEAACMRyACAADGIxABAADjEYgAAIDxCEQAAMB4BCIAAGA8AhEAADBecHeK169fr/Xr1+uDDz6QJH31q19Vbm6u7rzzTknS7bffrvLycr9j/uVf/kX5+fn259raWi1cuFCvv/66Bg0apNTUVOXl5Sk4+G+t7Ny5U1lZWTpy5IhiYmK0fPlyPfjgg37nXbdunZ566il5vV6NHz9eP//5zzV58uTuXA6uASOXFfd2CwCAa0C37hBFR0dr5cqVqqys1P79+/WNb3xD99xzj44cOWLXPPLIIzp+/Li9rVq1yh5rb29XSkqK2tratHv3br300ksqKChQbm6uXVNTU6OUlBRNnTpVVVVVysjI0MMPP6zt27fbNRs3blRWVpZWrFihAwcOaPz48fJ4PGpoaLicuQAAAIYKsCzLupwTDBkyRE899ZTS0tJ0++23a8KECVq9enWXtdu2bdPdd9+turo6RURESJLy8/OVnZ2txsZGhYSEKDs7W8XFxTp8+LB93OzZs9XU1KSSkhJJUlJSkiZNmqS1a9dKkjo6OhQTE6PFixdr2bJlF927z+eTy+VSc3OznE7nJc4AehN3iL44H6xM6e0WAEDSlfn5fcnPELW3t+vll19WS0uL3G63vX/Dhg0aNmyYxo4dq5ycHH3yySf2WEVFheLj4+0wJEkej0c+n8++y1RRUaHk5GS/7/J4PKqoqJAktbW1qbKy0q8mMDBQycnJds2naW1tlc/n89sAAAC69QyRJB06dEhut1tnzpzRoEGDtGXLFsXFxUmS7r//ft1www2KiorSwYMHlZ2drerqar3yyiuSJK/X6xeGJNmfvV7vZ9b4fD6dPn1aH3/8sdrb27usOXr06Gf2npeXpyeeeKK7lwwAAK5x3Q5EN910k6qqqtTc3Kxf/epXSk1NVXl5ueLi4vToo4/adfHx8Ro+fLimTZum999/XzfeeGOPNn4pcnJylJWVZX/2+XyKiYnpxY4AAMDVoNuBKCQkRKNGjZIkJSQkaN++fVqzZo1+8YtfXFCblJQkSXrvvfd04403KjIyUnv37vWrqa+vlyRFRkba/+3cd36N0+lU//79FRQUpKCgoC5rOs/xaRwOhxwORzeuFgAAmOCy1yHq6OhQa2trl2NVVVWSpOHDh0uS3G63Dh065Pc2WGlpqZxOp/1rN7fbrbKyMr/zlJaW2s8phYSEKCEhwa+mo6NDZWVlfs8yAQAAXKxu3SHKycnRnXfeqREjRujkyZMqLCzUzp07tX37dr3//vsqLCzUXXfdpaFDh+rgwYPKzMzUbbfdpnHjxkmSpk+frri4OM2bN0+rVq2S1+vV8uXLlZ6ebt+5WbBggdauXaulS5fqoYce0o4dO7Rp0yYVF//tbaKsrCylpqYqMTFRkydP1urVq9XS0qL58+f34NQAAABTdCsQNTQ06IEHHtDx48flcrk0btw4bd++Xd/85jd17Ngxvfbaa3Y4iYmJ0cyZM7V8+XL7+KCgIBUVFWnhwoVyu90aOHCgUlNT9eSTT9o1sbGxKi4uVmZmptasWaPo6Gi98MIL8ng8ds2sWbPU2Nio3Nxceb1eTZgwQSUlJRc8aA0AAHAxLnsdor6MdYj6PtYh+uKwDhGAq8VVtQ4RAADAtYJABAAAjEcgAgAAxiMQAQAA4xGIAACA8QhEAADAeAQiAABgPAIRAAAwHoEIAAAYj0AEAACMRyACAADGIxABAADjEYgAAIDxCEQAAMB4BCIAAGA8AhEAADAegQgAABiPQAQAAIxHIAIAAMYjEAEAAOMRiAAAgPEIRAAAwHgEIgAAYDwCEQAAMB6BCAAAGI9ABAAAjEcgAgAAxiMQAQAA4xGIAACA8QhEAADAeN0KROvXr9e4cePkdDrldDrldru1bds2e/zMmTNKT0/X0KFDNWjQIM2cOVP19fV+56itrVVKSooGDBig8PBwLVmyROfOnfOr2blzpyZOnCiHw6FRo0apoKDggl7WrVunkSNHKjQ0VElJSdq7d293LgUAAMDWrUAUHR2tlStXqrKyUvv379c3vvEN3XPPPTpy5IgkKTMzU6+++qo2b96s8vJy1dXV6d5777WPb29vV0pKitra2rR792699NJLKigoUG5url1TU1OjlJQUTZ06VVVVVcrIyNDDDz+s7du32zUbN25UVlaWVqxYoQMHDmj8+PHyeDxqaGi43PkAAAAGCrAsy7qcEwwZMkRPPfWU7rvvPl1//fUqLCzUfffdJ0k6evSoxowZo4qKCk2ZMkXbtm3T3Xffrbq6OkVEREiS8vPzlZ2drcbGRoWEhCg7O1vFxcU6fPiw/R2zZ89WU1OTSkpKJElJSUmaNGmS1q5dK0nq6OhQTEyMFi9erGXLll107z6fTy6XS83NzXI6nZczDeglI5cV93YLxvhgZUpvtwAAkq7Mz+9Lfoaovb1dL7/8slpaWuR2u1VZWamzZ88qOTnZrhk9erRGjBihiooKSVJFRYXi4+PtMCRJHo9HPp/PvstUUVHhd47Oms5ztLW1qbKy0q8mMDBQycnJds2naW1tlc/n89sAAAC6HYgOHTqkQYMGyeFwaMGCBdqyZYvi4uLk9XoVEhKisLAwv/qIiAh5vV5Jktfr9QtDneOdY59V4/P5dPr0af31r39Ve3t7lzWd5/g0eXl5crlc9hYTE9PdywcAANegbgeim266SVVVVdqzZ48WLlyo1NRUvf3221eitx6Xk5Oj5uZmezt27FhvtwQAAK4Cwd09ICQkRKNGjZIkJSQkaN++fVqzZo1mzZqltrY2NTU1+d0lqq+vV2RkpCQpMjLygrfBOt9CO7/m799Mq6+vl9PpVP/+/RUUFKSgoKAuazrP8WkcDoccDkd3LxkAAFzjLnsdoo6ODrW2tiohIUH9+vVTWVmZPVZdXa3a2lq53W5Jktvt1qFDh/zeBistLZXT6VRcXJxdc/45Oms6zxESEqKEhAS/mo6ODpWVldk1AAAA3dGtO0Q5OTm68847NWLECJ08eVKFhYXauXOntm/fLpfLpbS0NGVlZWnIkCFyOp1avHix3G63pkyZIkmaPn264uLiNG/ePK1atUper1fLly9Xenq6fedmwYIFWrt2rZYuXaqHHnpIO3bs0KZNm1Rc/Le3ibKyspSamqrExERNnjxZq1evVktLi+bPn9+DUwMAAEzRrUDU0NCgBx54QMePH5fL5dK4ceO0fft2ffOb35QkPfPMMwoMDNTMmTPV2toqj8ej5557zj4+KChIRUVFWrhwodxutwYOHKjU1FQ9+eSTdk1sbKyKi4uVmZmpNWvWKDo6Wi+88II8Ho9dM2vWLDU2Nio3N1der1cTJkxQSUnJBQ9aAwAAXIzLXoeoL2Mdor6PdYi+OKxDBOBqcVWtQwQAAHCtIBABAADjEYgAAIDxCEQAAMB4BCIAAGA8AhEAADAegQgAABiPQAQAAIxHIAIAAMYjEAEAAOMRiAAAgPEIRAAAwHgEIgAAYDwCEQAAMB6BCAAAGI9ABAAAjEcgAgAAxiMQAQAA4xGIAACA8QhEAADAeAQiAABgPAIRAAAwHoEIAAAYj0AEAACMRyACAADGIxABAADjEYgAAIDxCEQAAMB4BCIAAGA8AhEAADAegQgAABivW4EoLy9PkyZN0uDBgxUeHq4ZM2aourrar+b2229XQECA37ZgwQK/mtraWqWkpGjAgAEKDw/XkiVLdO7cOb+anTt3auLEiXI4HBo1apQKCgou6GfdunUaOXKkQkNDlZSUpL1793bncgAAACR1MxCVl5crPT1db775pkpLS3X27FlNnz5dLS0tfnWPPPKIjh8/bm+rVq2yx9rb25WSkqK2tjbt3r1bL730kgoKCpSbm2vX1NTUKCUlRVOnTlVVVZUyMjL08MMPa/v27XbNxo0blZWVpRUrVujAgQMaP368PB6PGhoaLnUuAACAoQIsy7Iu9eDGxkaFh4ervLxct912m6T/u0M0YcIErV69ustjtm3bprvvvlt1dXWKiIiQJOXn5ys7O1uNjY0KCQlRdna2iouLdfjwYfu42bNnq6mpSSUlJZKkpKQkTZo0SWvXrpUkdXR0KCYmRosXL9ayZcsuqn+fzyeXy6Xm5mY5nc5LnQb0opHLinu7BWN8sDKlt1sAAElX5uf3ZT1D1NzcLEkaMmSI3/4NGzZo2LBhGjt2rHJycvTJJ5/YYxUVFYqPj7fDkCR5PB75fD4dOXLErklOTvY7p8fjUUVFhSSpra1NlZWVfjWBgYFKTk62a7rS2toqn8/ntwEAAARf6oEdHR3KyMjQLbfcorFjx9r777//ft1www2KiorSwYMHlZ2drerqar3yyiuSJK/X6xeGJNmfvV7vZ9b4fD6dPn1aH3/8sdrb27usOXr06Kf2nJeXpyeeeOJSLxkAAFyjLjkQpaen6/Dhw3rjjTf89j/66KP2v+Pj4zV8+HBNmzZN77//vm688cZL77QH5OTkKCsry/7s8/kUExPTix0BAICrwSUFokWLFqmoqEi7du1SdHT0Z9YmJSVJkt577z3deOONioyMvOBtsPr6eklSZGSk/d/OfefXOJ1O9e/fX0FBQQoKCuqypvMcXXE4HHI4HBd3kQAAwBjdeobIsiwtWrRIW7Zs0Y4dOxQbG/u5x1RVVUmShg8fLklyu906dOiQ39tgpaWlcjqdiouLs2vKysr8zlNaWiq32y1JCgkJUUJCgl9NR0eHysrK7BoAAICL1a07ROnp6SosLNSvf/1rDR482H7mx+VyqX///nr//fdVWFiou+66S0OHDtXBgweVmZmp2267TePGjZMkTZ8+XXFxcZo3b55WrVolr9er5cuXKz093b57s2DBAq1du1ZLly7VQw89pB07dmjTpk0qLv7bG0VZWVlKTU1VYmKiJk+erNWrV6ulpUXz58/vqbkBAACG6FYgWr9+vaT/e7X+fC+++KIefPBBhYSE6LXXXrPDSUxMjGbOnKnly5fbtUFBQSoqKtLChQvldrs1cOBApaam6sknn7RrYmNjVVxcrMzMTK1Zs0bR0dF64YUX5PF47JpZs2apsbFRubm58nq9mjBhgkpKSi540BoAAODzXNY6RH0d6xD1faxD9MVhHSIAV4urbh0iAACAawGBCAAAGI9ABAAAjEcgAgAAxiMQAQAA4xGIAACA8QhEAADAeAQiAABgPAIRAAAwHoEIAAAYj0AEAACMRyACAADGIxABAADjEYgAAIDxCEQAAMB4BCIAAGA8AhEAADAegQgAABiPQAQAAIxHIAIAAMYjEAEAAOMRiAAAgPEIRAAAwHgEIgAAYDwCEQAAMB6BCAAAGI9ABAAAjEcgAgAAxiMQAQAA4xGIAACA8boViPLy8jRp0iQNHjxY4eHhmjFjhqqrq/1qzpw5o/T0dA0dOlSDBg3SzJkzVV9f71dTW1urlJQUDRgwQOHh4VqyZInOnTvnV7Nz505NnDhRDodDo0aNUkFBwQX9rFu3TiNHjlRoaKiSkpK0d+/e7lwOAACApG4GovLycqWnp+vNN99UaWmpzp49q+nTp6ulpcWuyczM1KuvvqrNmzervLxcdXV1uvfee+3x9vZ2paSkqK2tTbt379ZLL72kgoIC5ebm2jU1NTVKSUnR1KlTVVVVpYyMDD388MPavn27XbNx40ZlZWVpxYoVOnDggMaPHy+Px6OGhobLmQ8AAGCgAMuyrEs9uLGxUeHh4SovL9dtt92m5uZmXX/99SosLNR9990nSTp69KjGjBmjiooKTZkyRdu2bdPdd9+turo6RURESJLy8/OVnZ2txsZGhYSEKDs7W8XFxTp8+LD9XbNnz1ZTU5NKSkokSUlJSZo0aZLWrl0rSero6FBMTIwWL16sZcuWXVT/Pp9PLpdLzc3NcjqdlzoN6EUjlxX3dgvG+GBlSm+3AACSrszP78t6hqi5uVmSNGTIEElSZWWlzp49q+TkZLtm9OjRGjFihCoqKiRJFRUVio+Pt8OQJHk8Hvl8Ph05csSuOf8cnTWd52hra1NlZaVfTWBgoJKTk+2arrS2tsrn8/ltAAAAlxyIOjo6lJGRoVtuuUVjx46VJHm9XoWEhCgsLMyvNiIiQl6v1645Pwx1jneOfVaNz+fT6dOn9de//lXt7e1d1nSeoyt5eXlyuVz2FhMT0/0LBwAA15xLDkTp6ek6fPiwXn755Z7s54rKyclRc3OzvR07dqy3WwIAAFeB4Es5aNGiRSoqKtKuXbsUHR1t74+MjFRbW5uampr87hLV19crMjLSrvn7t8E630I7v+bv30yrr6+X0+lU//79FRQUpKCgoC5rOs/RFYfDIYfD0f0LBgAA17Ru3SGyLEuLFi3Sli1btGPHDsXGxvqNJyQkqF+/fiorK7P3VVdXq7a2Vm63W5Lkdrt16NAhv7fBSktL5XQ6FRcXZ9ecf47Oms5zhISEKCEhwa+mo6NDZWVldg0AAMDF6tYdovT0dBUWFurXv/61Bg8ebD+v43K51L9/f7lcLqWlpSkrK0tDhgyR0+nU4sWL5Xa7NWXKFEnS9OnTFRcXp3nz5mnVqlXyer1avny50tPT7bs3CxYs0Nq1a7V06VI99NBD2rFjhzZt2qTi4r+9UZSVlaXU1FQlJiZq8uTJWr16tVpaWjR//vyemhsAAGCIbgWi9evXS5Juv/12v/0vvviiHnzwQUnSM888o8DAQM2cOVOtra3yeDx67rnn7NqgoCAVFRVp4cKFcrvdGjhwoFJTU/Xkk0/aNbGxsSouLlZmZqbWrFmj6OhovfDCC/J4PHbNrFmz1NjYqNzcXHm9Xk2YMEElJSUXPGgNAADweS5rHaK+jnWI+j7WIfrisA4RgKvFVbcOEQAAwLWAQAQAAIxHIAIAAMYjEAEAAONd0sKMAMzTFx9g50FwABeLO0QAAMB4BCIAAGA8AhEAADAegQgAABiPQAQAAIxHIAIAAMYjEAEAAOMRiAAAgPEIRAAAwHgEIgAAYDwCEQAAMB6BCAAAGI9ABAAAjEcgAgAAxiMQAQAA4xGIAACA8QhEAADAeAQiAABgPAIRAAAwHoEIAAAYj0AEAACMRyACAADGIxABAADjEYgAAIDxCEQAAMB4BCIAAGC8bgeiXbt26Vvf+paioqIUEBCgrVu3+o0/+OCDCggI8NvuuOMOv5oTJ05o7ty5cjqdCgsLU1pamk6dOuVXc/DgQX39619XaGioYmJitGrVqgt62bx5s0aPHq3Q0FDFx8frt7/9bXcvBwAAoPuBqKWlRePHj9e6des+teaOO+7Q8ePH7e2Xv/yl3/jcuXN15MgRlZaWqqioSLt27dKjjz5qj/t8Pk2fPl033HCDKisr9dRTT+nxxx/X888/b9fs3r1bc+bMUVpamt566y3NmDFDM2bM0OHDh7t7SQAAwHABlmVZl3xwQIC2bNmiGTNm2PsefPBBNTU1XXDnqNM777yjuLg47du3T4mJiZKkkpIS3XXXXfrwww8VFRWl9evX6wc/+IG8Xq9CQkIkScuWLdPWrVt19OhRSdKsWbPU0tKioqIi+9xTpkzRhAkTlJ+ff1H9+3w+uVwuNTc3y+l0XsIMoLeNXFbc2y3gKvbBypTebgHAFXAlfn5fkWeIdu7cqfDwcN10001auHChPvroI3usoqJCYWFhdhiSpOTkZAUGBmrPnj12zW233WaHIUnyeDyqrq7Wxx9/bNckJyf7fa/H41FFRcWn9tXa2iqfz+e3AQAA9HgguuOOO/Tf//3fKisr009/+lOVl5frzjvvVHt7uyTJ6/UqPDzc75jg4GANGTJEXq/XromIiPCr6fz8eTWd413Jy8uTy+Wyt5iYmMu7WAAAcE0I7ukTzp492/53fHy8xo0bpxtvvFE7d+7UtGnTevrruiUnJ0dZWVn2Z5/PRygCAABX/rX7L33pSxo2bJjee+89SVJkZKQaGhr8as6dO6cTJ04oMjLSrqmvr/er6fz8eTWd411xOBxyOp1+GwAAwBUPRB9++KE++ugjDR8+XJLkdrvV1NSkyspKu2bHjh3q6OhQUlKSXbNr1y6dPXvWriktLdVNN92k6667zq4pKyvz+67S0lK53e4rfUkAAOAa0+1AdOrUKVVVVamqqkqSVFNTo6qqKtXW1urUqVNasmSJ3nzzTX3wwQcqKyvTPffco1GjRsnj8UiSxowZozvuuEOPPPKI9u7dq9///vdatGiRZs+eraioKEnS/fffr5CQEKWlpenIkSPauHGj1qxZ4/frrscee0wlJSV6+umndfToUT3++OPav3+/Fi1a1APTAgAATNLtQLR//37dfPPNuvnmmyVJWVlZuvnmm5Wbm6ugoCAdPHhQ3/72t/WVr3xFaWlpSkhI0P/+7//K4XDY59iwYYNGjx6tadOm6a677tKtt97qt8aQy+XS7373O9XU1CghIUHf/e53lZub67dW0de+9jUVFhbq+eef1/jx4/WrX/1KW7du1dixYy9nPgAAgIEuax2ivo51iPo+1iHCZ2EdIuDa1GfWIQIAAOhLCEQAAMB4BCIAAGA8AhEAADAegQgAABiPQAQAAIxHIAIAAMYjEAEAAOMRiAAAgPEIRAAAwHgEIgAAYDwCEQAAMB6BCAAAGI9ABAAAjEcgAgAAxiMQAQAA4xGIAACA8YJ7uwFcPUYuK+7tFgAA6BXcIQIAAMYjEAEAAOMRiAAAgPEIRAAAwHgEIgAAYDwCEQAAMB6BCAAAGI9ABAAAjEcgAgAAxiMQAQAA4xGIAACA8QhEAADAeN0ORLt27dK3vvUtRUVFKSAgQFu3bvUbtyxLubm5Gj58uPr376/k5GS9++67fjUnTpzQ3Llz5XQ6FRYWprS0NJ06dcqv5uDBg/r617+u0NBQxcTEaNWqVRf0snnzZo0ePVqhoaGKj4/Xb3/72+5eDgAAQPcDUUtLi8aPH69169Z1Ob5q1So9++yzys/P1549ezRw4EB5PB6dOXPGrpk7d66OHDmi0tJSFRUVadeuXXr00UftcZ/Pp+nTp+uGG25QZWWlnnrqKT3++ON6/vnn7Zrdu3drzpw5SktL01tvvaUZM2ZoxowZOnz4cHcvCQAAGC7Asizrkg8OCNCWLVs0Y8YMSf93dygqKkrf/e539b3vfU+S1NzcrIiICBUUFGj27Nl65513FBcXp3379ikxMVGSVFJSorvuuksffvihoqKitH79ev3gBz+Q1+tVSEiIJGnZsmXaunWrjh49KkmaNWuWWlpaVFRUZPczZcoUTZgwQfn5+RfVv8/nk8vlUnNzs5xO56VOwzVj5LLi3m4B6FEfrEzp7RYAXAFX4ud3jz5DVFNTI6/Xq+TkZHufy+VSUlKSKioqJEkVFRUKCwuzw5AkJScnKzAwUHv27LFrbrvtNjsMSZLH41F1dbU+/vhju+b87+ms6fyerrS2tsrn8/ltAAAAPRqIvF6vJCkiIsJvf0REhD3m9XoVHh7uNx4cHKwhQ4b41XR1jvO/49NqOse7kpeXJ5fLZW8xMTHdvUQAAHANMuots5ycHDU3N9vbsWPHerslAABwFejRQBQZGSlJqq+v99tfX19vj0VGRqqhocFv/Ny5czpx4oRfTVfnOP87Pq2mc7wrDodDTqfTbwMAAOjRQBQbG6vIyEiVlZXZ+3w+n/bs2SO32y1JcrvdampqUmVlpV2zY8cOdXR0KCkpya7ZtWuXzp49a9eUlpbqpptu0nXXXWfXnP89nTWd3wMAAHCxuh2ITp06paqqKlVVVUn6vwepq6qqVFtbq4CAAGVkZOjHP/6xfvOb3+jQoUN64IEHFBUVZb+JNmbMGN1xxx165JFHtHfvXv3+97/XokWLNHv2bEVFRUmS7r//foWEhCgtLU1HjhzRxo0btWbNGmVlZdl9PPbYYyopKdHTTz+to0eP6vHHH9f+/fu1aNGiy58VAABglODuHrB//35NnTrV/twZUlJTU1VQUKClS5eqpaVFjz76qJqamnTrrbeqpKREoaGh9jEbNmzQokWLNG3aNAUGBmrmzJl69tln7XGXy6Xf/e53Sk9PV0JCgoYNG6bc3Fy/tYq+9rWvqbCwUMuXL9f3v/99ffnLX9bWrVs1duzYS5oIAABgrstah6ivYx0if6xDhGsN6xAB16arfh0iAACAvohABAAAjEcgAgAAxiMQAQAA4xGIAACA8QhEAADAeAQiAABgPAIRAAAwHoEIAAAYj0AEAACMRyACAADGIxABAADjEYgAAIDxCEQAAMB4BCIAAGA8AhEAADAegQgAABiPQAQAAIxHIAIAAMYjEAEAAOMRiAAAgPEIRAAAwHgEIgAAYDwCEQAAMB6BCAAAGI9ABAAAjEcgAgAAxgvu7QYA4EoZuay4t1votg9WpvR2C4CRuEMEAACMRyACAADGIxABAADj9XggevzxxxUQEOC3jR492h4/c+aM0tPTNXToUA0aNEgzZ85UfX293zlqa2uVkpKiAQMGKDw8XEuWLNG5c+f8anbu3KmJEyfK4XBo1KhRKigo6OlLAQAAhrgid4i++tWv6vjx4/b2xhtv2GOZmZl69dVXtXnzZpWXl6uurk733nuvPd7e3q6UlBS1tbVp9+7deumll1RQUKDc3Fy7pqamRikpKZo6daqqqqqUkZGhhx9+WNu3b78SlwMAAK5xV+Qts+DgYEVGRl6wv7m5Wf/5n/+pwsJCfeMb35AkvfjiixozZozefPNNTZkyRb/73e/09ttv67XXXlNERIQmTJigH/3oR8rOztbjjz+ukJAQ5efnKzY2Vk8//bQkacyYMXrjjTf0zDPPyOPxXIlLAgAA17Arcofo3XffVVRUlL70pS9p7ty5qq2tlSRVVlbq7NmzSk5OtmtHjx6tESNGqKKiQpJUUVGh+Ph4RURE2DUej0c+n09Hjhyxa84/R2dN5zk+TWtrq3w+n98GAADQ44EoKSlJBQUFKikp0fr161VTU6Ovf/3rOnnypLxer0JCQhQWFuZ3TEREhLxeryTJ6/X6haHO8c6xz6rx+Xw6ffr0p/aWl5cnl8tlbzExMZd7uQAA4BrQ478yu/POO+1/jxs3TklJSbrhhhu0adMm9e/fv6e/rltycnKUlZVlf/b5fIQiAABw5V+7DwsL01e+8hW99957ioyMVFtbm5qamvxq6uvr7WeOIiMjL3jrrPPz59U4nc7PDF0Oh0NOp9NvAwAAuOKB6NSpU3r//fc1fPhwJSQkqF+/fiorK7PHq6urVVtbK7fbLUlyu906dOiQGhoa7JrS0lI5nU7FxcXZNeefo7Om8xwAAADd0eOB6Hvf+57Ky8v1wQcfaPfu3frOd76joKAgzZkzRy6XS2lpacrKytLrr7+uyspKzZ8/X263W1OmTJEkTZ8+XXFxcZo3b57+8Ic/aPv27Vq+fLnS09PlcDgkSQsWLNCf/vQnLV26VEePHtVzzz2nTZs2KTMzs6cvBwAAGKDHnyH68MMPNWfOHH300Ue6/vrrdeutt+rNN9/U9ddfL0l65plnFBgYqJkzZ6q1tVUej0fPPfecfXxQUJCKioq0cOFCud1uDRw4UKmpqXryySftmtjYWBUXFyszM1Nr1qxRdHS0XnjhBV65BwAAlyTAsiyrt5voLT6fTy6XS83NzTxPpL75l8GBaw1/7R74fFfi5zd/ywwAABiPQAQAAIxHIAIAAMYjEAEAAOMRiAAAgPEIRAAAwHgEIgAAYDwCEQAAMB6BCAAAGI9ABAAAjEcgAgAAxiMQAQAA4xGIAACA8QhEAADAeMG93QAAoG8buay4t1votg9WpvR2C7jKcIcIAAAYj0AEAACMRyACAADGIxABAADjEYgAAIDxCEQAAMB4vHZ/hfTF11ABADAVd4gAAIDxCEQAAMB4/MoMAK4i/Lod6B3cIQIAAMYjEAEAAOMRiAAAgPEIRAAAwHgEIgAAYLw+H4jWrVunkSNHKjQ0VElJSdq7d29vtwQAAPqYPh2INm7cqKysLK1YsUIHDhzQ+PHj5fF41NDQ0NutAQCAPqRPB6Kf/exneuSRRzR//nzFxcUpPz9fAwYM0H/913/1dmsAAKAP6bMLM7a1tamyslI5OTn2vsDAQCUnJ6uioqLLY1pbW9Xa2mp/bm5uliT5fL4e76+j9ZMePycAoGeMyNzc2y102+EnPL3dwlWj8+e2ZVk9ds4+G4j++te/qr29XREREX77IyIidPTo0S6PycvL0xNPPHHB/piYmCvSIwAAPcW1urc7uPqcPHlSLperR87VZwPRpcjJyVFWVpb9uaOjQydOnNDQoUMVEBDQi531TT6fTzExMTp27JicTmdvt9OnMZc9h7nsGcxjz2Eue07nXNbW1iogIEBRUVE9du4+G4iGDRumoKAg1dfX++2vr69XZGRkl8c4HA45HA6/fWFhYVeqRWM4nU7+R95DmMuew1z2DOax5zCXPcflcvX4XPbZh6pDQkKUkJCgsrIye19HR4fKysrkdrt7sTMAANDX9Nk7RJKUlZWl1NRUJSYmavLkyVq9erVaWlo0f/783m4NAAD0IX06EM2aNUuNjY3Kzc2V1+vVhAkTVFJScsGD1rgyHA6HVqxYccGvIdF9zGXPYS57BvPYc5jLnnMl5zLA6sl31gAAAPqgPvsMEQAAQE8hEAEAAOMRiAAAgPEIRAAAwHgEInymvLw8TZo0SYMHD1Z4eLhmzJih6upqv5ozZ84oPT1