multi_class_with_Distilbert.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": [],
      "machine_shape": "hm",
      "authorship_tag": "ABX9TyPBSCDroKW1kJdrskDDp0A2",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "gpuClass": "premium",
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/Duygubirol/NLP/blob/main/multi_class_with_Distilbert.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "GQ5ww4uKmS8U"
      },
      "outputs": [],
      "source": [
        "# Importing Libraries\n",
        "!pip install transformers\n",
        "from transformers import RobertaTokenizerFast, TFRobertaForSequenceClassification, TFTrainer, TFTrainingArguments, DataCollatorWithPadding\n",
        "import tensorflow as tf\n",
        "from sklearn.metrics import classification_report\n",
        "import numpy as np\n",
        "import pandas as pd\n",
        "from sklearn.model_selection import train_test_split\n",
        "from sklearn.metrics import confusion_matrix, matthews_corrcoef\n",
        "from matplotlib import pyplot as plt"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "\n",
        "# Read Data \n",
        "df = pd.read_csv('/content/df_all.csv')\n",
        "df.drop(columns=df.columns[0], axis=1, inplace=True)\n",
        "df.head()"
      ],
      "metadata": {
        "id": "-aS9T1S7vbFK",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 206
        },
        "outputId": "372a4ffd-4f03-45ed-fa44-e9ad8bceec7b"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "                                                text          class\n",
              "0   rt crooked hillary you didn't lose because yo...  explicit_hate\n",
              "1   rt estonian pm :  \" i don't want blacks in es...  explicit_hate\n",
              "2  are a bunch of religious bigots . i hope you e...  explicit_hate\n",
              "3  hamas is the only one who lies and all those l...  explicit_hate\n",
              "4  one of the questions never asked : if africans...  explicit_hate"
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-3e8721db-7955-4888-af5e-c1d7685e923e\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>text</th>\n",
              "      <th>class</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>rt crooked hillary you didn't lose because yo...</td>\n",
              "      <td>explicit_hate</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>rt estonian pm :  \" i don't want blacks in es...</td>\n",
              "      <td>explicit_hate</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>are a bunch of religious bigots . i hope you e...</td>\n",
              "      <td>explicit_hate</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>hamas is the only one who lies and all those l...</td>\n",
              "      <td>explicit_hate</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>one of the questions never asked : if africans...</td>\n",
              "      <td>explicit_hate</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-3e8721db-7955-4888-af5e-c1d7685e923e')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-3e8721db-7955-4888-af5e-c1d7685e923e button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-3e8721db-7955-4888-af5e-c1d7685e923e');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ]
          },
          "metadata": {},
          "execution_count": 50
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Check shape of dataframe \n",
        "df.shape"
      ],
      "metadata": {
        "id": "BwQC2Fnqv1vF",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "78161f8b-40b0-4183-d2de-a91953a9d361"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(64875, 2)"
            ]
          },
          "metadata": {},
          "execution_count": 51
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# drop NA \n",
        "df = df.dropna()"
      ],
      "metadata": {
        "id": "4uRj7mpFwMuL"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# Check distribution of classes\n",
        "df['class'].value_counts()"
      ],
      "metadata": {
        "id": "a1Akk8trwPJF",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "85910935-a6b9-496a-b540-7756c837e904"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "not_hate         35977\n",
              "offensive        19190\n",
              "implicit_hate     7100\n",
              "explicit_hate     2519\n",
              "Name: class, dtype: int64"
            ]
          },
          "metadata": {},
          "execution_count": 53
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# layers for multi class  classification \n",
        "\n",
        "# Copy Dataframe first\n",
        "df_multiclass = df.copy()\n",
        "\n",
        "# Replace classes \n",
        "df_multiclass['class'] = np.where((df_multiclass['class'] == 'not_hate'), 0, df_multiclass['class'])\n",
        "df_multiclass['class'] = np.where((df_multiclass['class'] == 'offensive'), 1, df_multiclass['class'])\n",
        "df_multiclass['class'] = np.where((df_multiclass['class'] == 'implicit_hate'), 2, df_multiclass['class'])\n",
        "df_multiclass['class'] = np.where((df_multiclass['class'] == 'explicit_hate'), 3, df_multiclass['class'])\n",
        "\n",
        "# show head \n",
        "df_multiclass.head()"
      ],
      "metadata": {
        "id": "0_FeV0vmwSIz",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 206
        },
        "outputId": "7c22793a-5951-4291-a39f-35d647adb664"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "                                                text class\n",
