BERT-WaimaiSentimentClassify.ipynb

{
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/DayDreamChaser/torch-action/blob/main/BERT%E2%80%94%E2%80%94transformers.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "e_fso1PElkCT"
      },
      "source": [
        "# BERT文本分类"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [],
      "metadata": {
        "id": "GObbqe_g0itK"
      }
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "QdnUpeb1lkCV"
      },
      "source": [
        "任务：微调BERT来处理文本情感分类任务\n",
        "\n",
        "数据集：美团外卖的用户评论数据集\n",
        "\n",
        "目标：把评论分成，好评(标签为1) 和 差评(标签为0)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 1,
      "metadata": {
        "id": "XNrvFDIclkCW"
      },
      "outputs": [],
      "source": [
        "#安装库\n",
        "# !pip install datasets\n",
        "# !pip install transformers[torch]\n",
        "# !pip install torchkeras"
      ]
    },
    {
      "cell_type": "code",
      "source": [],
      "metadata": {
        "id": "Abl67-9UFsi9"
      },
      "execution_count": 1,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Fqt1MTCalkCX"
      },
      "source": [
        "## 一，准备数据"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "nX2CwcRrlkCX"
      },
      "source": [
        "准备数据阶段主要需要用到的是datasets.Dataset 和transformers.AutoTokenizer。\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "kdOYBSRYlkCX"
      },
      "source": [
        "### 1，数据加载"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "h9VVb7yLlkCX"
      },
      "source": [
        "HuggingFace的datasets库提供了Dataset的功能。"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 2,
      "metadata": {
        "id": "EMYHMztLlkCY"
      },
      "outputs": [],
      "source": [
        "import numpy as np\n",
        "import pandas as pd\n",
        "import datasets\n",
        "\n",
        "import torch\n",
        "from torch.utils.data import DataLoader"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "import urllib.request\n",
        "import os\n",
        "\n",
        "url_base = 'https://raw.githubusercontent.com/SophonPlus/ChineseNlpCorpus/master/datasets/waimai_10k/'\n",
        "key_file = 'waimai_10k.csv'\n",
        "dataset_dir = \"data\"\n",
        "\n",
        "\n",
        "def _download(file_name):\n",
        "    file_path = dataset_dir + \"/\" + file_name\n",
        "    if os.path.exists(file_path):\n",
        "        return\n",
        "\n",
        "    print(\"Downloading \" + file_name + \" ... \")\n",
        "    urllib.request.urlretrieve(url_base + file_name, file_path)\n",
        "    print(\"Done\")\n",
        "\n",
        "_download(key_file)"
      ],
      "metadata": {
        "id": "cTuCcQim0HPs"
      },
      "execution_count": 3,
      "outputs": []
    },
    {
      "cell_type": "code",
      "execution_count": 4,
      "metadata": {
        "id": "z0ycQw0ulkCY"
      },
      "outputs": [],
      "source": [
        "df = pd.read_csv(\"data/waimai_10k.csv\")\n",
        "df = df.rename(columns={\"label\":\"labels\"})"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 5,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "alJcBzErlkCY",
        "outputId": "d20f5422-faeb-4612-8d35-343445661317"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "labels               1\n",
              "review    很快，好吃，味道足，量大\n",
              "Name: 0, dtype: object"
            ]
          },
          "metadata": {},
          "execution_count": 5
        }
      ],
      "source": [
        "df.loc[0]"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "df.head(10)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 363
        },
        "id": "P1Q3j-7U0sYi",
        "outputId": "35952111-4c16-4a03-c689-24b7744b8c5b"
      },
      "execution_count": 6,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "   labels                                review\n",
              "0       1                          很快，好吃，味道足，量大\n",
              "1       1                          没有送水没有送水没有送水\n",
              "2       1                              非常快，态度好。\n",
              "3       1                       方便，快捷，味道可口，快递给力\n",
              "4       1                          菜味道很棒！送餐很及时！\n",
              "5       1                     今天师傅是不是手抖了，微辣格外辣！\n",
              "6       1                      送餐快,态度也特别好,辛苦啦谢谢\n",
              "7       1  超级快就送到了，这么冷的天气骑士们辛苦了。谢谢你们。麻辣香锅依然很好吃。\n",
              "8       1            经过上次晚了2小时，这次超级快，20分钟就送到了……\n",
              "9       1                  最后五分钟订的，卖家特别好接单了，谢谢。"
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-c3316586-97bc-4ac0-a6b6-edb0073a6efd\" 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>labels</th>\n",
              "      <th>review</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>1</td>\n",
              "      <td>很快，好吃，味道足，量大</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>1</td>\n",
              "      <td>没有送水没有送水没有送水</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>1</td>\n",
              "      <td>非常快，态度好。</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>1</td>\n",
              "      <td>方便，快捷，味道可口，快递给力</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>1</td>\n",
              "      <td>菜味道很棒！送餐很及时！</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>5</th>\n",
              "      <td>1</td>\n",
              "      <td>今天师傅是不是手抖了，微辣格外辣！</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>6</th>\n",
              "      <td>1</td>\n",
              "      <td>送餐快,态度也特别好,辛苦啦谢谢</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>7</th>\n",
              "      <td>1</td>\n",
              "      <td>超级快就送到了，这么冷的天气骑士们辛苦了。谢谢你们。麻辣香锅依然很好吃。</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8</th>\n",
              "      <td>1</td>\n",
              "      <td>经过上次晚了2小时，这次超级快，20分钟就送到了……</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>9</th>\n",
              "      <td>1</td>\n",
              "      <td>最后五分钟订的，卖家特别好接单了，谢谢。</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>\n",
              "    <div class=\"colab-df-buttons\">\n",
              "\n",
              "  <div class=\"colab-df-container\">\n",
              "    <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-c3316586-97bc-4ac0-a6b6-edb0073a6efd')\"\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 -960 960 960\">\n",
              "    <path d=\"M120-120v-720h720v720H120Zm60-500h600v-160H180v160Zm220 220h160v-160H400v160Zm0 220h160v-160H400v160ZM180-400h160v-160H180v160Zm440 0h160v-160H620v160ZM180-180h160v-160H180v160Zm440 0h160v-160H620v160Z\"/>\n",
              "  </svg>\n",
              "    </button>\n",
              "\n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\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",
              "    .colab-df-buttons div {\n",
              "      margin-bottom: 4px;\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-c3316586-97bc-4ac0-a6b6-edb0073a6efd 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-c3316586-97bc-4ac0-a6b6-edb0073a6efd');\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",
              "\n",
              "\n",
              "<div id=\"df-92e0f83c-1a54-43b5-916d-878b6f1344a0\">\n",
              "  <button class=\"colab-df-quickchart\" onclick=\"quickchart('df-92e0f83c-1a54-43b5-916d-878b6f1344a0')\"\n",
              "            title=\"Suggest charts\"\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",
              "    <g>\n",
              "        <path d=\"M19 3H5c-1.1 0-2 .9-2 2v14c0 1.1.9 2 2 2h14c1.1 0 2-.9 2-2V5c0-1.1-.9-2-2-2zM9 17H7v-7h2v7zm4 0h-2V7h2v10zm4 0h-2v-4h2v4z\"/>\n",
              "    </g>\n",
              "</svg>\n",
              "  </button>\n",
              "\n",
              "<style>\n",
              "  .colab-df-quickchart {\n",
              "      --bg-color: #E8F0FE;\n",
              "      --fill-color: #1967D2;\n",
              "      --hover-bg-color: #E2EBFA;\n",
              "      --hover-fill-color: #174EA6;\n",
              "      --disabled-fill-color: #AAA;\n",
              "      --disabled-bg-color: #DDD;\n",
              "  }\n",
              "\n",
              "  [theme=dark] .colab-df-quickchart {\n",
              "      --bg-color: #3B4455;\n",
              "      --fill-color: #D2E3FC;\n",
              "      --hover-bg-color: #434B5C;\n",
              "      --hover-fill-color: #FFFFFF;\n",
              "      --disabled-bg-color: #3B4455;\n",
              "      --disabled-fill-color: #666;\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart {\n",
              "    background-color: var(--bg-color);\n",
              "    border: none;\n",
              "    border-radius: 50%;\n",