dQ4cO1aBBgzRz5swLFszEhVauXKmAgABlZGTY+5jLi/eXv/xF//zP/6yhQ4eqf//+io+P1/79++1xy7KUm5ur4cOHq3///kpOTta7777bix1ffdrb2/XDH/5QsbGx6t+/v2688Ub96Ec/8vv7UMxj13bt2qVvfetbioqKUkBAgLZu3eo3fjHzduLECc2dO1dOp1NhYWFKS0vTqVOnvsCruDp81lyePXtW2dnZio+P18CBAxUVFaUHHnhAdXV1fufoibkkEOEzlZeXKz09XW+++aZKS0t19uxZTZ8+XS0tLXZNZmamXn31VW3evFnl5eWqq6vTvffe24tdX/327dunX/ziFxo3bpzffuby4nz88ce65ZZb1K9fP23btk1vv/22nn76aV133XV2zapVq/Tss88qPz9fe/bs0cCBA+XxeHTmzJle7Pzq8tOf/lTr16/X2rVr9c477+inP/2pVq1apZ///Od2DfPYtZaWFo0fP17r1q3rcvxi5m3u3Lk6cuSISktLVVRUpF27dunRRx/9oi7hqvFZc/nJJ5/owIED+uEPf6gDBw7olVdeUXV1tb797W/71fXIXFpANzQ0NFiSrPLycsuyLKupqcnq16+ftXnzZrvmnXfesSRZFRUVvdXmVe3kyZPWl7/8Zau0tNT6h3/4B+uxxx6zLIu57I7s7Gzr1ltv/dTxjo4OKzIy0nrqqafsfU1NTZbD4bB++ctffhEt9gkpKSnWQw895Lfv3nvvtebOnWtZFvN4sSRZW7ZssT9fzLy9/fbbliRr3759ds22bdusgIAA6y9/+csX1vvV5u/nsit79+61JFl//vOfLcvqubnkDhG6pbm5WZI0ZMgQSVJlZaXOnj2r5ORku2b06NEaMWKEKioqeqXHq116erpSUlL85kxiLrvjN7/5jRITE/WP//iPCg8P180336z/+I//sMdramrk9Xr95tLlcikpKYm5PM/XvvY1lZWV6Y9//KMk6Q9/+IPeeOMN3XnnnZKYx0t1MfNWUVGhsLAwJSYm2jXJyckKDAzUnj17vvCe+5Lm5mYFBATYf4u0p+ayT69UjS9WR0eHMjIydMstt2js2LGSJK/Xq5CQkAv+SG5ERIS8Xm8vdHl1e/nll3XgwAHt27fvgjHm8uL96U9/0vr165WVlaXvf//72rdvn/7t3/5NISEhSk1Ntefr71etZy79LVu2TD6fT6NHj1ZQUJDa29v1k5/8RHPnzpUk5vESXcy8eb1ehYeH+40HBwdryJAhzO1nOHPmjLKzszVnzhz7j7v21FwSiHDR0tPTdfjwYb3xxhu93UqfdOzYMT322GMqLS1VaGhob7fTp3V0dCgxMVH//u//Lkm6+eabdfjwYeXn5ys1NbWXu+s7Nm3apA0bNqiwsFBf/epXVVVVpYyMDEVFRTGPuOqcPXtW//RP/yTLsrR+/foePz+/MsNFWbRokYqKivT6668rOjra3h8ZGam2tjY1NTX51dfX1ysyMvIL7vLqVllZqYaGBk2cOFHBwcEKDg5WeXm5nn32WQUHBysiIoK5vEjDhw9XXFyc374xY8aotrZWkuz5+vs39JhLf0uWLNGyZcs0e/ZsxcfHa968ecrMzFReXp4k5vFSXcy8RUZGqqGhwW/83LlzOnHiBHPbhc4w9Oc//1mlpaX23SGp5+aSQITPZFmWFi1apC1btmjHjh2KjY31G09ISFC/fv1UVlZm76uurlZtba3cbvcX3e5Vbdq0aTp06JCqqqrsLTExUXPnzrX/zVxenFtuueWC5R/++Mc/6oYbbpAkxcbGKjIy0m8ufT6f9uzZw1ye55NPPlFgoP+PgaCgIHV0dEhiHi/Vxcyb2+1WU1OTKisr7ZodO3aoo6NDSUlJX3jPV7POMPTuu+/qtdde09ChQ/3Ge2wuL+EhcBhk4cKFlsvlsnbu3GkdP37c3j755BO7ZsGCBdaIESOsHTt2WPv377fcbrfldrt7seu+4/y3zCyLubxYe/futYKDg62f/OQn1rvvvmtt2LDBGjBggPU///M/ds3KlSutsLAw69e//rV18OBB65577rFiY2Ot06dP92LnV5fU1FTr//2//2cVFRVZNTU11iuvvGINGzbMWrp0qV3DPHbt5MmT1ltvvWW99dZbliTrZz/7mfXWW2/Zbz5dzLzdcccd1s0332zt2bPHeuONN6wvf/nL1pw5c3rrknrNZ81lW1ub9e1vf9uKjo62qqqq/H4Otba22ufoibkkEOEzSepye/HFF+2a06dPW//6r/9qXXfdddaAAQOs73znO9bx48d7r+k+5O8DEXN58V599VVr7NixlsPhsEaPHm09//zzfuMdHR3WD3/4QysiIsJyOBzWtGnTrOrq6l7q9urk8/msxx57zBoxYoQVGhpqfelLX7J+8IMf+P2gYR679vrrr3f5/42pqamWZV3cvH300UfWnDlzrEGDBllOp9OaP3++dfLkyV64mt71WXNZU1PzqT+HXn/9dfscPTGXAZZ13pKkAAAABuIZIgAAYDwCEQAAMB6BCAAAGI9ABAAAjEcgAgAAxiMQAQAA4xGIAACA8QhEAADAeAQiAABgPAIRAAAwHoEIAAAYj0AEAACM9/8B5epgvKkTTIIAAAAASUVORK5CYII="
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "execution_count": 6
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-05-15T17:13:10.032414Z",
     "start_time": "2024-05-15T17:13:10.027516Z"
    }
   },
   "cell_type": "code",
   "source": [
    "train_titles[:2]"
   ],
   "id": "6c6fd3cfb68c15c7",
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['AMD Debuts Dual-Core Opteron Processor',\n",
       "       \"Wood's Suspension Upheld (Reuters)\"], dtype='<U115')"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "execution_count": 7