              "0   rt crooked hillary you didn't lose because yo...     3\n",
              "1   rt estonian pm :  \" i don't want blacks in es...     3\n",
              "2  are a bunch of religious bigots . i hope you e...     3\n",
              "3  hamas is the only one who lies and all those l...     3\n",
              "4  one of the questions never asked : if africans...     3"
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-7f31b62f-be92-4b42-96e6-d0574571e5ad\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>text</th>\n",
              "      <th>class</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>rt crooked hillary you didn't lose because yo...</td>\n",
              "      <td>3</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>rt estonian pm :  \" i don't want blacks in es...</td>\n",
              "      <td>3</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>are a bunch of religious bigots . i hope you e...</td>\n",
              "      <td>3</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>hamas is the only one who lies and all those l...</td>\n",
              "      <td>3</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>one of the questions never asked : if africans...</td>\n",
              "      <td>3</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-7f31b62f-be92-4b42-96e6-d0574571e5ad')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-7f31b62f-be92-4b42-96e6-d0574571e5ad button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-7f31b62f-be92-4b42-96e6-d0574571e5ad');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ]
          },
          "metadata": {},
          "execution_count": 54
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Show Distribution\n",
        "df_multiclass['class'].value_counts()"
      ],
      "metadata": {
        "id": "_eFNDvMfwV-y",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "baef1b6e-c21a-47e7-ac3f-a8964082eaa7"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0    35977\n",
              "1    19190\n",
              "2     7100\n",
              "3     2519\n",
              "Name: class, dtype: int64"
            ]
          },
          "metadata": {},
          "execution_count": 55
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "\n",
        "# Get max number of words \n",
        "num_words = df_multiclass['text'].apply(lambda x:len(str(x).split()))\n",
        "print('The comment with the most words consist of', num_words.max(),'words')"
      ],
      "metadata": {
        "id": "I5PdAT8RwYi2",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "53096d72-958b-4ba1-a257-97426c7bfa72"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "The comment with the most words consist of 268 words\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Visualize \n",
        "plt.hist(num_words, bins=15)\n",
        "plt.show()"
      ],
      "metadata": {
        "id": "syLkIZbmwbY2",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 592
        },
        "outputId": "4a9338df-4fe9-47c8-b5a8-299c1571b4e3"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 720x720 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "\n",
        "# Create X  \n",
        "X=list(df_multiclass['text']) \n",
        "\n",
        "# Create y \n",
        "y=list(df_multiclass['class']) \n",
        "\n",
        "# Train-Test Split\n",
        "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size = 0.20, random_state = 123) "
      ],
      "metadata": {
        "id": "QgbBBjtPwd5g"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "\n",
        "categories=sorted(list(set(y))) #set will return the unique different entries"
      ],
      "metadata": {
        "id": "OjEqvsp8wiBj"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "num_categories = len(categories)\n",
        "num_categories"
      ],
      "metadata": {
        "id": "5P2LuHTCwksv",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "46c7b5eb-cfeb-49f3-feb4-2253501d0339"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "4"
            ]
          },
          "metadata": {},
          "execution_count": 60
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "from transformers import AutoModelForSequenceClassification\n",
        "from transformers import TFAutoModelForSequenceClassification"
      ],
      "metadata": {
        "id": "FK-olOLx1PGP"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "from transformers import DistilBertTokenizerFast\n",
        "tokenizer = DistilBertTokenizerFast.from_pretrained('distilbert-base-uncased')\n",
        "\n",
        "# Tokenize X Train & Test Set \n",
        "train_input = tokenizer(X_train, truncation=True, padding=True, return_tensors='tf')\n",
        "test_input = tokenizer(X_test, truncation=True, padding=True, return_tensors='tf')"
      ],
      "metadata": {
        "id": "8fKRJmyYwoIe"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "train_dataset = tf.data.Dataset.from_tensor_slices((\n",
        "    dict(train_input),\n",
        "    y_train\n",
        "))\n",
        "\n",
        "test_dataset = tf.data.Dataset.from_tensor_slices((\n",
        "    dict(test_input),\n",
        "    y_test\n",
        "))"
      ],
      "metadata": {
        "id": "ZCqQgVqEwr0B"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "training_args = TFTrainingArguments(\n",
        "    output_dir='./results',          \n",
        "    num_train_epochs=3,              \n",