              "    cursor: pointer;\n",
              "    display: none;\n",
              "    fill: var(--fill-color);\n",
              "    height: 32px;\n",
              "    padding: 0;\n",
              "    width: 32px;\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart:hover {\n",
              "    background-color: var(--hover-bg-color);\n",
              "    box-shadow: 0 1px 2px rgba(60, 64, 67, 0.3), 0 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "    fill: var(--button-hover-fill-color);\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart-complete:disabled,\n",
              "  .colab-df-quickchart-complete:disabled:hover {\n",
              "    background-color: var(--disabled-bg-color);\n",
              "    fill: var(--disabled-fill-color);\n",
              "    box-shadow: none;\n",
              "  }\n",
              "\n",
              "  .colab-df-spinner {\n",
              "    border: 2px solid var(--fill-color);\n",
              "    border-color: transparent;\n",
              "    border-bottom-color: var(--fill-color);\n",
              "    animation:\n",
              "      spin 1s steps(1) infinite;\n",
              "  }\n",
              "\n",
              "  @keyframes spin {\n",
              "    0% {\n",
              "      border-color: transparent;\n",
              "      border-bottom-color: var(--fill-color);\n",
              "      border-left-color: var(--fill-color);\n",
              "    }\n",
              "    20% {\n",
              "      border-color: transparent;\n",
              "      border-left-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "    }\n",
              "    30% {\n",
              "      border-color: transparent;\n",
              "      border-left-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "      border-right-color: var(--fill-color);\n",
              "    }\n",
              "    40% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "    }\n",
              "    60% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "    }\n",
              "    80% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "      border-bottom-color: var(--fill-color);\n",
              "    }\n",
              "    90% {\n",
              "      border-color: transparent;\n",
              "      border-bottom-color: var(--fill-color);\n",
              "    }\n",
              "  }\n",
              "</style>\n",
              "\n",
              "  <script>\n",
              "    async function quickchart(key) {\n",
              "      const quickchartButtonEl =\n",
              "        document.querySelector('#' + key + ' button');\n",
              "      quickchartButtonEl.disabled = true;  // To prevent multiple clicks.\n",
              "      quickchartButtonEl.classList.add('colab-df-spinner');\n",
              "      try {\n",
              "        const charts = await google.colab.kernel.invokeFunction(\n",
              "            'suggestCharts', [key], {});\n",
              "      } catch (error) {\n",
              "        console.error('Error during call to suggestCharts:', error);\n",
              "      }\n",
              "      quickchartButtonEl.classList.remove('colab-df-spinner');\n",
              "      quickchartButtonEl.classList.add('colab-df-quickchart-complete');\n",
              "    }\n",
              "    (() => {\n",
              "      let quickchartButtonEl =\n",
              "        document.querySelector('#df-92e0f83c-1a54-43b5-916d-878b6f1344a0 button');\n",
              "      quickchartButtonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "    })();\n",
              "  </script>\n",
              "</div>\n",
              "    </div>\n",
              "  </div>\n"
            ],
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "dataframe",
              "variable_name": "df",
              "summary": "{\n  \"name\": \"df\",\n  \"rows\": 11987,\n  \"fields\": [\n    {\n      \"column\": \"labels\",\n      \"properties\": {\n        \"dtype\": \"number\",\n        \"std\": 0,\n        \"min\": 0,\n        \"max\": 1,\n        \"num_unique_values\": 2,\n        \"samples\": [\n          0,\n          1\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    },\n    {\n      \"column\": \"review\",\n      \"properties\": {\n        \"dtype\": \"string\",\n        \"num_unique_values\": 11980,\n        \"samples\": [\n          \"\\u7279\\u5927\\u676f\\u661f\\u51b0\\u4e50\\u548c\\u5927\\u676f\\u7684\\u62ff\\u94c1\\u4e00\\u4e2a\\u676f\\u578b\\uff0c\\u4e0b\\u6b21\\u8981\\u5f53\\u9762\\u786e\\u8ba4\\uff01\",\n          \"\\u4e2d\\u5348\\u5fd9\\u9001\\u9910\\u6162\\u53ef\\u4ee5\\u7406\\u89e3\\uff0c\\u5473\\u9053\\u4e00\\u822c\\u4e5f\\u53ef\\u4ee5\\u7406\\u89e3\\uff0c\\u4f46\\u7ca5\\u91cc\\u9762\\u957f\\u957f\\u7684\\u94a2\\u4e1d\\u7b97\\u600e\\u4e48\\u56de\\u4e8b\\uff1f\"\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    }\n  ]\n}"
            }
          },
          "metadata": {},
          "execution_count": 6
        }
      ]
    },
    {
      "source": [
        "# @title label\n",
        "from matplotlib import pyplot as plt\n",
        "df['labels'].plot(kind='hist', bins=3, title='labels')\n",
        "plt.gca().spines[['top', 'right',]].set_visible(False)"
      ],
      "cell_type": "code",
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 452
        },
        "id": "CozesKFc00VC",
        "outputId": "94a14460-b75a-4ac2-a37b-ebbcb3eebcee"
      }
    },
    {
      "source": [
        "df['len'] = df['review'].apply(len)\n",
        "df['len'].describe()"
      ],
      "cell_type": "code",
      "execution_count": 8,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "count    11987.000000\n",
              "mean        25.048553\n",
              "std         24.680028\n",
              "min          5.000000\n",
              "25%         11.000000\n",
              "50%         17.000000\n",
              "75%         30.000000\n",
              "max        463.000000\n",
              "Name: len, dtype: float64"
            ]
          },
          "metadata": {},
          "execution_count": 8
        }
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "-bQbePFP0wqa",
        "outputId": "2bf2c9c7-9989-4300-b13b-6795a6b18297"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "平均长度25.05， 中位数17， 上四分位30，最大463\n",
        "取最大长度50截断"
      ],
      "metadata": {
        "id": "owM18o1n366J"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "df['len'].plot(kind='hist', bins=20, title='review len')\n",
        "plt.gca().spines[['top', 'right',]].set_visible(False)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 452
        },
        "id": "B25gIvM81XrV",
        "outputId": "e942a501-5d36-4bda-910e-79ebcd0d6c9f"
      },
      "execution_count": 9,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 10,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "-UTFi_9glkCZ",
        "outputId": "b75a06e7-c83d-42c6-f70e-c1fed28a65a8"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "10                量大，好吃，每次点的都够吃两次\n",
              "11                     挺辣的，吃着还可以吧\n",
              "12                    味道好，送餐快，分量足\n",
              "13                     量足，好吃，送餐也快\n",
              "14         特别好吃，量特大，而且送餐特别快，特别特别棒\n",
              "15                     口感好的很，速度快！\n",
              "16           相当好吃的香锅，分量够足，味道也没的说。\n",
              "17    好吃！速度！包装也有品质，不出家门就能吃到餐厅的味道！\n",
              "18                味道好极啦，送餐很快师傅辛苦啦\n",
              "19                  量大味道好，送餐师傅都很好\n",
              "Name: review, dtype: object"
            ]
          },
          "metadata": {},
          "execution_count": 10
        }
      ],
      "source": [
        "df[10:20][\"review\"]"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 11,
      "metadata": {
        "id": "djw7AyoylkCZ",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 470
        },
        "outputId": "40f6a40a-c4e0-4fec-bb28-3432a85511cd"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<Axes: title={'center': 'labels'}, ylabel='Frequency'>"
            ]
          },
          "metadata": {},
          "execution_count": 11
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ],
      "source": [
        "neg_df = df[df['labels']==0]\n",
        "pos_df = df[df['labels']==1]\n",
        "\n",
        "pos_df = pos_df.sample(n=1000, random_state=200)  # 1000条\n",
        "neg_df = neg_df.sample(n=1000, random_state=200)  # 1000条\n",
        "df = pd.concat([neg_df, pos_df], axis=0)\n",
        "\n",
        "df['labels'].plot(kind='hist', bins=3, title='labels')"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "ds = datasets.Dataset.from_pandas(df)\n",