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-05-15T17:26:37.184545Z",
     "start_time": "2024-05-15T17:26:35.282986Z"
    }
   },
   "cell_type": "code",
   "source": [
    "from reviews import wrangle_data\n",
    "word_dimension = 9000 # about 95% fluency\n",
    "sequence_length = 45\n",
    "\n",
    "# oov == out of value token\n",
    "tokenizer = tf.keras.preprocessing.text.Tokenizer(num_words=word_dimension, oov_token='~~~')\n",
    "tokenizer.fit_on_texts(train_titles)\n",
    "\n",
    "train_data, train_labels, train_tokens = wrangle_data(tokenizer, train_titles, train_labels, sequence_length)\n",
    "test_data, test_labels, test_tokens = wrangle_data(tokenizer, test_titles, test_labels, sequence_length)"
   ],
   "id": "21e993b570183285",
   "outputs": [],
   "execution_count": 24
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-05-15T17:26:37.189079Z",
     "start_time": "2024-05-15T17:26:37.185713Z"
    }
   },
   "cell_type": "code",
   "source": [
    "train_labels.shape"
   ],
   "id": "434f8331000b3eb5",
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(120000,)"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "execution_count": 25
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-05-15T17:26:37.481897Z",
     "start_time": "2024-05-15T17:26:37.479418Z"
    }
   },
   "cell_type": "code",
   "source": [
    "import math\n",
    "\n",
    "def fourth_root(n):\n",
    "    return math.pow(n, 0.25)"
   ],
   "id": "9c98b6b99ed9ad9",
   "outputs": [],
   "execution_count": 26
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-05-15T17:26:38.372238Z",
     "start_time": "2024-05-15T17:26:38.312432Z"
    }
   },
   "cell_type": "code",
   "source": [
    "from reviews import save_model, save_history\n",
    "# Build the model\n",
    "\n",
    "def compile_model(model):\n",
    "    model.compile(optimizer='adam', loss='sparse_categorical_crossentropy', metrics=['accuracy'])\n",
    "    print(model.summary())\n",
    "    return model\n",
    "\n",
    "def cnn_model(word_dimension, embedding_dimension, sequence_length):\n",
    "    model = tf.keras.Sequential([\n",
    "        tf.keras.layers.Embedding(word_dimension, embedding_dimension, input_length=sequence_length),\n",
    "        tf.keras.layers.Conv1D(filters=128, kernel_size=8, activation='relu'),\n",
    "        tf.keras.layers.GlobalAveragePooling1D(),\n",
    "        tf.keras.layers.Dense(32, activation='relu'),\n",
    "        tf.keras.layers.Dropout(0.5),\n",
    "        tf.keras.layers.Dense(4, activation='sigmoid')\n",
    "    ])\n",
    "\n",
    "    return compile_model(model)\n",
    "\n",
    "# CNN model\n",
    "model_name = \"ag-news-cnn\"\n",
    "\n",
    "earlystop = tf.keras.callbacks.EarlyStopping('val_loss', patience=3, restore_best_weights=True)\n",
    "checkpoint = tf.keras.callbacks.ModelCheckpoint(filepath=f'ckpts/reviews/{model_name}-'+'{epoch:02d}-{val_accuracy:.4f}')\n",
    "embedding_dimension = int(fourth_root(word_dimension))\n",
    "\n",
    "cnn = cnn_model(word_dimension, embedding_dimension, sequence_length)"
   ],
   "id": "487cd08daefc8719",