        "    per_device_train_batch_size=16,  \n",
        "    per_device_eval_batch_size=16,   \n",
        "    warmup_steps = 500,\n",
        "    weight_decay=0.01,\n",
        "    logging_steps=10,\n",
        "    eval_steps = 10\n",
        ")"
      ],
      "metadata": {
        "id": "8cxvMDfFwvkn"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "from transformers import TFDistilBertForSequenceClassification, TFTrainer, TFTrainingArguments"
      ],
      "metadata": {
        "id": "wdXQn8gO1p_D"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "from sklearn.metrics import f1_score, roc_auc_score"
      ],
      "metadata": {
        "id": "hX4ZDMqK7O3c"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "with training_args.strategy.scope():\n",
        "    model = TFDistilBertForSequenceClassification.from_pretrained(\"distilbert-base-uncased\",num_labels=num_categories)\n",
        "\n",
        "trainer = TFTrainer(\n",
        "    model=model,                         \n",
        "    args=training_args,                  \n",
        "    train_dataset=train_dataset,         \n",
        "    eval_dataset=test_dataset\n",
        ")\n",
        "\n",
        "trainer.train()"
      ],
      "metadata": {
        "id": "dr5nf3WowyFS",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "3a4dd129-0e2a-42f9-a568-305d8ed9dd8d"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "Some layers from the model checkpoint at distilbert-base-uncased were not used when initializing TFDistilBertForSequenceClassification: ['vocab_layer_norm', 'vocab_transform', 'vocab_projector', 'activation_13']\n",
            "- This IS expected if you are initializing TFDistilBertForSequenceClassification from the checkpoint of a model trained on another task or with another architecture (e.g. initializing a BertForSequenceClassification model from a BertForPreTraining model).\n",
            "- This IS NOT expected if you are initializing TFDistilBertForSequenceClassification from the checkpoint of a model that you expect to be exactly identical (initializing a BertForSequenceClassification model from a BertForSequenceClassification model).\n",
            "Some layers of TFDistilBertForSequenceClassification were not initialized from the model checkpoint at distilbert-base-uncased and are newly initialized: ['pre_classifier', 'dropout_59', 'classifier']\n",
            "You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference.\n",
            "/usr/local/lib/python3.8/dist-packages/transformers/trainer_tf.py:115: FutureWarning: The class `TFTrainer` is deprecated and will be removed in version 5 of Transformers. We recommend using native Keras instead, by calling methods like `fit()` and `predict()` directly on the model object. Detailed examples of the Keras style can be found in our examples at https://github.com/huggingface/transformers/tree/main/examples/tensorflow\n",
            "  warnings.warn(\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "output = trainer.predict(test_dataset)[0]\n",
        "output = np.argmax(output, axis = - 1)"
      ],
      "metadata": {
        "id": "yY0qIXUd86q3"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "cm=confusion_matrix(y_test,output)\n",
        "cm"
      ],
      "metadata": {
        "id": "Lhps_u69w25f",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "b22eb9fa-69da-4c57-f8af-1a96be37964c"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([[6604,   91,  445,   47],\n",
              "       [  64, 3671,    7,   93],\n",
              "       [ 571,    6,  813,   45],\n",
              "       [  70,  161,  102,  168]])"
            ]
          },
          "metadata": {},
          "execution_count": 69
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "print(classification_report(y_test, output))"
      ],
      "metadata": {
        "id": "ZaSYVUVb5iDe",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "a8d67c32-b64c-4215-ba89-d937c6994639"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       0.90      0.92      0.91      7187\n",
            "           1       0.93      0.96      0.95      3835\n",
            "           2       0.59      0.57      0.58      1435\n",
            "           3       0.48      0.34      0.39       501\n",
            "\n",
            "    accuracy                           0.87     12958\n",
            "   macro avg       0.73      0.69      0.71     12958\n",
            "weighted avg       0.86      0.87      0.86     12958\n",
            "\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "from sklearn import metrics\n",
        "plt.rcParams['figure.figsize'] = (10, 10)\n",
        "confusion_matrix = metrics.confusion_matrix(y_test, output)\n",
        "\n",
        "cm_display = metrics.ConfusionMatrixDisplay(confusion_matrix = confusion_matrix, display_labels = [\"not_hate\", \"offensive\", \"implicit\", \"explicit\" ])\n",
        "\n",
        "cm_display.plot()"
      ],
      "metadata": {
        "id": "D9w__9vd5poD",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 592
        },
        "outputId": "05a5b4f0-2dd8-46a7-c25e-b51ab75aa043"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<sklearn.metrics._plot.confusion_matrix.ConfusionMatrixDisplay at 0x7fbbf127e2b0>"
            ]
          },
          "metadata": {},
          "execution_count": 71
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 720x720 with 2 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    }
  ]
}