TestBERTModel_debut__vs_all.ipynb

{
  "cells": [
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "492db99a",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "492db99a",
        "outputId": "e505b261-82ac-4a8e-b59d-7c90fed55ab4"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Drive already mounted at /content/drive; to attempt to forcibly remount, call drive.mount(\"/content/drive\", force_remount=True).\n"
          ]
        }
      ],
      "source": [
        "import json\n",
        "import os\n",
        "import re\n",
        "# drive à importer : https://drive.google.com/drive/folders/1vn_RM47LA_HdpQwZqdwqNRPm7CXyeeV3?usp=sharing\n",
        "# \n",
        "#1 chargement des datas\n",
        "\n",
        "#Nous allons tester la classification binaire \n",
        "from google.colab import drive\n",
        "drive.mount('/content/drive')"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "from matplotlib import pyplot as plt\n",
        "import seaborn as sn\n",
        "from sklearn.metrics import confusion_matrix, classification_report\n",
        "import numpy as np\n"
      ],
      "metadata": {
        "id": "EQuKDaHmTZA2"
      },
      "id": "EQuKDaHmTZA2",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "47bcf75a",
      "metadata": {
        "id": "47bcf75a"
      },
      "outputs": [],
      "source": [
        "# Définit le chemin vers le dossier contenant les différentes types de datas\n",
        "\n",
        "###WhisperX\n",
        "\n",
        "#whisperX \n",
        "folder_path_whisperX_ad = \"/content/drive/MyDrive/Alzheimer_transcripts/transcriptions/whisperX/ad\"\n",
        "folder_path_whisperX_cn = \"/content/drive/MyDrive/Alzheimer_transcripts/transcriptions/whisperX/cn\"\n",
        "\n",
        "#whisperX with segmented 30sec audios length\n",
        "folder_path_whisperX_ad_seg = \"/content/drive/MyDrive/Alzheimer_transcripts/transcriptions/whisperX_seg/ad\"\n",
        "folder_path_whisperX_cn_seg = \"/content/drive/MyDrive/Alzheimer_transcripts/transcriptions/whisperX_seg/cn\"\n",
        "\n",
        "#whisperX with segmented 30sec audios length and whistespaces representation \n",
        "folder_path_whisperX_ad_ws_seg = \"/content/drive/MyDrive/Alzheimer_transcripts/transcriptions/whisperX_seg/white_spaces_ad\"\n",
        "folder_path_whisperX_cn_ws_seg = \"/content/drive/MyDrive/Alzheimer_transcripts/transcriptions/whisperX_seg/white_spaces_cn\"\n",
        "\n",
        "\n",
        "#whisperX with whistespaces representation \n",
        "folder_path_whisperX_ad_ws = \"/content/drive/MyDrive/Alzheimer_transcripts/transcriptions/whisperX/white_spaces_ad\"\n",
        "folder_path_whisperX_cn_ws = \"/content/drive/MyDrive/Alzheimer_transcripts/transcriptions/whisperX/white_spaces_cn\"\n",
        "\n",
        "\n",
        "\n",
        "###Whisper classic\n",
        "folder_path_whisper_ad = \"/content/drive/MyDrive/Alzheimer_transcripts/transcriptions/whisper/ad\"\n",
        "folder_path_whisper_cn = \"/content/drive/MyDrive/Alzheimer_transcripts/transcriptions/whisper/cn\"\n",
        "\n",
        "###Whisper classic with segmented 30sec audios length\n",
        "folder_path_whisper_ad_seg = \"/content/drive/MyDrive/Alzheimer_transcripts/transcriptions/whisper_seg/ad\"\n",
        "folder_path_whisper_cn_seg = \"/content/drive/MyDrive/Alzheimer_transcripts/transcriptions/whisper_seg/cn\"\n",
        "\n",
        "\n",
        "\n",
        "#### Only debut of audio (index 0 of segmented)\n",
        "folder_path_whisper_debut = \"/content/drive/MyDrive/Alzheimer_transcripts/transcriptions/whisper_seg/debut\"\n",
        "folder_path_whisperX_debut = \"/content/drive/MyDrive/Alzheimer_transcripts/transcriptions/whisperX_seg/debut\"\n",
        "folder_path_whisperX_ws_debut = \"/content/drive/MyDrive/Alzheimer_transcripts/transcriptions/whisperX_seg/debut_white_spaces\"\n",
        "#creer dossier avec \n",
        "\n",
        "\n",
        "###Whisper \n",
        "\n",
        "###WhisperX\n",
        "\n",
        "###\n",
        "# # Liste tous les fichiers dans les 2 dossiers\n",
        "# def list_files(folder_ad, folder_cn):\n",
        "#   files = []\n",
        "#   for data in [folder_ad, folder_cn]:\n",
        "#       files.extend(os.listdir(data))    \n",
        "#   return files   \n",
        "\n",
        "# files_whisperX = list_files(folder_path_whisperX_ad, folder_path_whisperX_cn)\n",
        "# print(\"taille whisperX files : \", len(files_whisperX))\n",
        "# files_whisperX_seg = list_files(folder_path_whisperX_ad_seg, folder_path_whisperX_cn_seg)\n",
        "# print(\"taille whisperX_seg files : \",len(files_whisperX_seg))\n",
        "# files_whisperX_seg_ws = list_files(folder_path_whisperX_ad_ws_seg, folder_path_whisperX_cn_ws_seg)\n",
        "# print(\"taille whisperX_seg_ws files : \",len(files_whisperX_seg_ws))\n",
        "# files_whisperX_ws = list_files(folder_path_whisperX_ad_ws, folder_path_whisperX_cn_ws)\n",
        "# print(\"taille whisperX_ws files : \",len(files_whisperX_ws))\n",
        "\n",
        "\n",
        "# files_whisper = list_files(folder_path_whisper_ad, folder_path_whisper_cn)\n",
        "# print(\"taille whisper files : \", len(files_whisper))\n",
        "# files_whisper_seg = list_files(folder_path_whisper_ad_seg, folder_path_whisper_cn_seg)\n",
        "# print(\"taille whisper_seg files : \", len(files_whisper_seg))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "6d6d201e",
      "metadata": {
        "id": "6d6d201e"
      },
      "outputs": [],
      "source": [
        "def from_json_to_dict(folder_ad, folder_cn):\n",
        "    data_ad = []\n",
        "    data_cn = []\n",
        "    for dir_path in [folder_ad, folder_cn]:\n",
        "        for file_name in os.listdir(dir_path):\n",
        "            if file_name.endswith('.json'):\n",
        "                file_path = os.path.join(dir_path, file_name)\n",
        "                with open(file_path, 'r') as f:\n",
        "                    data = json.load(f)\n",
        "                    if dir_path == folder_ad:\n",
        "                        data_ad.append(data)\n",
        "                    elif dir_path == folder_cn:\n",
        "                        data_cn.append(data)\n",
        "    return data_ad, data_cn             \n",
        "\n",
        "                \n",
        "whisperX_data_ad, whisperX_data_cn = from_json_to_dict(folder_path_whisperX_ad, folder_path_whisperX_cn)\n",
        "whisperX_seg_data_ad, whisperX_seg_data_cn = from_json_to_dict(folder_path_whisperX_ad_seg, folder_path_whisperX_cn_seg)\n",
        "whisperX_seg_ws_data_ad, whisperX_seg_ws_data_cn = from_json_to_dict(folder_path_whisperX_ad_ws_seg, folder_path_whisperX_cn_ws_seg)\n",
        "whisperX_ws_data_ad, whisperX_ws_data_cn = from_json_to_dict(folder_path_whisperX_ad_ws, folder_path_whisperX_cn_ws)\n",
        "\n",
        "\n",
        "whisper_seg_data_ad, whisper_seg_data_cn = from_json_to_dict(folder_path_whisper_ad_seg, folder_path_whisper_cn_seg)\n"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "whisper_data_ad, whisper_data_cn = from_json_to_dict(folder_path_whisper_ad, folder_path_whisper_cn)"
      ],
      "metadata": {
        "id": "zHk7mTKnKYTD"
      },
      "id": "zHk7mTKnKYTD",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "def from_json_to_dict_debut(folder_debut):\n",
        "  data_ad = []\n",
        "  data_cn = []\n",
        "  for file_name in os.listdir(folder_debut):\n",
        "    if file_name.endswith('.json'):\n",
        "        file_path = os.path.join(folder_debut, file_name)\n",
        "        with open(file_path, 'r') as f:\n",
        "          data = json.load(f)\n",