        "ds"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "GMNvKNGbDmZs",
        "outputId": "0e6b910b-f83f-49b8-ada1-031d4f9b4ad5"
      },
      "execution_count": 12,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Dataset({\n",
              "    features: ['labels', 'review', 'len', '__index_level_0__'],\n",
              "    num_rows: 2000\n",
              "})"
            ]
          },
          "metadata": {},
          "execution_count": 12
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "OyV-3meFlkCZ"
      },
      "source": [
        "### 2，文本分词"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "WN_jG2AplkCa"
      },
      "source": [
        "transformers库使用tokenizer进行文本分词。\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 13,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "awY6AYmQlkCa",
        "outputId": "bc987056-df83-4637-a90b-e0c5f485cc11"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "/usr/local/lib/python3.10/dist-packages/huggingface_hub/utils/_token.py:88: UserWarning: \n",
            "The secret `HF_TOKEN` does not exist in your Colab secrets.\n",
            "To authenticate with the Hugging Face Hub, create a token in your settings tab (https://huggingface.co/settings/tokens), set it as secret in your Google Colab and restart your session.\n",
            "You will be able to reuse this secret in all of your notebooks.\n",
            "Please note that authentication is recommended but still optional to access public models or datasets.\n",
            "  warnings.warn(\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "BertTokenizerFast(name_or_path='hfl/chinese-roberta-wwm-ext-large', vocab_size=21128, model_max_length=1000000000000000019884624838656, is_fast=True, padding_side='right', truncation_side='right', special_tokens={'unk_token': '[UNK]', 'sep_token': '[SEP]', 'pad_token': '[PAD]', 'cls_token': '[CLS]', 'mask_token': '[MASK]'}, clean_up_tokenization_spaces=True),  added_tokens_decoder={\n",
            "\t0: AddedToken(\"[PAD]\", rstrip=False, lstrip=False, single_word=False, normalized=False, special=True),\n",
            "\t100: AddedToken(\"[UNK]\", rstrip=False, lstrip=False, single_word=False, normalized=False, special=True),\n",
            "\t101: AddedToken(\"[CLS]\", rstrip=False, lstrip=False, single_word=False, normalized=False, special=True),\n",
            "\t102: AddedToken(\"[SEP]\", rstrip=False, lstrip=False, single_word=False, normalized=False, special=True),\n",
            "\t103: AddedToken(\"[MASK]\", rstrip=False, lstrip=False, single_word=False, normalized=False, special=True),\n",
            "}\n"
          ]
        }
      ],
      "source": [
        "from transformers import AutoTokenizer #BertTokenizer\n",
        "\n",
        "model_id = \"hfl/chinese-roberta-wwm-ext-large\"\n",
        "tokenizer = AutoTokenizer.from_pretrained(model_id) #需要和模型一致\n",
        "print(tokenizer)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 22,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "gpu0POc6lkCa",
        "outputId": "998d0c63-84d1-4dd9-86d2-ceb160be5ea7"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "一定不要点炒面,2小时终于坨成了石头\n",
            "{'input_ids': [101, 671, 2137, 679, 6206, 4157, 4143, 7481, 117, 123, 2207, 3198, 5303, 754, 1789, 2768, 749, 4767, 1928, 102], 'token_type_ids': [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], 'attention_mask': [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1]}\n"
          ]
        }
      ],
      "source": [
        "#tokenizer可以使用 __call__,encode,encode_plus,batch_encode_plus等方法编码\n",
        "#可以使用decode,batch_decode等方法进行解码\n",
        "sample_txt = ds['review'][0]\n",
        "print(sample_txt)\n",
        "text_codes = tokenizer(text=sample_txt,\n",
        "                       text_pair = None,\n",
        "                       max_length = 100,\n",
        "                       truncation = True,\n",
        "                       padding= 'do_not_pad') #可选'longest','max_length','do_not_pad'\n",
        "\n",
        "#input_ids是编码后的数字，token_type_ids表示来自第1个句子还是第2个句子\n",
        "#attention_mask在padding的位置是0其它位置是1\n",
        "print(text_codes)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 15,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "2OsblOwqlkCb",
        "outputId": "303ceb9a-ba2f-4cc6-821d-6d3c11def24d"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[CLS]\n",
            "[SEP]\n"
          ]
        }
      ],
      "source": [
        "print(tokenizer.decode(text_codes[\"input_ids\"][0]))\n",
        "print(tokenizer.decode(text_codes[\"input_ids\"][-1]))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 24,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "b68ZLNtmlkCb",
        "outputId": "01a42a41-6088-4ca8-8f7d-f5ce28257e48"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "['[CLS]',\n",
              " '一',\n",
              " '定',\n",
              " '不',\n",
              " '要',\n",
              " '点',\n",
              " '炒',\n",
              " '面',\n",
              " ',',\n",
              " '2',\n",
              " '小',\n",
              " '时',\n",
              " '终',\n",
              " '于',\n",
              " '坨',\n",
              " '成',\n",
              " '了',\n",
              " '石',\n",
              " '头',\n",
              " '[SEP]']"
            ]
          },
          "metadata": {},
          "execution_count": 24
        }
      ],
      "source": [
        "tokenizer.batch_decode(text_codes[\"input_ids\"])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 25,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "IxBjSucUlkCb",
        "outputId": "13724e86-7b95-4e7b-d181-68697de19f7f"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "tokens= ['一', '定', '不', '要', '点', '炒', '面', ',', '2', '小', '时', '终', '于', '坨', '成', '了', '石', '头']\n",
            "ids =  [671, 2137, 679, 6206, 4157, 4143, 7481, 117, 123, 2207, 3198, 5303, 754, 1789, 2768, 749, 4767, 1928]\n"
          ]
        }
      ],
      "source": [
        "tokens = tokenizer.tokenize(sample_txt)\n",
        "print(\"tokens=\",tokens)\n",
        "\n",
        "ids = tokenizer.convert_tokens_to_ids(tokens)\n",
        "print(\"ids = \",ids)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "1_Rwy7CKlkCc"
      },
      "source": [
        "### 3，传入DataLoader"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 26,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 49,
          "referenced_widgets": [
            "5e17e92f944c4baba8f833dc39536f8f",
            "b118110b6c1a4e35a19e8e4f7e4aa66e",
            "14905759328a47d0a32f94344c741a5a",
            "23471a88434642798422fb3dd147e8ed",
            "32f6f165951b433082b30d795f7d8c98",
            "6902740078df4375bc31a6fc630cbd7e",
            "34c652c518f84dc18a7444b4ed6c4fee",
            "a9a9f8db1b804941b88ee1987bd6f4c6",
            "0a6e5118777348eea83ad7c38cf9fe29",
            "f923393b93554bc48360f415004f8560",
            "3731ccc31e1a4349b52c2a6c4a91e0a8"
          ]
        },
        "id": "CG6Ei4PllkCc",
        "outputId": "7c6485f2-689d-4637-e746-b6aa665006e0"
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "Map (num_proc=4):   0%|          | 0/2000 [00:00<?, ? examples/s]"
            ],
            "application/vnd.jupyter.widget-view+json": {
              "version_major": 2,
              "version_minor": 0,
              "model_id": "5e17e92f944c4baba8f833dc39536f8f"
            }
          },
          "metadata": {}
        }
      ],
      "source": [
        "# 进行批量Tokenize\n",
        "ds_encoded = ds.map(lambda x:tokenizer(x[\"review\"], max_length=50, truncation=True, padding='max_length'),\n",
        "                    batched=True,\n",
        "                    batch_size=16,\n",
        "                    num_proc=4) #支持批处理和多进程map"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 27,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "2BoB4p7IlkCc",
        "outputId": "25d3cb77-6be7-451e-b353-a27321efe643"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "{'labels': tensor(0),\n",
              " 'input_ids': tensor([ 101,  671, 2137,  679, 6206, 4157, 4143, 7481,  117,  123, 2207, 3198,\n",
              "         5303,  754, 1789, 2768,  749, 4767, 1928,  102,    0,    0,    0,    0,\n",
              "            0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,\n",
              "            0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,\n",
              "            0,    0]),\n",
              " 'token_type_ids': tensor([0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
              "         0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