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_5\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " embedding_5 (Embedding)     (None, 45, 9)             81000     \n",
      "                                                                 \n",
      " conv1d_5 (Conv1D)           (None, 38, 128)           9344      \n",
      "                                                                 \n",
      " global_average_pooling1d_5  (None, 128)               0         \n",
      "  (GlobalAveragePooling1D)                                       \n",
      "                                                                 \n",
      " dense_10 (Dense)            (None, 32)                4128      \n",
      "                                                                 \n",
      " dropout_5 (Dropout)         (None, 32)                0         \n",
      "                                                                 \n",
      " dense_11 (Dense)            (None, 4)                 132       \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 94604 (369.55 KB)\n",
      "Trainable params: 94604 (369.55 KB)\n",
      "Non-trainable params: 0 (0.00 Byte)\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    }
   ],
   "execution_count": 27
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-05-15T17:27:14.810346Z",
     "start_time": "2024-05-15T17:26:40.278227Z"
    }
   },
   "cell_type": "code",
   "source": [
    "\n",
    "history_cnn = cnn.fit(train_data, train_labels, validation_split=0.1, epochs=25, callbacks=[earlystop, checkpoint],\n",
    "                        batch_size=64)\n",
    "\n",
    "save_model(cnn, model_name, history_cnn, test_data, test_labels)\n",
    "save_history(history_cnn, model_name)"
   ],
   "id": "d28492d80f441436",
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/25\n",
      "1671/1688 [============================>.] - ETA: 0s - loss: 0.7756 - accuracy: 0.6966INFO:tensorflow:Assets written to: ckpts/reviews/ag-news-cnn-01-0.8480/assets\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: ckpts/reviews/ag-news-cnn-01-0.8480/assets\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1688/1688 [==============================] - 5s 3ms/step - loss: 0.7733 - accuracy: 0.6980 - val_loss: 0.4628 - val_accuracy: 0.8480\n",
      "Epoch 2/25\n",
      "1682/1688 [============================>.] - ETA: 0s - loss: 0.4765 - accuracy: 0.8476INFO:tensorflow:Assets written to: ckpts/reviews/ag-news-cnn-02-0.8627/assets\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: ckpts/reviews/ag-news-cnn-02-0.8627/assets\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1688/1688 [==============================] - 5s 3ms/step - loss: 0.4762 - accuracy: 0.8476 - val_loss: 0.4144 - val_accuracy: 0.8627\n",
      "Epoch 3/25\n",
      "1679/1688 [============================>.] - ETA: 0s - loss: 0.4147 - accuracy: 0.8666INFO:tensorflow:Assets written to: ckpts/reviews/ag-news-cnn-03-0.8633/assets\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: ckpts/reviews/ag-news-cnn-03-0.8633/assets\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1688/1688 [==============================] - 6s 3ms/step - loss: 0.4151 - accuracy: 0.8666 - val_loss: 0.4062 - val_accuracy: 0.8633\n",