        "          data_ad.append(data) if data[\"label\"] else data_cn.append(data)\n",
        "  return data_ad, data_cn"
      ],
      "metadata": {
        "id": "bl4e2LUTsx6N"
      },
      "id": "bl4e2LUTsx6N",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "whisper_debut_data_ad, whisper_debut_data_cn = from_json_to_dict_debut(folder_path_whisper_debut)\n",
        "\n",
        "whisperX_debut_data_ad, whisperX_debut_data_cn = from_json_to_dict_debut(folder_path_whisperX_debut)\n",
        "whisper_debut_ws_data_ad, whisper_debut_ws_data_cn = from_json_to_dict_debut(folder_path_whisperX_ws_debut)"
      ],
      "metadata": {
        "id": "JBm8ktEwtgXa"
      },
      "id": "JBm8ktEwtgXa",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [],
      "metadata": {
        "id": "8kLWE6kBuD7l"
      },
      "id": "8kLWE6kBuD7l",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "from random import choice\n",
        "\n",
        "print(\"\\nexemple d'une data ad avec whisperX :\\n \", choice(whisperX_data_ad))\n",
        "print(\"\\nexemple d'une data cn avec whisperX :\\n \", choice(whisperX_data_cn))\n",
        "\n",
        "print(\"\\nexemple d'une data ad avec whisperX_ws :\\n \", choice(whisperX_ws_data_ad))"
      ],
      "metadata": {
        "id": "gakJZ4yaItVp",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "1dbefcfd-b769-41ff-adca-40321fc6edb6"
      },
      "id": "gakJZ4yaItVp",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "\n",
            "exemple d'une data ad avec whisperX :\n",
            "  {'text': \"And here's the picture. The kid's still turning over. She's trying to steal cookies out of the cookie jar. The little girl's reaching up to grab hers. The mother's doing the dishes but let the water run over all over the floor. And she's standing in the puddle. Thank you. Mm-hmm. Drying the dishes. Did I say that before? Yeah. Kids laughing. The other kid with the storm falling over. Break his neck. Mm-hmm. There's no action to the curtains and grass is growing out there I guess you could call that action. Leaves are probably glowing. Others in their own little world. Water's still running. Okay. He's scanning the map. Like he made them. The chair is falling over the stool. Okay. Good. I think the little kid kicked it and pushed him over. Okay. Instead of holding him up she gave a shove. Okay. Good.\", 'label': 1}\n",
            "\n",
            "exemple d'une data cn avec whisperX :\n",
            "  {'text': \"And then here on this one, tell me everything you see going on in this picture. Well, I see the mother is not paying attention and her water is running over her sink. The boy has climbed up on a ladder to steal some cookies and he's falling over. The girl is laughing. The grass is growing outside. Nobody seems to care. They're just going on with their daily activities. Anything else you see? No right or wrong answers, but is there anything else you see going on in the picture? No, not really.\", 'label': 0}\n",
            "\n",
            "exemple d'une data ad avec whisperX_ws :\n",
            "  {'text': \"Tell me everything that you see going on in that picture. .. Well, the lady's drying her dishes. Then the little boy's up on the ladder reaching for a cookie jar. And a little girl is down on the floor reaching up toward him. And I see a sink, the dishes, cupboard. . and then there's a walkway in the back. I guess that's the walkway I call. . Anything else? What else is going on in that picture? The water's running over the sink. .. Anything else? . I said little boys climbing up on a cookie for the cookies.\", 'label': 1}\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "dd4c6b38",
      "metadata": {
        "id": "dd4c6b38"
      },
      "outputs": [],
      "source": [
        "#delete des keys autre que 'text' et 'label'\n",
        "keys_to_keep = ['text', 'label']\n",
        "\n",
        "def delete_keys(d, keys_to_keep):\n",
        "    for key in list(d.keys()):\n",
        "        if key not in keys_to_keep:\n",
        "            del d[key]\n",
        "\n",
        "for data in (whisperX_data_ad + whisperX_data_cn + \\\n",
        "             whisperX_seg_data_ad + whisperX_seg_data_cn + \\\n",
        "             whisperX_seg_ws_data_ad + whisperX_seg_ws_data_cn +\\\n",
        "             whisperX_ws_data_ad + whisperX_ws_data_cn +\\\n",
        "             whisper_seg_data_ad + whisper_seg_data_cn +\\\n",
        "             whisper_data_ad + whisper_data_cn)+\\\n",
        "             whisper_debut_data_ad + whisper_debut_data_cn+\\\n",
        "             whisperX_debut_data_ad + whisperX_debut_data_cn+\\\n",
        "             whisper_debut_ws_data_ad + whisper_debut_ws_data_cn:\n",
        "    delete_keys(data, keys_to_keep)\n",
        "    "
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "41335178",
      "metadata": {
        "id": "41335178"
      },
      "outputs": [],
      "source": []
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "6eaa6fac",
      "metadata": {
        "id": "6eaa6fac"
      },
      "outputs": [],
      "source": [
        "### nous avons 6 types de données différentes, on leur attribue un numéros pour la lisibilité des variables:\n",
        "# 0 : whisper\n",
        "# 1 : whisper_seg\n",
        "\n",
        "# 2 : whisperX\n",
        "# 3 : whisperX_seg\n",
        "# 4 : whisperX_seg_ws\n",
        "# 5 : whisperX_ws\n",
        "\n",
        "## à cela s'ajoute seulement les segments debut pour faire une comparaison (premier segment de 30 secondes)\n",
        "# 6 : whisper debut \n",
        "\n",
        "# 7 : whisperX debut \n",
        "# 8 : whisperX_ws debut \n",
        "from sklearn.model_selection import train_test_split\n",
        "\n",
        "data_0 = whisper_data_ad + whisper_data_cn\n",
        "data_1 = whisper_seg_data_ad + whisper_seg_data_cn \n",
        "data_2 =whisperX_data_ad + whisperX_data_cn\n",
        "data_3 = whisperX_seg_data_ad + whisperX_seg_data_cn\n",
        "data_4 = whisperX_seg_ws_data_ad + whisperX_seg_ws_data_cn\n",
        "data_5 = whisperX_ws_data_ad + whisperX_ws_data_cn\n",
        "\n",
        "data_6 = whisper_debut_data_ad + whisper_debut_data_cn\n",
        "data_7 = whisperX_debut_data_ad + whisperX_debut_data_cn\n",
        "data_8 = whisper_debut_ws_data_ad + whisper_debut_ws_data_cn\n",
        "\n",
        "texts_0, labels_0 = zip(*[(d['text'], d['label']) for d in data_0])\n",
        "texts_1, labels_1 = zip(*[(d['text'], d['label']) for d in data_1])\n",
        "texts_2, labels_2 = zip(*[(d['text'], d['label']) for d in data_2])\n",
        "texts_3, labels_3 = zip(*[(d['text'], d['label']) for d in data_3])\n",
        "texts_4, labels_4 = zip(*[(d['text'], d['label']) for d in data_4])\n",
        "texts_5, labels_5 = zip(*[(d['text'], d['label']) for d in data_5])\n",
        "\n",
        "texts_6, labels_6 = zip(*[(d['text'], d['label']) for d in data_6])\n",
        "texts_7, labels_7 = zip(*[(d['text'], d['label']) for d in data_7])\n",
        "texts_8, labels_8 = zip(*[(d['text'], d['label']) for d in data_8])\n",
        "\n",
        "x_train_0, x_test_0, y_train_0, y_test_0 = train_test_split(texts_0, labels_0, test_size = 0.25, random_state = 0, stratify = labels_0)\n",
        "x_train_1, x_test_1, y_train_1, y_test_1 = train_test_split(texts_1, labels_1, test_size = 0.25, random_state = 0, stratify = labels_1)\n",
        "x_train_2, x_test_2, y_train_2, y_test_2 = train_test_split(texts_2, labels_2, test_size = 0.25, random_state = 0, stratify = labels_2)\n",
        "x_train_3, x_test_3, y_train_3, y_test_3 = train_test_split(texts_3, labels_3, test_size = 0.25, random_state = 0, stratify = labels_3)\n",
        "x_train_4, x_test_4, y_train_4, y_test_4 = train_test_split(texts_4, labels_4, test_size = 0.25, random_state = 0, stratify = labels_4)\n",