              "         0, 0]),\n",
              " 'attention_mask': tensor([1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0,\n",
              "         0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
              "         0, 0])}"
            ]
          },
          "metadata": {},
          "execution_count": 27
        }
      ],
      "source": [
        "#转换成pytorch中的tensor\n",
        "ds_encoded.set_format(type=\"torch\",\n",
        "           columns = [\"input_ids\",'attention_mask','token_type_ids','labels'])\n",
        "\n",
        "#ds_encoded.reset_format()\n",
        "ds_encoded[0]"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 28,
      "metadata": {
        "id": "-X7XA4PqlkCc",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "abac576a-d7c3-4117-fa2f-0ac62917a540"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Dataset({\n",
            "    features: ['labels', 'review', 'len', '__index_level_0__', 'input_ids', 'token_type_ids', 'attention_mask'],\n",
            "    num_rows: 1280\n",
            "}) Dataset({\n",
            "    features: ['labels', 'review', 'len', '__index_level_0__', 'input_ids', 'token_type_ids', 'attention_mask'],\n",
            "    num_rows: 320\n",
            "}) Dataset({\n",
            "    features: ['labels', 'review', 'len', '__index_level_0__', 'input_ids', 'token_type_ids', 'attention_mask'],\n",
            "    num_rows: 400\n",
            "})\n"
          ]
        }
      ],
      "source": [
        "#分割成训练集和测试集\n",
        "ds_train_val, ds_test = ds_encoded.train_test_split(test_size=0.2).values()\n",
        "ds_train, ds_val = ds_train_val.train_test_split(test_size=0.2).values()\n",
        "\n",
        "print(ds_train, ds_val, ds_test)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 30,
      "metadata": {
        "id": "H5PMwHTTlkCc"
      },
      "outputs": [],
      "source": [
        "#在collate_fn中可以做动态批处理(dynamic batching)\n",
        "def collate_fn(examples):\n",
        "    return tokenizer.pad(examples) #return_tensors='pt'\n",
        "\n",
        "#以下方式等价\n",
        "#from transformers import DataCollatorWithPadding\n",
        "#collate_fn = DataCollatorWithPadding(tokenizer=tokenizer)\n",
        "\n",
        "batch_size = 32\n",
        "dl_train = torch.utils.data.DataLoader(ds_train, batch_size=batch_size, collate_fn = collate_fn)\n",
        "dl_val = torch.utils.data.DataLoader(ds_val, batch_size=batch_size,  collate_fn = collate_fn)\n",
        "dl_test = torch.utils.data.DataLoader(ds_test, batch_size=batch_size,  collate_fn = collate_fn)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 32,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "e068uKxHlkCd",
        "outputId": "219c389c-2e8d-47bb-ee55-6864a0f23eb2"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "{'labels': tensor([1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 0, 0, 0, 1, 0, 0, 1, 0, 1, 0, 0, 0, 0,\n",
            "        0, 1, 0, 0, 1, 0, 0, 1]), 'input_ids': tensor([[ 101, 4343, 5489,  ...,    0,    0,    0],\n",
            "        [ 101, 3739, 6963,  ...,    0,    0,    0],\n",
            "        [ 101, 7883, 5597,  ...,    0,    0,    0],\n",
            "        ...,\n",
            "        [ 101, 4636, 2428,  ...,  679,  671,  102],\n",
            "        [ 101,  671,  819,  ...,    0,    0,    0],\n",
            "        [ 101, 1348, 2563,  ...,    0,    0,    0]]), 'token_type_ids': tensor([[0, 0, 0,  ..., 0, 0, 0],\n",
            "        [0, 0, 0,  ..., 0, 0, 0],\n",
            "        [0, 0, 0,  ..., 0, 0, 0],\n",
            "        ...,\n",
            "        [0, 0, 0,  ..., 0, 0, 0],\n",
            "        [0, 0, 0,  ..., 0, 0, 0],\n",
            "        [0, 0, 0,  ..., 0, 0, 0]]), 'attention_mask': tensor([[1, 1, 1,  ..., 0, 0, 0],\n",
            "        [1, 1, 1,  ..., 0, 0, 0],\n",
            "        [1, 1, 1,  ..., 0, 0, 0],\n",
            "        ...,\n",
            "        [1, 1, 1,  ..., 1, 1, 1],\n",
            "        [1, 1, 1,  ..., 0, 0, 0],\n",
            "        [1, 1, 1,  ..., 0, 0, 0]])}\n"
          ]
        }
      ],
      "source": [
        "for batch in dl_train:\n",
        "  print(batch)\n",
        "  break"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "mGZ7N4YAlkCd"
      },
      "source": [
        "## 二，定义模型"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "9Vn9hIyRlkCd"
      },
      "source": [
        "一个完整的模型(Model)包括模型架构(Architecture)和模型权重(Checkpoints/Weights)。\n",
        "\n",
        "transformers提供了3种指定模型架构的方法。\n",
        "\n",
        "* 第1种是指定模型架构(如: from transformers import BertModel)\n",
        "\n",
        "* 第2种是自动推断模型架构(如: from transformers import AutoModel)\n",
        "\n",
        "* 第3种是自动推断模型架构并自动添加Head (如: from transformers import AutoModelForSequenceClassification )\n",
        "\n",
        "第1种方案和第2种方案用户可以灵活地根据自己要做的任务设计Head，并且需要对基础模型有一定的了解。\n",
        "\n",
        "\n",
        "此处我们使用第3种方案。\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 33,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "X4IwVlsNlkCe",
        "outputId": "3c60830f-244b-4026-d4ba-f6e0cf889b3b"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "Some weights of BertForSequenceClassification were not initialized from the model checkpoint at hfl/chinese-roberta-wwm-ext-large and are newly initialized: ['classifier.bias', 'classifier.weight']\n",
            "You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference.\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "dict_keys(['bert', 'dropout', 'classifier'])"
            ]
          },
          "metadata": {},
          "execution_count": 33
        }
      ],
      "source": [
        "from transformers import AutoModelForSequenceClassification\n",
        "\n",
        "#加载模型 (会添加针对特定任务类型的Head)\n",
        "model = AutoModelForSequenceClassification.from_pretrained(model_id, num_labels=2)\n",
        "\n",
        "dict(model.named_children()).keys()"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "model"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "L62FcYsA43EG",
        "outputId": "7d0ea800-1f16-472c-c811-0893036e0d35"
      },
      "execution_count": 34,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "BertForSequenceClassification(\n",
              "  (bert): BertModel(\n",
              "    (embeddings): BertEmbeddings(\n",
              "      (word_embeddings): Embedding(21128, 1024, padding_idx=0)\n",
              "      (position_embeddings): Embedding(512, 1024)\n",
              "      (token_type_embeddings): Embedding(2, 1024)\n",
              "      (LayerNorm): LayerNorm((1024,), eps=1e-12, elementwise_affine=True)\n",
              "      (dropout): Dropout(p=0.1, inplace=False)\n",
              "    )\n",
              "    (encoder): BertEncoder(\n",
              "      (layer): ModuleList(\n",
              "        (0-23): 24 x BertLayer(\n",
              "          (attention): BertAttention(\n",
              "            (self): BertSelfAttention(\n",
              "              (query): Linear(in_features=1024, out_features=1024, bias=True)\n",
              "              (key): Linear(in_features=1024, out_features=1024, bias=True)\n",
              "              (value): Linear(in_features=1024, out_features=1024, bias=True)\n",
              "              (dropout): Dropout(p=0.1, inplace=False)\n",
              "            )\n",
              "            (output): BertSelfOutput(\n",
              "              (dense): Linear(in_features=1024, out_features=1024, bias=True)\n",
              "              (LayerNorm): LayerNorm((1024,), eps=1e-12, elementwise_affine=True)\n",
              "              (dropout): Dropout(p=0.1, inplace=False)\n",
              "            )\n",
              "          )\n",
              "          (intermediate): BertIntermediate(\n",
              "            (dense): Linear(in_features=1024, out_features=4096, bias=True)\n",
              "            (intermediate_act_fn): GELUActivation()\n",
              "          )\n",
              "          (output): BertOutput(\n",
              "            (dense): Linear(in_features=4096, out_features=1024, bias=True)\n",
              "            (LayerNorm): LayerNorm((1024,), eps=1e-12, elementwise_affine=True)\n",
              "            (dropout): Dropout(p=0.1, inplace=False)\n",
              "          )\n",
              "        )\n",
              "      )\n",
              "    )\n",
              "    (pooler): BertPooler(\n",
              "      (dense): Linear(in_features=1024, out_features=1024, bias=True)\n",
              "      (activation): Tanh()\n",
              "    )\n",