      "Epoch 4/25\n",
      "1678/1688 [============================>.] - ETA: 0s - loss: 0.3848 - accuracy: 0.8729INFO:tensorflow:Assets written to: ckpts/reviews/ag-news-cnn-04-0.8633/assets\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: ckpts/reviews/ag-news-cnn-04-0.8633/assets\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1688/1688 [==============================] - 6s 4ms/step - loss: 0.3850 - accuracy: 0.8728 - val_loss: 0.4065 - val_accuracy: 0.8633\n",
      "Epoch 5/25\n",
      "1670/1688 [============================>.] - ETA: 0s - loss: 0.3627 - accuracy: 0.8787INFO:tensorflow:Assets written to: ckpts/reviews/ag-news-cnn-05-0.8633/assets\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: ckpts/reviews/ag-news-cnn-05-0.8633/assets\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1688/1688 [==============================] - 6s 3ms/step - loss: 0.3625 - accuracy: 0.8786 - val_loss: 0.4220 - val_accuracy: 0.8633\n",
      "Epoch 6/25\n",
      "1669/1688 [============================>.] - ETA: 0s - loss: 0.3419 - accuracy: 0.8855INFO:tensorflow:Assets written to: ckpts/reviews/ag-news-cnn-06-0.8629/assets\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: ckpts/reviews/ag-news-cnn-06-0.8629/assets\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1688/1688 [==============================] - 6s 3ms/step - loss: 0.3419 - accuracy: 0.8854 - val_loss: 0.4288 - val_accuracy: 0.8629\n",
      "238/238 [==============================] - 0s 770us/step - loss: 0.4116 - accuracy: 0.8613\n",
      "INFO:tensorflow:Assets written to: models/reviews-ag-news-cnn-6-0.8613.tf/assets\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: models/reviews-ag-news-cnn-6-0.8613.tf/assets\n"
     ]
    }
   ],
   "execution_count": 28
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-05-15T17:27:43.238644Z",
     "start_time": "2024-05-15T17:27:42.905768Z"
    }
   },
   "cell_type": "code",
   "source": [
    "# plot the training history\n",
    "from history import plot_history\n",
    "plot_history(history_cnn)"
   ],
   "id": "de5182b493738c9",
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<Figure size 2400x800 with 2 Axes>"
      ],
      "image/png": "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"
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "execution_count": 30
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-05-15T17:28:02.947920Z",
     "start_time": "2024-05-15T17:28:02.723481Z"
    }
   },
   "cell_type": "code",
   "source": [
    "# evaluate on test data\n",
    "loss, accuracy = cnn.evaluate(test_data, test_labels)\n",
    "print(f'Test Loss: {loss}, Test Accuracy: {accuracy}')"
   ],
   "id": "729ddac16fbff9ff",
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "238/238 [==============================] - 0s 729us/step - loss: 0.4116 - accuracy: 0.8613\n",
      "Test Loss: 0.41159871220588684, Test Accuracy: 0.8613157868385315\n"
     ]
    }
   ],
   "execution_count": 31
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}