        "x_train_5, x_test_5, y_train_5, y_test_5 = train_test_split(texts_5, labels_5, test_size = 0.25, random_state = 0, stratify = labels_5)\n",
        "\n",
        "x_train_6, x_test_6, y_train_6, y_test_6 = train_test_split(texts_6, labels_6, test_size = 0.25, random_state = 0, stratify = labels_6)\n",
        "x_train_7, x_test_7, y_train_7, y_test_7 = train_test_split(texts_7, labels_7, test_size = 0.25, random_state = 0, stratify = labels_7)\n",
        "x_train_8, x_test_8, y_train_8, y_test_8 = train_test_split(texts_8, labels_8, test_size = 0.25, random_state = 0, stratify = labels_8)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "a0c7074a",
      "metadata": {
        "id": "a0c7074a",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "905ca2a8-bddf-4175-9043-221797cd3fca"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            " What do you see going on here? Just look at this picture and tell me what you see going on. Well, you see the little boy stealing cookies. And the under girl is going to pick him up. I think she is. Because he is going to take a whole damn fall. And this girl... You have to find something here. Yeah, what else do you see? Well, cookie jars. They get smashed. But that little girl don't pick her up. And the woman is already getting it. Anything else? What's going on right here? What do you see going on over here? Anything? Yeah, she is in warm water and it's on all sides. And over there is cookies. Copyright. Is that all? Yeah, it's about all right. See? Okay.\n",
            "\n",
            "\n",
            " Here are two cups for coffee or something like that. Okay, thank you very much indeed.\n",
            "\n",
            "\n",
            "What do you see going on here? Just look at this picture and tell me what you see going on. Well, you see the little boy stealing cookies. And the under girl is going to pick him up. I think she is. Because he is going to take a whole damn fall. And this girl... You gotta find something here. Yeah, what else did you see? Well, cookie jars, they get smashed, but that little girl don't pick it. And the woman is already getting it. Anything else? What's going on right here? What do you see going on over here? Anything? Yeah, she's in warm water and it's on also. And over there is cookies. Copyright. Is that all? Yeah, it's about all right, see? Okay.\n",
            "\n",
            "\n",
            "Here are two cups for coffee or something like that. or something like that and... Okay. That's fine. Okay. Okay, thank you very much indeed.\n"
          ]
        }
      ],
      "source": [
        "print(x_train_0[22])\n",
        "print(\"\\n\")\n",
        "print(x_train_1[22])\n",
        "print(\"\\n\")\n",
        "print(x_train_2[22])\n",
        "print(\"\\n\")\n",
        "print(x_train_3[22])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "bb10ff99",
      "metadata": {
        "id": "bb10ff99",
        "outputId": "b4310e92-9c81-4a2f-b7fd-913039f62651",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Looking in indexes: https://pypi.org/simple, https://us-python.pkg.dev/colab-wheels/public/simple/\n",
            "Requirement already satisfied: transformers in /usr/local/lib/python3.8/dist-packages (4.26.1)\n",
            "Requirement already satisfied: pyyaml>=5.1 in /usr/local/lib/python3.8/dist-packages (from transformers) (5.4.1)\n",
            "Requirement already satisfied: requests in /usr/local/lib/python3.8/dist-packages (from transformers) (2.25.1)\n",
            "Requirement already satisfied: huggingface-hub<1.0,>=0.11.0 in /usr/local/lib/python3.8/dist-packages (from transformers) (0.12.1)\n",
            "Requirement already satisfied: tqdm>=4.27 in /usr/local/lib/python3.8/dist-packages (from transformers) (4.64.1)\n",
            "Requirement already satisfied: numpy>=1.17 in /usr/local/lib/python3.8/dist-packages (from transformers) (1.22.4)\n",
            "Requirement already satisfied: regex!=2019.12.17 in /usr/local/lib/python3.8/dist-packages (from transformers) (2022.6.2)\n",
            "Requirement already satisfied: filelock in /usr/local/lib/python3.8/dist-packages (from transformers) (3.9.0)\n",
            "Requirement already satisfied: tokenizers!=0.11.3,<0.14,>=0.11.1 in /usr/local/lib/python3.8/dist-packages (from transformers) (0.13.2)\n",
            "Requirement already satisfied: packaging>=20.0 in /usr/local/lib/python3.8/dist-packages (from transformers) (23.0)\n",
            "Requirement already satisfied: typing-extensions>=3.7.4.3 in /usr/local/lib/python3.8/dist-packages (from huggingface-hub<1.0,>=0.11.0->transformers) (4.5.0)\n",
            "Requirement already satisfied: urllib3<1.27,>=1.21.1 in /usr/local/lib/python3.8/dist-packages (from requests->transformers) (1.26.14)\n",
            "Requirement already satisfied: chardet<5,>=3.0.2 in /usr/local/lib/python3.8/dist-packages (from requests->transformers) (4.0.0)\n",
            "Requirement already satisfied: certifi>=2017.4.17 in /usr/local/lib/python3.8/dist-packages (from requests->transformers) (2022.12.7)\n",
            "Requirement already satisfied: idna<3,>=2.5 in /usr/local/lib/python3.8/dist-packages (from requests->transformers) (2.10)\n",
            "Looking in indexes: https://pypi.org/simple, https://us-python.pkg.dev/colab-wheels/public/simple/\n",
            "Requirement already satisfied: tensorflow_text in /usr/local/lib/python3.8/dist-packages (2.11.0)\n",
            "Requirement already satisfied: tensorflow-hub>=0.8.0 in /usr/local/lib/python3.8/dist-packages (from tensorflow_text) (0.12.0)\n",
            "Requirement already satisfied: tensorflow<2.12,>=2.11.0 in /usr/local/lib/python3.8/dist-packages (from tensorflow_text) (2.11.0)\n",
            "Requirement already satisfied: protobuf<3.20,>=3.9.2 in /usr/local/lib/python3.8/dist-packages (from tensorflow<2.12,>=2.11.0->tensorflow_text) (3.19.6)\n",
            "Requirement already satisfied: keras<2.12,>=2.11.0 in /usr/local/lib/python3.8/dist-packages (from tensorflow<2.12,>=2.11.0->tensorflow_text) (2.11.0)\n",
            "Requirement already satisfied: tensorflow-io-gcs-filesystem>=0.23.1 in /usr/local/lib/python3.8/dist-packages (from tensorflow<2.12,>=2.11.0->tensorflow_text) (0.31.0)\n",
            "Requirement already satisfied: astunparse>=1.6.0 in /usr/local/lib/python3.8/dist-packages (from tensorflow<2.12,>=2.11.0->tensorflow_text) (1.6.3)\n",
            "Requirement already satisfied: termcolor>=1.1.0 in /usr/local/lib/python3.8/dist-packages (from tensorflow<2.12,>=2.11.0->tensorflow_text) (2.2.0)\n",
            "Requirement already satisfied: gast<=0.4.0,>=0.2.1 in /usr/local/lib/python3.8/dist-packages (from tensorflow<2.12,>=2.11.0->tensorflow_text) (0.4.0)\n",
            "Requirement already satisfied: packaging in /usr/local/lib/python3.8/dist-packages (from tensorflow<2.12,>=2.11.0->tensorflow_text) (23.0)\n",
            "Requirement already satisfied: tensorflow-estimator<2.12,>=2.11.0 in /usr/local/lib/python3.8/dist-packages (from tensorflow<2.12,>=2.11.0->tensorflow_text) (2.11.0)\n",
            "Requirement already satisfied: absl-py>=1.0.0 in /usr/local/lib/python3.8/dist-packages (from tensorflow<2.12,>=2.11.0->tensorflow_text) (1.4.0)\n",
            "Requirement already satisfied: opt-einsum>=2.3.2 in /usr/local/lib/python3.8/dist-packages (from tensorflow<2.12,>=2.11.0->tensorflow_text) (3.3.0)\n",
            "Requirement already satisfied: typing-extensions>=3.6.6 in /usr/local/lib/python3.8/dist-packages (from tensorflow<2.12,>=2.11.0->tensorflow_text) (4.5.0)\n",
            "Requirement already satisfied: h5py>=2.9.0 in /usr/local/lib/python3.8/dist-packages (from tensorflow<2.12,>=2.11.0->tensorflow_text) (3.1.0)\n",