              "  )\n",
              "  (dropout): Dropout(p=0.1, inplace=False)\n",
              "  (classifier): Linear(in_features=1024, out_features=2, bias=True)\n",
              ")"
            ]
          },
          "metadata": {},
          "execution_count": 34
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "device = 'cuda' if torch.cuda.is_available() else 'cpu'\n",
        "device"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        },
        "id": "NFr4Tg91FdOj",
        "outputId": "81d81560-a639-42c1-b89c-8d284f953ff7"
      },
      "execution_count": 35,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "'cuda'"
            ],
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "string"
            }
          },
          "metadata": {},
          "execution_count": 35
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ULn1JyQulkCe"
      },
      "source": [
        "我们可以用一个batch的数据去试算一下"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 37,
      "metadata": {
        "id": "Db3PO-melkCe"
      },
      "outputs": [],
      "source": [
        "model = model.to(device)\n",
        "output = model(**batch.to(device))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 38,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "bBy4l-XjlkCe",
        "outputId": "a30d48ba-03be-486b-d1ce-64d42bcfd5f6"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "tensor(0.6930, device='cuda:0', grad_fn=<NllLossBackward0>)"
            ]
          },
          "metadata": {},
          "execution_count": 38
        }
      ],
      "source": [
        "output.loss"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 39,
      "metadata": {
        "id": "A_HjBtpVlkCf",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "b14b0bc7-6a5f-49cb-8f0b-565397e5f314"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "SequenceClassifierOutput(loss=tensor(0.6930, device='cuda:0', grad_fn=<NllLossBackward0>), logits=tensor([[ 0.2274,  0.0879],\n",
              "        [ 0.2534, -0.0826],\n",
              "        [ 0.2366,  0.1563],\n",
              "        [ 0.3862,  0.0725],\n",
              "        [ 0.1837,  0.1254],\n",
              "        [ 0.2206, -0.1636],\n",
              "        [ 0.3930,  0.0513],\n",
              "        [ 0.3388, -0.1072],\n",
              "        [ 0.2994, -0.0508],\n",
              "        [ 0.2531, -0.0265],\n",
              "        [ 0.4523,  0.0403],\n",
              "        [ 0.2359,  0.1121],\n",
              "        [ 0.4592, -0.0588],\n",
              "        [ 0.2201, -0.0724],\n",
              "        [ 0.2708,  0.1066],\n",
              "        [ 0.3107, -0.0762],\n",
              "        [ 0.1405,  0.1901],\n",
              "        [ 0.2418,  0.0127],\n",
              "        [ 0.2504,  0.1760],\n",
              "        [ 0.3721,  0.2222],\n",
              "        [ 0.2607, -0.0076],\n",
              "        [ 0.3155,  0.0731],\n",
              "        [ 0.1676,  0.0458],\n",
              "        [ 0.1966, -0.1905],\n",
              "        [ 0.2911,  0.1291],\n",
              "        [ 0.2567,  0.1316],\n",
              "        [ 0.3107,  0.0167],\n",
              "        [ 0.2838,  0.0387],\n",
              "        [ 0.4149,  0.0059],\n",
              "        [ 0.1522,  0.0682],\n",
              "        [ 0.2589,  0.1399],\n",
              "        [ 0.1181, -0.2881]], device='cuda:0', grad_fn=<AddmmBackward0>), hidden_states=None, attentions=None)"
            ]
          },
          "metadata": {},
          "execution_count": 39
        }
      ],
      "source": [
        "output"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "cSDboKVIlkCf"
      },
      "source": [
        "## 三，训练模型"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "!pip install accelerate"
      ],
      "metadata": {
        "id": "i_9WP02e5R0T"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "20vuo5OQlkCf"
      },
      "source": [
        "下面使用我们的梦中情炉 torchkeras 来实现最优雅的微调训练循环。🤗🤗\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 40,
      "metadata": {
        "id": "Lc70XraplkCf"
      },
      "outputs": [],
      "source": [
        "from torchkeras import KerasModel\n",
        "\n",
        "#我们需要修改StepRunner以适应transformers的数据集格式\n",
        "\n",
        "class StepRunner:\n",
        "    def __init__(self, net, loss_fn, accelerator, stage = \"train\", metrics_dict = None,\n",
        "                 optimizer = None, lr_scheduler = None\n",
        "                 ):\n",
        "        self.net,self.loss_fn,self.metrics_dict,self.stage = net,loss_fn,metrics_dict,stage\n",
        "        self.optimizer,self.lr_scheduler = optimizer,lr_scheduler\n",
        "        self.accelerator = accelerator\n",
        "        if self.stage=='train':\n",
        "            self.net.train()\n",
        "        else:\n",
        "            self.net.eval()\n",
        "\n",
        "    def __call__(self, batch):\n",
        "\n",
        "        out = self.net(**batch.to(device))\n",
        "\n",
        "        #loss\n",
        "        loss = out.loss\n",
        "\n",
        "        #preds\n",
        "        preds =(out.logits).argmax(axis=1)\n",
        "\n",
        "        #backward()\n",
        "        if self.optimizer is not None and self.stage==\"train\":\n",
        "            self.accelerator.backward(loss)\n",
        "            self.optimizer.step()\n",
        "            if self.lr_scheduler is not None:\n",
        "                self.lr_scheduler.step()\n",
        "            self.optimizer.zero_grad()\n",
        "\n",
        "        all_loss = self.accelerator.gather(loss).sum()\n",
        "\n",
        "        labels = batch['labels']\n",
        "        acc = (preds==labels).sum()/((labels>-1).sum())\n",
        "\n",
        "        all_acc = self.accelerator.gather(acc).mean()\n",
        "\n",
        "        #losses\n",
        "        step_losses = {self.stage+\"_loss\":all_loss.item(), self.stage+'_acc':all_acc.item()}\n",
        "\n",
        "        #metrics\n",
        "        step_metrics = {}\n",
        "\n",
        "        if self.stage==\"train\":\n",
        "            if self.optimizer is not None:\n",
        "                step_metrics['lr'] = self.optimizer.state_dict()['param_groups'][0]['lr']\n",
        "            else:\n",
        "                step_metrics['lr'] = 0.0\n",
        "        return step_losses,step_metrics\n",
        "\n",
        "KerasModel.StepRunner = StepRunner"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 41,
      "metadata": {
        "id": "yF4BoqZHlkCg"
      },
      "outputs": [],
      "source": [
        "model = model.to(device)\n",
        "\n",
        "optimizer = torch.optim.AdamW(model.parameters(), lr=3e-5)\n",
        "keras_model = KerasModel(model,loss_fn=None,optimizer = optimizer)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 42,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 973
        },
        "id": "jaydv06LlkCh",
        "outputId": "c668cebe-7bcb-4dd3-8b06-19c0afaac13f"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "\u001b[0;31m<<<<<< ⚡️ cuda is used >>>>>>\u001b[0m\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 600x400 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ],
            "text/html": [
              "\n",
              "<style>\n",
              "    /* background: */\n",
              "    progress::-webkit-progress-bar {background-color: #CDCDCD; width: 100%;}\n",
              "    progress {background-color: #CDCDCD;}\n",
              "\n",
              "    /* value: */\n",
              "    progress::-webkit-progress-value {background-color: #00BFFF  !important;}\n",
              "    progress::-moz-progress-bar {background-color: #00BFFF  !important;}\n",
              "    progress {color: #00BFFF ;}\n",
              "\n",
              "    /* optional */\n",
              "    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
              "        background: #000000;\n",
              "    }\n",
              "</style>\n"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ],
            "text/html": [
              "\n",
              "    <div>\n",
              "      <progress value='14' class='progress-bar-interrupted' max='20' style='width:300px; height:20px; vertical-align: middle;'></progress>\n",
              "      70.00% [14/20] [09:34<04:06]\n",
              "      <br>\n",
              "      ████████████████████100.00% [10/10] [val_loss=0.7394, val_acc=0.8719]\n",
              "    </div>\n",
              "    "
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "\u001b[0;31m<<<<<< val_acc without improvement in 10 epoch,early stopping >>>>>> \n",