            "Requirement already satisfied: six>=1.12.0 in /usr/local/lib/python3.8/dist-packages (from tensorflow<2.12,>=2.11.0->tensorflow_text) (1.15.0)\n",
            "Requirement already satisfied: flatbuffers>=2.0 in /usr/local/lib/python3.8/dist-packages (from tensorflow<2.12,>=2.11.0->tensorflow_text) (23.1.21)\n",
            "Requirement already satisfied: libclang>=13.0.0 in /usr/local/lib/python3.8/dist-packages (from tensorflow<2.12,>=2.11.0->tensorflow_text) (15.0.6.1)\n",
            "Requirement already satisfied: google-pasta>=0.1.1 in /usr/local/lib/python3.8/dist-packages (from tensorflow<2.12,>=2.11.0->tensorflow_text) (0.2.0)\n",
            "Requirement already satisfied: tensorboard<2.12,>=2.11 in /usr/local/lib/python3.8/dist-packages (from tensorflow<2.12,>=2.11.0->tensorflow_text) (2.11.2)\n",
            "Requirement already satisfied: numpy>=1.20 in /usr/local/lib/python3.8/dist-packages (from tensorflow<2.12,>=2.11.0->tensorflow_text) (1.22.4)\n",
            "Requirement already satisfied: grpcio<2.0,>=1.24.3 in /usr/local/lib/python3.8/dist-packages (from tensorflow<2.12,>=2.11.0->tensorflow_text) (1.51.3)\n",
            "Requirement already satisfied: wrapt>=1.11.0 in /usr/local/lib/python3.8/dist-packages (from tensorflow<2.12,>=2.11.0->tensorflow_text) (1.15.0)\n",
            "Requirement already satisfied: setuptools in /usr/local/lib/python3.8/dist-packages (from tensorflow<2.12,>=2.11.0->tensorflow_text) (57.4.0)\n",
            "Requirement already satisfied: wheel<1.0,>=0.23.0 in /usr/local/lib/python3.8/dist-packages (from astunparse>=1.6.0->tensorflow<2.12,>=2.11.0->tensorflow_text) (0.38.4)\n",
            "Requirement already satisfied: google-auth<3,>=1.6.3 in /usr/local/lib/python3.8/dist-packages (from tensorboard<2.12,>=2.11->tensorflow<2.12,>=2.11.0->tensorflow_text) (2.16.1)\n",
            "Requirement already satisfied: tensorboard-data-server<0.7.0,>=0.6.0 in /usr/local/lib/python3.8/dist-packages (from tensorboard<2.12,>=2.11->tensorflow<2.12,>=2.11.0->tensorflow_text) (0.6.1)\n",
            "Requirement already satisfied: requests<3,>=2.21.0 in /usr/local/lib/python3.8/dist-packages (from tensorboard<2.12,>=2.11->tensorflow<2.12,>=2.11.0->tensorflow_text) (2.25.1)\n",
            "Requirement already satisfied: markdown>=2.6.8 in /usr/local/lib/python3.8/dist-packages (from tensorboard<2.12,>=2.11->tensorflow<2.12,>=2.11.0->tensorflow_text) (3.4.1)\n",
            "Requirement already satisfied: werkzeug>=1.0.1 in /usr/local/lib/python3.8/dist-packages (from tensorboard<2.12,>=2.11->tensorflow<2.12,>=2.11.0->tensorflow_text) (2.2.3)\n",
            "Requirement already satisfied: tensorboard-plugin-wit>=1.6.0 in /usr/local/lib/python3.8/dist-packages (from tensorboard<2.12,>=2.11->tensorflow<2.12,>=2.11.0->tensorflow_text) (1.8.1)\n",
            "Requirement already satisfied: google-auth-oauthlib<0.5,>=0.4.1 in /usr/local/lib/python3.8/dist-packages (from tensorboard<2.12,>=2.11->tensorflow<2.12,>=2.11.0->tensorflow_text) (0.4.6)\n",
            "Requirement already satisfied: rsa<5,>=3.1.4 in /usr/local/lib/python3.8/dist-packages (from google-auth<3,>=1.6.3->tensorboard<2.12,>=2.11->tensorflow<2.12,>=2.11.0->tensorflow_text) (4.9)\n",
            "Requirement already satisfied: cachetools<6.0,>=2.0.0 in /usr/local/lib/python3.8/dist-packages (from google-auth<3,>=1.6.3->tensorboard<2.12,>=2.11->tensorflow<2.12,>=2.11.0->tensorflow_text) (5.3.0)\n",
            "Requirement already satisfied: pyasn1-modules>=0.2.1 in /usr/local/lib/python3.8/dist-packages (from google-auth<3,>=1.6.3->tensorboard<2.12,>=2.11->tensorflow<2.12,>=2.11.0->tensorflow_text) (0.2.8)\n",
            "Requirement already satisfied: requests-oauthlib>=0.7.0 in /usr/local/lib/python3.8/dist-packages (from google-auth-oauthlib<0.5,>=0.4.1->tensorboard<2.12,>=2.11->tensorflow<2.12,>=2.11.0->tensorflow_text) (1.3.1)\n",
            "Requirement already satisfied: importlib-metadata>=4.4 in /usr/local/lib/python3.8/dist-packages (from markdown>=2.6.8->tensorboard<2.12,>=2.11->tensorflow<2.12,>=2.11.0->tensorflow_text) (6.0.0)\n",
            "Requirement already satisfied: urllib3<1.27,>=1.21.1 in /usr/local/lib/python3.8/dist-packages (from requests<3,>=2.21.0->tensorboard<2.12,>=2.11->tensorflow<2.12,>=2.11.0->tensorflow_text) (1.26.14)\n",
            "Requirement already satisfied: idna<3,>=2.5 in /usr/local/lib/python3.8/dist-packages (from requests<3,>=2.21.0->tensorboard<2.12,>=2.11->tensorflow<2.12,>=2.11.0->tensorflow_text) (2.10)\n",
            "Requirement already satisfied: certifi>=2017.4.17 in /usr/local/lib/python3.8/dist-packages (from requests<3,>=2.21.0->tensorboard<2.12,>=2.11->tensorflow<2.12,>=2.11.0->tensorflow_text) (2022.12.7)\n",
            "Requirement already satisfied: chardet<5,>=3.0.2 in /usr/local/lib/python3.8/dist-packages (from requests<3,>=2.21.0->tensorboard<2.12,>=2.11->tensorflow<2.12,>=2.11.0->tensorflow_text) (4.0.0)\n",
            "Requirement already satisfied: MarkupSafe>=2.1.1 in /usr/local/lib/python3.8/dist-packages (from werkzeug>=1.0.1->tensorboard<2.12,>=2.11->tensorflow<2.12,>=2.11.0->tensorflow_text) (2.1.2)\n",
            "Requirement already satisfied: zipp>=0.5 in /usr/local/lib/python3.8/dist-packages (from importlib-metadata>=4.4->markdown>=2.6.8->tensorboard<2.12,>=2.11->tensorflow<2.12,>=2.11.0->tensorflow_text) (3.15.0)\n",
            "Requirement already satisfied: pyasn1<0.5.0,>=0.4.6 in /usr/local/lib/python3.8/dist-packages (from pyasn1-modules>=0.2.1->google-auth<3,>=1.6.3->tensorboard<2.12,>=2.11->tensorflow<2.12,>=2.11.0->tensorflow_text) (0.4.8)\n",
            "Requirement already satisfied: oauthlib>=3.0.0 in /usr/local/lib/python3.8/dist-packages (from requests-oauthlib>=0.7.0->google-auth-oauthlib<0.5,>=0.4.1->tensorboard<2.12,>=2.11->tensorflow<2.12,>=2.11.0->tensorflow_text) (3.2.2)\n"
          ]
        }
      ],
      "source": [
        "#!pip install transformers\n",
        "#!pip install tensorflow_hub\n",
        "!pip install tensorflow_text\n",
        "#from transformers import BertTokenizer, TFBertModel\n",
        "\n",
        "import tensorflow as tf\n",
        "import tensorflow_hub as hub\n",
        "import tensorflow_text as text\n",
        "#tokenizer = BertTokenizer.from_pretrained('bert-base-uncased')\n",
        "#bert_model = TFBertModel.from_pretrained(\"bert-base-uncased\")\n",
        "# bert_preprocess = hub.KerasLayer(\"https://tfhub.dev/tensorflow/bert_en_uncased_preprocess/3\")\n",
        "# bert_encoder = hub.KerasLayer(\"https://tfhub.dev/tensorflow/bert_en_uncased_L-12_H-768_A-12/4\")"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "356a85da",
      "metadata": {
        "scrolled": true,
        "id": "356a85da"
      },
      "outputs": [],
      "source": [
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "a5bdd5a2",
      "metadata": {
        "id": "a5bdd5a2"
      },
      "outputs": [],
      "source": [
        "# # Bert layers\n",
        "# text_input = tf.keras.layers.Input(shape=(), dtype=tf.string, name='text')\n",
        "# preprocessed_text = bert_preprocess(text_input)\n",
        "# outputs = bert_encoder(preprocessed_text)\n",
        "\n",
        "# # Neural network layers\n",
        "# l = tf.keras.layers.Dropout(0.01, name=\"dropout\")(outputs['pooled_output'])\n",
        "# l2 = tf.keras.layers.Dense(1, activation='sigmoid', name=\"output\")(l)\n",
        "\n",
        "# # Use inputs and outputs to construct a final model\n",
        "# model = tf.keras.Model(inputs=[text_input], outputs = [l2])\n",
        "\n",
        "def build_classifier_model():\n",
        "  bert_preprocess = hub.KerasLayer(\"https://tfhub.dev/tensorflow/bert_en_uncased_preprocess/3\", name='preprocessing')\n",
        "  bert_encoder = hub.KerasLayer(\"https://tfhub.dev/tensorflow/bert_en_uncased_L-12_H-768_A-12/4\", trainable=True, name='BERT_encoder')\n",
        "  text_input = tf.keras.layers.Input(shape=(), dtype=tf.string, name='text')\n",
        "  preprocessing_layer = bert_preprocess\n",
        "  encoder_inputs = preprocessing_layer(text_input)\n",
        "  encoder = bert_encoder\n",
        "  outputs = encoder(encoder_inputs)\n",
        "  net = outputs['pooled_output']\n",
        "  \n",
        "  net = tf.keras.layers.Dropout(0.1)(net)\n",
        "  net = tf.keras.layers.Dense(1, activation=\"sigmoid\", name='classifier')(net)\n",
        "  return tf.keras.Model(text_input, net)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "f50b669e",
      "metadata": {
        "id": "f50b669e"
      },
      "outputs": [],
      "source": [
        "#model_0 = build_classifier_model()\n",
        "model_1 = build_classifier_model()\n",
        "#model_2 = build_classifier_model()\n",
        "#model_3 = build_classifier_model()\n",
        "#model_4 = build_classifier_model()\n",
        "#model_5 = build_classifier_model()\n",
        "model_6 = build_classifier_model()\n",
        "#model_5 = build_classifier_model()\n",
        "model_1.summary()"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "!pip install -q tf-models-official\n",
        "\n",
        "from official.nlp import optimization\n",
        "\n",
        "# loss = tf.keras.losses.BinaryCrossentropy(from_logits=True)\n",
        "metrics = tf.metrics.BinaryAccuracy()\n",
        "\n",
        "epochs = 10\n",
        "steps_per_epoch = 200\n",
        "num_train_steps = steps_per_epoch * epochs\n",
        "num_warmup_steps = int(0.1*num_train_steps)\n",
        "init_lr = 3e-5\n",
        "# METRICS = [\n",
        "#       tf.keras.metrics.Accuracy(name='accuracy'),\n",
        "#       tf.keras.metrics.Precision(name='precision'),\n",
        "#       tf.keras.metrics.Recall(name='recall')\n",
        "# ]\n",
        "optimizer = optimization.create_optimizer(init_lr=init_lr,\n",
        "                                          num_train_steps=num_train_steps,\n",
        "                                          num_warmup_steps=num_warmup_steps,\n",
        "                                          optimizer_type='adamw')\n"
      ],
      "metadata": {
        "id": "wmli2cMtNS-D"
      },
      "id": "wmli2cMtNS-D",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "#training models 1\n",
        "#model_1.reset_states()\n",
        "#del model_1\n",
        "model_0 = build_classifier_model()\n",
        "model_0.compile(optimizer=optimizer,\n",
        "              loss='binary_crossentropy',\n",
        "              metrics=metrics)\n",
        "model_0.fit(x_train_0, y_train_0, epochs=epochs)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "AA56eOu1Nb_K",
        "outputId": "8d64d882-4bb7-4e29-d475-0e3209d4eedb"
      },
      "id": "AA56eOu1Nb_K",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/20\n",
            "4/4 [==============================] - 29s 865ms/step - loss: 0.7204 - binary_accuracy: 0.4516\n",
            "Epoch 2/20\n",
            "4/4 [==============================] - 4s 913ms/step - loss: 0.7273 - binary_accuracy: 0.4758\n",
            "Epoch 3/20\n",
            "4/4 [==============================] - 4s 853ms/step - loss: 0.6975 - binary_accuracy: 0.5484\n",
            "Epoch 4/20\n",
            "4/4 [==============================] - 3s 861ms/step - loss: 0.7044 - binary_accuracy: 0.4839\n",
            "Epoch 5/20\n",
            "4/4 [==============================] - 4s 921ms/step - loss: 0.6838 - binary_accuracy: 0.5403\n",
            "Epoch 6/20\n",
            "4/4 [==============================] - 4s 879ms/step - loss: 0.6974 - binary_accuracy: 0.5161\n",
            "Epoch 7/20\n",
            "4/4 [==============================] - 4s 871ms/step - loss: 0.6651 - binary_accuracy: 0.6129\n",
            "Epoch 8/20\n",
            "4/4 [==============================] - 4s 921ms/step - loss: 0.6419 - binary_accuracy: 0.6694\n",
            "Epoch 9/20\n",
            "4/4 [==============================] - 4s 937ms/step - loss: 0.6254 - binary_accuracy: 0.7097\n",
            "Epoch 10/20\n",
            "4/4 [==============================] - 4s 884ms/step - loss: 0.6306 - binary_accuracy: 0.6613\n",
            "Epoch 11/20\n",
            "4/4 [==============================] - 4s 887ms/step - loss: 0.6334 - binary_accuracy: 0.6452\n",
            "Epoch 12/20\n",
            "4/4 [==============================] - 4s 943ms/step - loss: 0.5781 - binary_accuracy: 0.7823\n",
            "Epoch 13/20\n",
            "4/4 [==============================] - 4s 879ms/step - loss: 0.5727 - binary_accuracy: 0.7177\n",
            "Epoch 14/20\n",
            "4/4 [==============================] - 4s 907ms/step - loss: 0.5511 - binary_accuracy: 0.7661\n",
            "Epoch 15/20\n",
            "4/4 [==============================] - 4s 933ms/step - loss: 0.4904 - binary_accuracy: 0.8306\n",
            "Epoch 16/20\n",
            "4/4 [==============================] - 4s 871ms/step - loss: 0.4928 - binary_accuracy: 0.8065\n",
            "Epoch 17/20\n",
            "4/4 [==============================] - 4s 869ms/step - loss: 0.4385 - binary_accuracy: 0.8306\n",
            "Epoch 18/20\n",
            "4/4 [==============================] - 4s 918ms/step - loss: 0.3979 - binary_accuracy: 0.8548\n",
            "Epoch 19/20\n",
            "4/4 [==============================] - 4s 863ms/step - loss: 0.3739 - binary_accuracy: 0.8790\n",
            "Epoch 20/20\n",
            "4/4 [==============================] - 4s 875ms/step - loss: 0.3415 - binary_accuracy: 0.8871\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<keras.callbacks.History at 0x7f317545f400>"
            ]
          },
          "metadata": {},
          "execution_count": 37
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "#train avec tout test avec tout\n",
        "model_0.evaluate(x_test_0, y_test_0)\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ZFRZZ2FJSt6q",
        "outputId": "9a0893f5-0ee7-42f4-f9b2-db06fe19b611"
      },
      "id": "ZFRZZ2FJSt6q",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "2/2 [==============================] - 2s 107ms/step - loss: 0.6160 - binary_accuracy: 0.7143\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[0.615959882736206, 0.7142857313156128]"
            ]
          },
          "metadata": {},
          "execution_count": 38
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "#train avec tout test avec debut\n",
        "model_0.evaluate(x_test_6, y_test_6)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "33vClhg5iAh_",
        "outputId": "7b2ff7c3-68f7-492c-a5e6-fa2298862a31"
      },
      "id": "33vClhg5iAh_",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "2/2 [==============================] - 0s 107ms/step - loss: 0.4587 - binary_accuracy: 0.7619\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[0.45868268609046936, 0.761904776096344]"
            ]
          },
          "metadata": {},
          "execution_count": 39
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "def get_matrice_cm(model, x_test, y_test ):\n",
        "  y_predicted = model.predict(x_test)\n",
        "  y_predicted = np.round(y_predicted).flatten()\n",
        "  cm = confusion_matrix(y_test, y_predicted)\n",
        "  return cm\n",
        "\n",
        "def plot_2_cm(cm1, cm2, title1, title2):\n",
        "  # Créer une grille de subplots pour afficher les matrices de confusion\n",
        "  fig, axes = plt.subplots(nrows=1, ncols=2, figsize=(20, 4))\n",
        "\n",
        "  # Afficher chaque matrice de confusion dans un subplot différent\n",