            "\u001b[0m\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "    epoch  train_loss  train_acc       lr  val_loss   val_acc\n",
              "0       1    0.400042   0.815625  0.00003  0.298466  0.878125\n",
              "1       2    0.178673   0.942187  0.00003  0.293229  0.887500\n",
              "2       3    0.079503   0.975000  0.00003  0.435754  0.875000\n",
              "3       4    0.047774   0.987500  0.00003  0.483032  0.890625\n",
              "4       5    0.030902   0.987500  0.00003  0.593966  0.856250\n",
              "5       6    0.044966   0.985938  0.00003  0.639174  0.868750\n",
              "6       7    0.044582   0.989062  0.00003  0.647959  0.859375\n",
              "7       8    0.073322   0.980469  0.00003  0.604683  0.828125\n",
              "8       9    0.040865   0.985938  0.00003  0.535952  0.884375\n",
              "9      10    0.036645   0.989844  0.00003  0.567870  0.843750\n",
              "10     11    0.014979   0.996094  0.00003  0.616570  0.875000\n",
              "11     12    0.003719   0.999219  0.00003  0.646139  0.884375\n",
              "12     13    0.000601   1.000000  0.00003  0.699452  0.881250\n",
              "13     14    0.000337   1.000000  0.00003  0.739432  0.871875"
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-9ad9d0b7-43c5-4cb4-9e1f-c016c34a71a4\" 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>epoch</th>\n",
              "      <th>train_loss</th>\n",
              "      <th>train_acc</th>\n",
              "      <th>lr</th>\n",
              "      <th>val_loss</th>\n",
              "      <th>val_acc</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>1</td>\n",
              "      <td>0.400042</td>\n",
              "      <td>0.815625</td>\n",
              "      <td>0.00003</td>\n",
              "      <td>0.298466</td>\n",
              "      <td>0.878125</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>2</td>\n",
              "      <td>0.178673</td>\n",
              "      <td>0.942187</td>\n",
              "      <td>0.00003</td>\n",
              "      <td>0.293229</td>\n",
              "      <td>0.887500</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>3</td>\n",
              "      <td>0.079503</td>\n",
              "      <td>0.975000</td>\n",
              "      <td>0.00003</td>\n",
              "      <td>0.435754</td>\n",
              "      <td>0.875000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>4</td>\n",
              "      <td>0.047774</td>\n",
              "      <td>0.987500</td>\n",
              "      <td>0.00003</td>\n",
              "      <td>0.483032</td>\n",
              "      <td>0.890625</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>5</td>\n",
              "      <td>0.030902</td>\n",
              "      <td>0.987500</td>\n",
              "      <td>0.00003</td>\n",
              "      <td>0.593966</td>\n",
              "      <td>0.856250</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>5</th>\n",
              "      <td>6</td>\n",
              "      <td>0.044966</td>\n",
              "      <td>0.985938</td>\n",
              "      <td>0.00003</td>\n",
              "      <td>0.639174</td>\n",
              "      <td>0.868750</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>6</th>\n",
              "      <td>7</td>\n",
              "      <td>0.044582</td>\n",
              "      <td>0.989062</td>\n",
              "      <td>0.00003</td>\n",
              "      <td>0.647959</td>\n",
              "      <td>0.859375</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>7</th>\n",
              "      <td>8</td>\n",
              "      <td>0.073322</td>\n",
              "      <td>0.980469</td>\n",
              "      <td>0.00003</td>\n",
              "      <td>0.604683</td>\n",
              "      <td>0.828125</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8</th>\n",
              "      <td>9</td>\n",
              "      <td>0.040865</td>\n",
              "      <td>0.985938</td>\n",
              "      <td>0.00003</td>\n",
              "      <td>0.535952</td>\n",
              "      <td>0.884375</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>9</th>\n",
              "      <td>10</td>\n",
              "      <td>0.036645</td>\n",
              "      <td>0.989844</td>\n",
              "      <td>0.00003</td>\n",
              "      <td>0.567870</td>\n",
              "      <td>0.843750</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>10</th>\n",
              "      <td>11</td>\n",
              "      <td>0.014979</td>\n",
              "      <td>0.996094</td>\n",
              "      <td>0.00003</td>\n",
              "      <td>0.616570</td>\n",
              "      <td>0.875000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>11</th>\n",
              "      <td>12</td>\n",
              "      <td>0.003719</td>\n",
              "      <td>0.999219</td>\n",
              "      <td>0.00003</td>\n",
              "      <td>0.646139</td>\n",
              "      <td>0.884375</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>12</th>\n",
              "      <td>13</td>\n",
              "      <td>0.000601</td>\n",
              "      <td>1.000000</td>\n",
              "      <td>0.00003</td>\n",
              "      <td>0.699452</td>\n",
              "      <td>0.881250</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>13</th>\n",
              "      <td>14</td>\n",
              "      <td>0.000337</td>\n",
              "      <td>1.000000</td>\n",
              "      <td>0.00003</td>\n",
              "      <td>0.739432</td>\n",
              "      <td>0.871875</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>\n",
              "    <div class=\"colab-df-buttons\">\n",
              "\n",
              "  <div class=\"colab-df-container\">\n",
              "    <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-9ad9d0b7-43c5-4cb4-9e1f-c016c34a71a4')\"\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 -960 960 960\">\n",
              "    <path d=\"M120-120v-720h720v720H120Zm60-500h600v-160H180v160Zm220 220h160v-160H400v160Zm0 220h160v-160H400v160ZM180-400h160v-160H180v160Zm440 0h160v-160H620v160ZM180-180h160v-160H180v160Zm440 0h160v-160H620v160Z\"/>\n",
              "  </svg>\n",
              "    </button>\n",
              "\n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\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",
              "    .colab-df-buttons div {\n",
              "      margin-bottom: 4px;\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-9ad9d0b7-43c5-4cb4-9e1f-c016c34a71a4 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-9ad9d0b7-43c5-4cb4-9e1f-c016c34a71a4');\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",
              "\n",
              "\n",
              "<div id=\"df-a2970215-9ad7-4add-8882-8ab4fe349a85\">\n",
              "  <button class=\"colab-df-quickchart\" onclick=\"quickchart('df-a2970215-9ad7-4add-8882-8ab4fe349a85')\"\n",
              "            title=\"Suggest charts\"\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",
              "    <g>\n",
              "        <path d=\"M19 3H5c-1.1 0-2 .9-2 2v14c0 1.1.9 2 2 2h14c1.1 0 2-.9 2-2V5c0-1.1-.9-2-2-2zM9 17H7v-7h2v7zm4 0h-2V7h2v10zm4 0h-2v-4h2v4z\"/>\n",
              "    </g>\n",
              "</svg>\n",
              "  </button>\n",
              "\n",
              "<style>\n",
              "  .colab-df-quickchart {\n",
              "      --bg-color: #E8F0FE;\n",
              "      --fill-color: #1967D2;\n",
              "      --hover-bg-color: #E2EBFA;\n",
              "      --hover-fill-color: #174EA6;\n",
              "      --disabled-fill-color: #AAA;\n",
              "      --disabled-bg-color: #DDD;\n",
              "  }\n",
              "\n",
              "  [theme=dark] .colab-df-quickchart {\n",
              "      --bg-color: #3B4455;\n",
              "      --fill-color: #D2E3FC;\n",
              "      --hover-bg-color: #434B5C;\n",
              "      --hover-fill-color: #FFFFFF;\n",
              "      --disabled-bg-color: #3B4455;\n",
              "      --disabled-fill-color: #666;\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart {\n",
              "    background-color: var(--bg-color);\n",
              "    border: none;\n",
              "    border-radius: 50%;\n",
              "    cursor: pointer;\n",
              "    display: none;\n",
              "    fill: var(--fill-color);\n",
              "    height: 32px;\n",
              "    padding: 0;\n",