        "  sn.heatmap(cm1, ax=axes[0], annot=True, cmap=\"Blues\", fmt='g')\n",
        "  sn.heatmap(cm2, ax=axes[1], annot=True, cmap=\"Blues\", fmt='g')\n",
        "  # sn.heatmap(cm_4, ax=axes[2], annot=True, cmap=\"Blues\", fmt='g')\n",
        "\n",
        "  axes[0].set_xlabel('Prédiction')\n",
        "  axes[0].set_ylabel('Réel')\n",
        "  axes[0].set_title(title1)\n",
        "\n",
        "  axes[1].set_xlabel('Prédiction')\n",
        "  axes[1].set_ylabel('Réel')\n",
        "  axes[1].set_title(title2)\n"
      ],
      "metadata": {
        "id": "eg72EkmKTPHJ"
      },
      "id": "eg72EkmKTPHJ",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "cm_1 = get_matrice_cm(model_0, x_test_0, y_test_0)\n",
        "cm_1_b = get_matrice_cm(model_0, x_test_6, y_test_6)"
      ],
      "metadata": {
        "id": "VEIB8vImyfcZ"
      },
      "id": "VEIB8vImyfcZ",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [],
      "metadata": {
        "id": "hluSHKHW-eZ9"
      },
      "id": "hluSHKHW-eZ9",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# # Créer une grille de subplots pour afficher les matrices de confusion\n",
        "# fig, axes = plt.subplots(nrows=1, ncols=2, figsize=(20, 4))\n",
        "\n",
        "# # Afficher chaque matrice de confusion dans un subplot différent\n",
        "# sn.heatmap(cm_1, ax=axes[0], annot=True, cmap=\"Blues\", fmt='g')\n",
        "# sn.heatmap(cm_1_b, ax=axes[1], annot=True, cmap=\"Blues\", fmt='g')\n",
        "# # sn.heatmap(cm_4, ax=axes[2], annot=True, cmap=\"Blues\", fmt='g')\n",
        "\n",
        "# axes[0].set_xlabel('Prédiction')\n",
        "# axes[0].set_ylabel('Réel')\n",
        "# axes[0].set_title('Matrice de confusion modele_1 : données de whisper avec test sur whisper')\n",
        "\n",
        "# axes[1].set_xlabel('Prédiction')\n",
        "# axes[1].set_ylabel('Réel')\n",
        "# axes[1].set_title('Matrice de confusion modele_1 : données de whisper avec test sur whisper_debut')\n",
        "\n",
        "plot_2_cm(cm_1, cm_1_b, 'Matrice de confusion modele_1 : données de whisper avec test sur whisper', 'Matrice de confusion modele_1 : données de whisper avec test sur whisper_debut')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 312
        },
        "id": "cCqoVEw2tNNu",
        "outputId": "5c90aeb5-7067-455f-f1a3-c5d84697616a"
      },
      "id": "cCqoVEw2tNNu",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Text(0.5, 1.0, 'Matrice de confusion modele_1 : données de whisper avec test sur whisper_debut')"
            ]
          },
          "metadata": {},
          "execution_count": 46
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 1440x288 with 4 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "#training models 1 mais avec les données de seg_debut\n",
        "model_1.reset_states()\n",
        "del model_1\n",
        "\n",
        "model_1_bis = build_classifier_model()\n",
        "model_1_bis.compile(optimizer=optimizer,\n",
        "              loss='binary_crossentropy',\n",
        "              metrics=metrics)\n",
        "model_1_bis.fit(x_train_6, y_train_6, epochs=epochs)"
      ],
      "metadata": {
        "id": "etkZN8qdTenw",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "220a6bf3-1cc1-42b8-8f9f-af83fa18b943"
      },
      "id": "etkZN8qdTenw",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/20\n",
            "4/4 [==============================] - 23s 900ms/step - loss: 0.7108 - binary_accuracy: 0.5542\n",
            "Epoch 2/20\n",
            "4/4 [==============================] - 3s 868ms/step - loss: 0.6829 - binary_accuracy: 0.5000\n",
            "Epoch 3/20\n",
            "4/4 [==============================] - 4s 872ms/step - loss: 0.6295 - binary_accuracy: 0.6613\n",
            "Epoch 4/20\n",
            "4/4 [==============================] - 4s 906ms/step - loss: 0.5724 - binary_accuracy: 0.7742\n",
            "Epoch 5/20\n",
            "4/4 [==============================] - 4s 928ms/step - loss: 0.5049 - binary_accuracy: 0.8306\n",
            "Epoch 6/20\n",
            "4/4 [==============================] - 4s 917ms/step - loss: 0.4369 - binary_accuracy: 0.8629\n",
            "Epoch 7/20\n",
            "4/4 [==============================] - 3s 867ms/step - loss: 0.3534 - binary_accuracy: 0.8710\n",
            "Epoch 8/20\n",
            "4/4 [==============================] - 4s 921ms/step - loss: 0.2718 - binary_accuracy: 0.9274\n",
            "Epoch 9/20\n",
            "4/4 [==============================] - 4s 855ms/step - loss: 0.2143 - binary_accuracy: 0.9597\n",
            "Epoch 10/20\n",
            "4/4 [==============================] - 3s 857ms/step - loss: 0.1526 - binary_accuracy: 0.9758\n",
            "Epoch 11/20\n",
            "4/4 [==============================] - 4s 971ms/step - loss: 0.1158 - binary_accuracy: 0.9839\n",
            "Epoch 12/20\n",
            "4/4 [==============================] - 4s 923ms/step - loss: 0.0754 - binary_accuracy: 0.9919\n",
            "Epoch 13/20\n",
            "4/4 [==============================] - 3s 849ms/step - loss: 0.0477 - binary_accuracy: 1.0000\n",
            "Epoch 14/20\n",
            "4/4 [==============================] - 3s 851ms/step - loss: 0.0257 - binary_accuracy: 1.0000\n",
            "Epoch 15/20\n",
            "4/4 [==============================] - 4s 901ms/step - loss: 0.0169 - binary_accuracy: 1.0000\n",
            "Epoch 16/20\n",
            "4/4 [==============================] - 3s 850ms/step - loss: 0.0111 - binary_accuracy: 1.0000\n",
            "Epoch 17/20\n",
            "4/4 [==============================] - 4s 935ms/step - loss: 0.0072 - binary_accuracy: 1.0000\n",
            "Epoch 18/20\n",
            "4/4 [==============================] - 4s 885ms/step - loss: 0.0047 - binary_accuracy: 1.0000\n",
            "Epoch 19/20\n",
            "4/4 [==============================] - 4s 918ms/step - loss: 0.0038 - binary_accuracy: 1.0000\n",
            "Epoch 20/20\n",
            "4/4 [==============================] - 3s 860ms/step - loss: 0.0031 - binary_accuracy: 1.0000\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<keras.callbacks.History at 0x7f3071fb1250>"
            ]
          },
          "metadata": {},
          "execution_count": 45
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "#train avec début test avec tout\n",
        "cm_2 = get_matrice_cm(model_1_bis, x_test_0, y_test_0)\n",
        "model_1_bis.evaluate(x_test_0, y_test_0)"
      ],
      "metadata": {
        "id": "E5g6bZGfTg4s",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "bb0e82f0-336d-4dbc-9f81-bdce06355a5a"
      },
      "id": "E5g6bZGfTg4s",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "2/2 [==============================] - 1s 363ms/step\n",
            "2/2 [==============================] - 1s 105ms/step - loss: 0.4390 - binary_accuracy: 0.9048\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[0.4390411376953125, 0.9047619104385376]"
            ]
          },
          "metadata": {},
          "execution_count": 57
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "#train avec début et test avec début\n",
        "cm_2_bis = get_matrice_cm(model_1_bis, x_test_6, y_test_6)\n",
        "model_1_bis.evaluate(x_test_6, y_test_6)"
      ],
      "metadata": {
        "id": "T4oIPlSaTnWC",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "ff765af6-f3fe-4d32-90a5-d568bc33e4df"
      },
      "id": "T4oIPlSaTnWC",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "2/2 [==============================] - 0s 277ms/step\n",
            "2/2 [==============================] - 0s 103ms/step - loss: 0.8104 - binary_accuracy: 0.7857\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[0.8103622198104858, 0.7857142686843872]"