              "    width: 32px;\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart:hover {\n",
              "    background-color: var(--hover-bg-color);\n",
              "    box-shadow: 0 1px 2px rgba(60, 64, 67, 0.3), 0 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "    fill: var(--button-hover-fill-color);\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart-complete:disabled,\n",
              "  .colab-df-quickchart-complete:disabled:hover {\n",
              "    background-color: var(--disabled-bg-color);\n",
              "    fill: var(--disabled-fill-color);\n",
              "    box-shadow: none;\n",
              "  }\n",
              "\n",
              "  .colab-df-spinner {\n",
              "    border: 2px solid var(--fill-color);\n",
              "    border-color: transparent;\n",
              "    border-bottom-color: var(--fill-color);\n",
              "    animation:\n",
              "      spin 1s steps(1) infinite;\n",
              "  }\n",
              "\n",
              "  @keyframes spin {\n",
              "    0% {\n",
              "      border-color: transparent;\n",
              "      border-bottom-color: var(--fill-color);\n",
              "      border-left-color: var(--fill-color);\n",
              "    }\n",
              "    20% {\n",
              "      border-color: transparent;\n",
              "      border-left-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "    }\n",
              "    30% {\n",
              "      border-color: transparent;\n",
              "      border-left-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "      border-right-color: var(--fill-color);\n",
              "    }\n",
              "    40% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "    }\n",
              "    60% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "    }\n",
              "    80% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "      border-bottom-color: var(--fill-color);\n",
              "    }\n",
              "    90% {\n",
              "      border-color: transparent;\n",
              "      border-bottom-color: var(--fill-color);\n",
              "    }\n",
              "  }\n",
              "</style>\n",
              "\n",
              "  <script>\n",
              "    async function quickchart(key) {\n",
              "      const quickchartButtonEl =\n",
              "        document.querySelector('#' + key + ' button');\n",
              "      quickchartButtonEl.disabled = true;  // To prevent multiple clicks.\n",
              "      quickchartButtonEl.classList.add('colab-df-spinner');\n",
              "      try {\n",
              "        const charts = await google.colab.kernel.invokeFunction(\n",
              "            'suggestCharts', [key], {});\n",
              "      } catch (error) {\n",
              "        console.error('Error during call to suggestCharts:', error);\n",
              "      }\n",
              "      quickchartButtonEl.classList.remove('colab-df-spinner');\n",
              "      quickchartButtonEl.classList.add('colab-df-quickchart-complete');\n",
              "    }\n",
              "    (() => {\n",
              "      let quickchartButtonEl =\n",
              "        document.querySelector('#df-a2970215-9ad7-4add-8882-8ab4fe349a85 button');\n",
              "      quickchartButtonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "    })();\n",
              "  </script>\n",
              "</div>\n",
              "    </div>\n",
              "  </div>\n"
            ],
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "dataframe",
              "summary": "{\n  \"name\": \")\",\n  \"rows\": 14,\n  \"fields\": [\n    {\n      \"column\": \"epoch\",\n      \"properties\": {\n        \"dtype\": \"number\",\n        \"std\": 4,\n        \"min\": 1,\n        \"max\": 14,\n        \"num_unique_values\": 14,\n        \"samples\": [\n          10,\n          12,\n          1\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    },\n    {\n      \"column\": \"train_loss\",\n      \"properties\": {\n        \"dtype\": \"number\",\n        \"std\": 0.10494132493218161,\n        \"min\": 0.00033662353343970607,\n        \"max\": 0.4000417148694396,\n        \"num_unique_values\": 14,\n        \"samples\": [\n          0.036644862263347025,\n          0.0037191421099123543,\n          0.4000417148694396\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    },\n    {\n      \"column\": \"train_acc\",\n      \"properties\": {\n        \"dtype\": \"number\",\n        \"std\": 0.0478305198392267,\n        \"min\": 0.815625,\n        \"max\": 1.0,\n        \"num_unique_values\": 11,\n        \"samples\": [\n          0.9890625,\n          0.815625,\n          0.99921875\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    },\n    {\n      \"column\": \"lr\",\n      \"properties\": {\n        \"dtype\": \"number\",\n        \"std\": 7.032061045426913e-21,\n        \"min\": 3e-05,\n        \"max\": 3e-05,\n        \"num_unique_values\": 1,\n        \"samples\": [\n          3e-05\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    },\n    {\n      \"column\": \"val_loss\",\n      \"properties\": {\n        \"dtype\": \"number\",\n        \"std\": 0.13612279774108013,\n        \"min\": 0.29322877824306487,\n        \"max\": 0.7394324883818626,\n        \"num_unique_values\": 14,\n        \"samples\": [\n          0.5678697794675827\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    },\n    {\n      \"column\": \"val_acc\",\n      \"properties\": {\n        \"dtype\": \"number\",\n        \"std\": 0.01783631168215775,\n        \"min\": 0.828125,\n        \"max\": 0.890625,\n        \"num_unique_values\": 12,\n        \"samples\": [\n          0.88125\n        ],\n        \"semantic_type\": \"\",\n        \"description\": \"\"\n      }\n    }\n  ]\n}"
            }
          },
          "metadata": {},
          "execution_count": 42
        }
      ],
      "source": [
        "keras_model.fit(\n",
        "    train_data = dl_train,\n",
        "    val_data= dl_val,\n",
        "    ckpt_path='bert_waimai.pt',\n",
        "    epochs=20,\n",
        "    patience=10,\n",
        "    monitor=\"val_acc\",\n",
        "    mode=\"max\",\n",
        "    plot = True,\n",
        "    wandb = False,\n",
        "    quiet = True\n",
        ")\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "qnd93wyElkCh"
      },
      "source": [
        "## 四，评估模型"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "LJQb0j3xlkCh"
      },
      "source": [
        "可以使用huggingFace的evaluate库来进行模型评估。\n",
        "\n",
        "通过evaluate的load方法可以加载一些常用的评估指标。\n",
        "\n",
        "可以用add_batch逐批次地往这些评估指标上添加数据，最后用compute计算评估结果。"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 43,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "qSxYOgaglkCi",
        "outputId": "6894a5c6-365f-47c0-f6fa-5d6a3b6b0806"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Collecting evaluate\n",
            "  Downloading evaluate-0.4.1-py3-none-any.whl (84 kB)\n",
            "\u001b[?25l     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m0.0/84.1 kB\u001b[0m \u001b[31m?\u001b[0m eta \u001b[36m-:--:--\u001b[0m\r\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m84.1/84.1 kB\u001b[0m \u001b[31m3.0 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[?25hRequirement already satisfied: datasets>=2.0.0 in /usr/local/lib/python3.10/dist-packages (from evaluate) (2.18.0)\n",
            "Requirement already satisfied: numpy>=1.17 in /usr/local/lib/python3.10/dist-packages (from evaluate) (1.25.2)\n",
            "Requirement already satisfied: dill in /usr/local/lib/python3.10/dist-packages (from evaluate) (0.3.8)\n",
            "Requirement already satisfied: pandas in /usr/local/lib/python3.10/dist-packages (from evaluate) (2.0.3)\n",
            "Requirement already satisfied: requests>=2.19.0 in /usr/local/lib/python3.10/dist-packages (from evaluate) (2.31.0)\n",
            "Requirement already satisfied: tqdm>=4.62.1 in /usr/local/lib/python3.10/dist-packages (from evaluate) (4.66.2)\n",
            "Requirement already satisfied: xxhash in /usr/local/lib/python3.10/dist-packages (from evaluate) (3.4.1)\n",
            "Requirement already satisfied: multiprocess in /usr/local/lib/python3.10/dist-packages (from evaluate) (0.70.16)\n",
            "Requirement already satisfied: fsspec[http]>=2021.05.0 in /usr/local/lib/python3.10/dist-packages (from evaluate) (2023.6.0)\n",
            "Requirement already satisfied: huggingface-hub>=0.7.0 in /usr/local/lib/python3.10/dist-packages (from evaluate) (0.20.3)\n",