            ]
          },
          "metadata": {},
          "execution_count": 58
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Créer une grille de subplots pour afficher les matrices de confusion\n",
        "fig, axes = plt.subplots(nrows=1, ncols=2, figsize=(25, 4))\n",
        "\n",
        "# Afficher chaque matrice de confusion dans un subplot différent\n",
        "sn.heatmap(cm_2, ax=axes[0], annot=True, cmap=\"Blues\", fmt='g')\n",
        "sn.heatmap(cm_2_bis, ax=axes[1], annot=True, cmap=\"Blues\", fmt='g')\n",
        "# sn.heatmap(cm_4, ax=axes[2], annot=True, cmap=\"Blues\", fmt='g')\n",
        "\n",
        "axes[0].set_xlabel('Prédiction')\n",
        "axes[0].set_ylabel('Réel')\n",
        "axes[0].set_title('Matrice de confusion modele_1_bis : train avec données de whisper_debut avec test sur whisper')\n",
        "\n",
        "axes[1].set_xlabel('Prédiction')\n",
        "axes[1].set_ylabel('Réel')\n",
        "axes[1].set_title('Matrice de confusion modele_1_bis : train avec données de whisper_debut avec test sur whisper_debut')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 312
        },
        "id": "Go-j3HXKxC8u",
        "outputId": "c8d6eab0-fe11-4a4e-a3f4-0558c1b5e499"
      },
      "id": "Go-j3HXKxC8u",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Text(0.5, 1.0, 'Matrice de confusion modele_1_bis : train avec données de whisper_debut avec test sur whisper_debut')"
            ]
          },
          "metadata": {},
          "execution_count": 60
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 1800x288 with 4 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "#ici\n",
        "###same avec whisperX\n",
        "\n",
        "#training models 2 mais avec les données de seg_debut\n",
        "model_2 = build_classifier_model()\n",
        "model_2.compile(optimizer=optimizer,\n",
        "              loss='binary_crossentropy',\n",
        "              metrics=metrics)\n",
        "model_2.fit(x_train_2, y_train_2, epochs=epochs)"
      ],
      "metadata": {
        "id": "0I-ah1iqTszA",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "f7867fdd-2789-492c-9bf6-0cc4462a4ab0"
      },
      "id": "0I-ah1iqTszA",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/20\n",
            "4/4 [==============================] - 24s 910ms/step - loss: 0.7665 - binary_accuracy: 0.5242\n",
            "Epoch 2/20\n",
            "4/4 [==============================] - 4s 875ms/step - loss: 0.7672 - binary_accuracy: 0.5242\n",
            "Epoch 3/20\n",
            "4/4 [==============================] - 4s 875ms/step - loss: 0.7721 - binary_accuracy: 0.5242\n",
            "Epoch 4/20\n",
            "4/4 [==============================] - 4s 907ms/step - loss: 0.7292 - binary_accuracy: 0.5403\n",
            "Epoch 5/20\n",
            "4/4 [==============================] - 4s 943ms/step - loss: 0.7244 - binary_accuracy: 0.5565\n",
            "Epoch 6/20\n",
            "4/4 [==============================] - 4s 886ms/step - loss: 0.7222 - binary_accuracy: 0.5242\n",
            "Epoch 7/20\n",
            "4/4 [==============================] - 4s 891ms/step - loss: 0.6616 - binary_accuracy: 0.5645\n",
            "Epoch 8/20\n",
            "4/4 [==============================] - 4s 926ms/step - loss: 0.6687 - binary_accuracy: 0.5806\n",
            "Epoch 9/20\n",
            "4/4 [==============================] - 4s 935ms/step - loss: 0.6581 - binary_accuracy: 0.6048\n",
            "Epoch 10/20\n",
            "4/4 [==============================] - 4s 876ms/step - loss: 0.6319 - binary_accuracy: 0.6774\n",
            "Epoch 11/20\n",
            "4/4 [==============================] - 4s 891ms/step - loss: 0.6035 - binary_accuracy: 0.6935\n",
            "Epoch 12/20\n",
            "4/4 [==============================] - 4s 929ms/step - loss: 0.5889 - binary_accuracy: 0.7016\n",
            "Epoch 13/20\n",
            "4/4 [==============================] - 4s 950ms/step - loss: 0.5498 - binary_accuracy: 0.7903\n",
            "Epoch 14/20\n",
            "4/4 [==============================] - 3s 868ms/step - loss: 0.5250 - binary_accuracy: 0.7661\n",
            "Epoch 15/20\n",
            "4/4 [==============================] - 4s 931ms/step - loss: 0.4965 - binary_accuracy: 0.8145\n",
            "Epoch 16/20\n",
            "4/4 [==============================] - 4s 870ms/step - loss: 0.4915 - binary_accuracy: 0.8226\n",
            "Epoch 17/20\n",
            "4/4 [==============================] - 4s 871ms/step - loss: 0.4333 - binary_accuracy: 0.8629\n",
            "Epoch 18/20\n",
            "4/4 [==============================] - 4s 912ms/step - loss: 0.3997 - binary_accuracy: 0.8629\n",
            "Epoch 19/20\n",
            "4/4 [==============================] - 4s 910ms/step - loss: 0.3667 - binary_accuracy: 0.8871\n",
            "Epoch 20/20\n",
            "4/4 [==============================] - 4s 874ms/step - loss: 0.3099 - binary_accuracy: 0.9113\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<keras.callbacks.History at 0x7fe3c0f04b50>"
            ]
          },
          "metadata": {},
          "execution_count": 43
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "#train avec tout test avec tout\n",
        "cm_3 = get_matrice_cm(model_2, x_test_2, y_test_2)\n",
        "model_2.evaluate(x_test_2, y_test_2)\n"
      ],
      "metadata": {
        "id": "TMfESJEkTvcl",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "94488298-0feb-4a9f-8ca8-1905314ba45b"
      },
      "id": "TMfESJEkTvcl",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "2/2 [==============================] - 1s 327ms/step\n",
            "2/2 [==============================] - 1s 105ms/step - loss: 0.5470 - binary_accuracy: 0.7381\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[0.5470249652862549, 0.738095223903656]"
            ]
          },
          "metadata": {},
          "execution_count": 44
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "#train avec tout test avec debut\n",
        "cm_3_bis = get_matrice_cm(model_2, x_test_7, y_test_7)\n",
        "model_2.evaluate(x_test_7, y_test_7)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "_wGVLr85NoQl",
        "outputId": "02680440-f34a-428d-a4c5-120ce37e03a5"
      },
      "id": "_wGVLr85NoQl",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "2/2 [==============================] - 0s 304ms/step\n",
            "2/2 [==============================] - 1s 114ms/step - loss: 0.5715 - binary_accuracy: 0.6905\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[0.5715210437774658, 0.6904761791229248]"
            ]
          },
          "metadata": {},
          "execution_count": 45
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Créer une grille de subplots pour afficher les matrices de confusion\n",
        "# fig, axes = plt.subplots(nrows=1, ncols=2, figsize=(25, 4))\n",
        "\n",
        "# # Afficher chaque matrice de confusion dans un subplot différent\n",
        "# sn.heatmap(cm_3, ax=axes[0], annot=True, cmap=\"Blues\", fmt='g')\n",
        "# sn.heatmap(cm_3_bis, ax=axes[1], annot=True, cmap=\"Blues\", fmt='g')\n",
        "# # sn.heatmap(cm_4, ax=axes[2], annot=True, cmap=\"Blues\", fmt='g')\n",
        "\n",
        "# axes[0].set_xlabel('Prédiction')\n",
        "# axes[0].set_ylabel('Réel')\n",
        "# axes[0].set_title('Matrice de confusion modele_2_bis : train avec données de whisperX avec test sur whisperX')\n",
        "\n",
        "# axes[1].set_xlabel('Prédiction')\n",
        "# axes[1].set_ylabel('Réel')\n",
        "# axes[1].set_title('Matrice de confusion modele_2_bis : train avec données de whisperX avec test sur whisperX_debut')\n",
        "\n",
        "plot_2_cm(cm_3, cm_3_bis, 'Matrice de confusion modele_2_bis : train avec données de whisperX avec test sur whisperX', 'Matrice de confusion modele_2_bis : train avec données de whisperX avec test sur whisperX_debut')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 312
        },
        "id": "Vu2HesnYsSHB",
        "outputId": "2ee1a7d2-16b7-44dd-90d4-fe324d83a230"
      },
      "id": "Vu2HesnYsSHB",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Text(0.5, 1.0, 'Matrice de confusion modele_2_bis : train avec données de whisperX avec test sur whisperX_debut')"
            ]
          },
          "metadata": {},
          "execution_count": 46
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 1800x288 with 4 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "#train avec whisperX_debut\n",
        "\n",
        "model_7 = build_classifier_model()\n",
        "model_7.compile(optimizer=optimizer,\n",
        "              loss='binary_crossentropy',\n",
        "              metrics=metrics)\n",
        "model_7.fit(x_train_7, y_train_7, epochs=20)"
      ],
      "metadata": {
        "id": "HPqwboZhsS68",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "0fbdb62a-d5d9-44b5-f151-845e5fff0827"
      },
      "id": "HPqwboZhsS68",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "WARNING:tensorflow:Please fix your imports. Module tensorflow.python.training.tracking.data_structures has been moved to tensorflow.python.trackable.data_structures. The old module will be deleted in version 2.11.\n",
            "WARNING:tensorflow:From /usr/local/lib/python3.8/dist-packages/tensorflow/python/autograph/pyct/static_analysis/liveness.py:83: Analyzer.lamba_check (from tensorflow.python.autograph.pyct.static_analysis.liveness) is deprecated and will be removed after 2023-09-23.\n",
            "Instructions for updating:\n",
            "Lambda fuctions will be no more assumed to be used in the statement where they are used, or at least in the same block. https://github.com/tensorflow/tensorflow/issues/56089\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/20\n",
            "4/4 [==============================] - 252s 56s/step - loss: 0.6967 - binary_accuracy: 0.5000\n",
            "Epoch 2/20\n",
            "4/4 [==============================] - 213s 53s/step - loss: 0.6746 - binary_accuracy: 0.5726\n",
            "Epoch 3/20\n",
            "4/4 [==============================] - 215s 53s/step - loss: 0.6955 - binary_accuracy: 0.5403\n",
            "Epoch 4/20\n",
            "4/4 [==============================] - 211s 52s/step - loss: 0.7127 - binary_accuracy: 0.5161\n",
            "Epoch 5/20\n",
            "4/4 [==============================] - 211s 52s/step - loss: 0.6702 - binary_accuracy: 0.6048\n",
            "Epoch 6/20\n",
            "4/4 [==============================] - 211s 52s/step - loss: 0.6534 - binary_accuracy: 0.7097\n",
            "Epoch 7/20\n",
            "4/4 [==============================] - 210s 52s/step - loss: 0.6491 - binary_accuracy: 0.6532\n",
            "Epoch 8/20\n",
            "4/4 [==============================] - 215s 54s/step - loss: 0.6389 - binary_accuracy: 0.6290\n",
            "Epoch 9/20\n",
            "4/4 [==============================] - 212s 52s/step - loss: 0.5966 - binary_accuracy: 0.6694\n",
            "Epoch 10/20\n",
            "4/4 [==============================] - 214s 53s/step - loss: 0.5378 - binary_accuracy: 0.7500\n",
            "Epoch 11/20\n",
            "4/4 [==============================] - 213s 53s/step - loss: 0.4532 - binary_accuracy: 0.8226\n",
            "Epoch 12/20\n",
            "4/4 [==============================] - 216s 54s/step - loss: 0.4463 - binary_accuracy: 0.8145\n",
            "Epoch 13/20\n",
            "4/4 [==============================] - 214s 53s/step - loss: 0.4203 - binary_accuracy: 0.8468\n",
            "Epoch 14/20\n",
            "4/4 [==============================] - 214s 53s/step - loss: 0.3602 - binary_accuracy: 0.8548\n",
            "Epoch 15/20\n",
            "4/4 [==============================] - 214s 53s/step - loss: 0.3244 - binary_accuracy: 0.9032\n",
            "Epoch 16/20\n",
            "4/4 [==============================] - 213s 53s/step - loss: 0.2745 - binary_accuracy: 0.9113\n",
            "Epoch 17/20\n",
            "4/4 [==============================] - 216s 53s/step - loss: 0.2755 - binary_accuracy: 0.9032\n",
            "Epoch 18/20\n",
            "4/4 [==============================] - 213s 53s/step - loss: 0.2075 - binary_accuracy: 0.9677\n",
            "Epoch 19/20\n",
            "4/4 [==============================] - 221s 54s/step - loss: 0.1601 - binary_accuracy: 0.9677\n",
            "Epoch 20/20\n",
            "4/4 [==============================] - 216s 53s/step - loss: 0.1517 - binary_accuracy: 0.9677\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<keras.callbacks.History at 0x7f5970d31760>"
            ]
          },
          "metadata": {},
          "execution_count": 16
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "#test avec debut_7\n",
        "cm_7 = get_matrice_cm(model_7, x_test_7, y_test_7)\n",
        "model_7.evaluate(x_test_7, y_test_7)"
      ],
      "metadata": {
        "id": "aw1V23xaN6_i",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "ac6ecf4b-7bcc-4c81-85e6-5cf24d9a9743"
      },
      "id": "aw1V23xaN6_i",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "2/2 [==============================] - 22s 4s/step\n",
            "2/2 [==============================] - 23s 7s/step - loss: 0.6074 - binary_accuracy: 0.6905\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[0.6073777079582214, 0.6904761791229248]"
            ]
          },
          "metadata": {},
          "execution_count": 20
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "cm_7_bis = get_matrice_cm(model_7, x_test_2, y_test_2)\n",
        "model_7.evaluate(x_test_2, y_test_2)\n"
      ],
      "metadata": {
        "id": "XOXibT_CN7a0",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "6cb51b65-8127-4d9c-e472-94c6ec9083a2"
      },
      "id": "XOXibT_CN7a0",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "2/2 [==============================] - 21s 4s/step\n",
            "2/2 [==============================] - 24s 7s/step - loss: 0.3550 - binary_accuracy: 0.8571\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[0.35501670837402344, 0.8571428656578064]"
            ]
          },
          "metadata": {},
          "execution_count": 21
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "plot_2_cm(cm_7, cm_7_bis, 'Matrice de confusion modele_7: train avec données de whisperX_debut avec test sur whisperX_debut'\\\n",
        "          , 'Matrice de confusion modele_7_bis: train avec données de whisperX_debut avec test sur whisperX')\n"
      ],
      "metadata": {
        "id": "mhLCYlVhOBvB",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "outputId": "3851f6f1-2d87-4ff0-98b0-97be56ef220e"
      },
      "id": "mhLCYlVhOBvB",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 1440x288 with 4 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [],
      "metadata": {
        "id": "Xh4hui1aOCop"
      },
      "id": "Xh4hui1aOCop",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [],
      "metadata": {
        "id": "jWORxttXODzd"
      },
      "id": "jWORxttXODzd",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [],
      "metadata": {
        "id": "Itpxrs4QOWop"
      },
      "id": "Itpxrs4QOWop",
      "execution_count": null,
      "outputs": []
    }
  ],
  "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": "3.9.16"
    },
    "colab": {
      "provenance": []
    },
    "accelerator": "GPU",
    "gpuClass": "standard"
  },
  "nbformat": 4,
  "nbformat_minor": 5
}