            "Requirement already satisfied: packaging in /usr/local/lib/python3.10/dist-packages (from evaluate) (24.0)\n",
            "Collecting responses<0.19 (from evaluate)\n",
            "  Downloading responses-0.18.0-py3-none-any.whl (38 kB)\n",
            "Requirement already satisfied: filelock in /usr/local/lib/python3.10/dist-packages (from datasets>=2.0.0->evaluate) (3.13.3)\n",
            "Requirement already satisfied: pyarrow>=12.0.0 in /usr/local/lib/python3.10/dist-packages (from datasets>=2.0.0->evaluate) (14.0.2)\n",
            "Requirement already satisfied: pyarrow-hotfix in /usr/local/lib/python3.10/dist-packages (from datasets>=2.0.0->evaluate) (0.6)\n",
            "Requirement already satisfied: aiohttp in /usr/local/lib/python3.10/dist-packages (from datasets>=2.0.0->evaluate) (3.9.3)\n",
            "Requirement already satisfied: pyyaml>=5.1 in /usr/local/lib/python3.10/dist-packages (from datasets>=2.0.0->evaluate) (6.0.1)\n",
            "Requirement already satisfied: typing-extensions>=3.7.4.3 in /usr/local/lib/python3.10/dist-packages (from huggingface-hub>=0.7.0->evaluate) (4.10.0)\n",
            "Requirement already satisfied: charset-normalizer<4,>=2 in /usr/local/lib/python3.10/dist-packages (from requests>=2.19.0->evaluate) (3.3.2)\n",
            "Requirement already satisfied: idna<4,>=2.5 in /usr/local/lib/python3.10/dist-packages (from requests>=2.19.0->evaluate) (3.6)\n",
            "Requirement already satisfied: urllib3<3,>=1.21.1 in /usr/local/lib/python3.10/dist-packages (from requests>=2.19.0->evaluate) (2.0.7)\n",
            "Requirement already satisfied: certifi>=2017.4.17 in /usr/local/lib/python3.10/dist-packages (from requests>=2.19.0->evaluate) (2024.2.2)\n",
            "Requirement already satisfied: python-dateutil>=2.8.2 in /usr/local/lib/python3.10/dist-packages (from pandas->evaluate) (2.8.2)\n",
            "Requirement already satisfied: pytz>=2020.1 in /usr/local/lib/python3.10/dist-packages (from pandas->evaluate) (2023.4)\n",
            "Requirement already satisfied: tzdata>=2022.1 in /usr/local/lib/python3.10/dist-packages (from pandas->evaluate) (2024.1)\n",
            "Requirement already satisfied: aiosignal>=1.1.2 in /usr/local/lib/python3.10/dist-packages (from aiohttp->datasets>=2.0.0->evaluate) (1.3.1)\n",
            "Requirement already satisfied: attrs>=17.3.0 in /usr/local/lib/python3.10/dist-packages (from aiohttp->datasets>=2.0.0->evaluate) (23.2.0)\n",
            "Requirement already satisfied: frozenlist>=1.1.1 in /usr/local/lib/python3.10/dist-packages (from aiohttp->datasets>=2.0.0->evaluate) (1.4.1)\n",
            "Requirement already satisfied: multidict<7.0,>=4.5 in /usr/local/lib/python3.10/dist-packages (from aiohttp->datasets>=2.0.0->evaluate) (6.0.5)\n",
            "Requirement already satisfied: yarl<2.0,>=1.0 in /usr/local/lib/python3.10/dist-packages (from aiohttp->datasets>=2.0.0->evaluate) (1.9.4)\n",
            "Requirement already satisfied: async-timeout<5.0,>=4.0 in /usr/local/lib/python3.10/dist-packages (from aiohttp->datasets>=2.0.0->evaluate) (4.0.3)\n",
            "Requirement already satisfied: six>=1.5 in /usr/local/lib/python3.10/dist-packages (from python-dateutil>=2.8.2->pandas->evaluate) (1.16.0)\n",
            "Installing collected packages: responses, evaluate\n",
            "Successfully installed evaluate-0.4.1 responses-0.18.0\n"
          ]
        }
      ],
      "source": [
        "!pip install evaluate"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 46,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "nFiwXx3DlkCi",
        "outputId": "a73feeb6-f3f4-4b40-9a31-b7b18f9f9543"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "{'accuracy': 0.8975}"
            ]
          },
          "metadata": {},
          "execution_count": 46
        }
      ],
      "source": [
        "import evaluate\n",
        "metric = evaluate.load(\"accuracy\")\n",
        "model.eval()\n",
        "model = model.to(device)\n",
        "\n",
        "dl_test = keras_model.accelerator.prepare(dl_test)\n",
        "for batch in dl_test:\n",
        "    with torch.no_grad():\n",
        "        outputs = model(**batch.to(device))\n",
        "\n",
        "    logits = outputs.logits\n",
        "    predictions = torch.argmax(logits, dim=-1)\n",
        "    metric.add_batch(predictions=predictions, references=batch[\"labels\"])\n",
        "\n",
        "metric.compute()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "wQieWgAglkCi"
      },
      "source": [
        "## 五，使用模型"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 47,
      "metadata": {
        "id": "C4g4OBtllkCi"
      },
      "outputs": [],
      "source": [
        "texts = [\"今天师傅是不是手抖了，微辣格外辣 \",\"这他妈也太难吃了吧\",\"没有送水没有送水没有送水\",\"还行吧\"]\n",
        "batch = tokenizer(texts, padding=True, return_tensors=\"pt\")\n",
        "batch = {k:v.to(keras_model.accelerator.device) for k,v in batch.items()}"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 48,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "SD1dwTUFlkCj",
        "outputId": "8693260c-8bc3-4eed-d5d5-57dc58f17c55"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "tensor([2.2612e-02, 7.8786e-04, 5.4986e-04, 9.7478e-01], device='cuda:0',\n",
            "       grad_fn=<SelectBackward0>)\n"
          ]
        }
      ],
      "source": [
        "from torch import nn\n",
        "logits = model(**batch).logits\n",
        "scores = nn.Softmax(dim=-1)(logits)[:,-1]\n",
        "print(scores)\n",
        "#可以看到得分与人的预期是高度一致的"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "MOtcTaYJlkCj"
      },
      "source": [
        "也可以用pipeline把tokenizer和model组装在一起"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 49,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Gp_9hpXJlkCj",
        "outputId": "9e57c2fc-b034-4db7-db1e-4371f627f1bd"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[{'label': 'LABEL_1', 'score': 0.9840770959854126}]"
            ]
          },
          "metadata": {},
          "execution_count": 49
        }
      ],
      "source": [
        "from transformers import pipeline\n",
        "classifier = pipeline(task=\"text-classification\", tokenizer=tokenizer, model=model.cpu())\n",
        "classifier(\"挺好吃的哦\")"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "DeDvgFEllkCk"
      },
      "outputs": [],
      "source": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "xiznE6K6lkCk"
      },
      "source": [
        "## 六，保存模型"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "EhBI24iNlkCk"
      },
      "source": [
        "保存model和tokenizer之后，我们可以用一个pipeline加载，并进行批量预测。\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 50,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "y9P6WsIGlkCk",
        "outputId": "ce7c244b-3946-48cc-f4bb-13b69bc75faa"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "('waimai_10k_bert/tokenizer_config.json',\n",
              " 'waimai_10k_bert/special_tokens_map.json',\n",
              " 'waimai_10k_bert/vocab.txt',\n",
              " 'waimai_10k_bert/added_tokens.json',\n",
              " 'waimai_10k_bert/tokenizer.json')"
            ]
          },
          "metadata": {},
          "execution_count": 50
        }
      ],
      "source": [
        "model.config.id2label = {0:\"差评\",1:\"好评\"}\n",
        "model.save_pretrained(\"waimai_10k_bert\")\n",
        "tokenizer.save_pretrained(\"waimai_10k_bert\")"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 51,
      "metadata": {
        "id": "MsNkC9p6lkCl"
      },
      "outputs": [],
      "source": [
        "from transformers import pipeline\n",
        "classifier = pipeline(\"text-classification\",model=\"waimai_10k_bert\")\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 52,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Bc22iDgnlkCl",
        "outputId": "cc29d235-ebad-429a-b31d-4710200c1a28"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[{'label': '好评', 'score': 0.9777551293373108},\n",
              " {'label': '差评', 'score': 0.9994437098503113}]"
            ]
          },
          "metadata": {},
          "execution_count": 52
        }
      ],
      "source": [
        "classifier([\"味道还不错，下次再来\",\"我去，吃了我吐了三天\"])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 53,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "X3RzZIV-lkCl",
        "outputId": "6571f5cd-e85a-4348-c499-0fa92e49d8eb"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[{'label': '好评', 'score': 0.9777470231056213},\n",
              " {'label': '差评', 'score': 0.8687992691993713}]"
            ]
          },
          "metadata": {},
          "execution_count": 53
        }
      ],
      "source": [
        "classifier([\"用户：你好，我想买点理财\",\"想鼓励一下作者杀人\"])"
      ]
    },
    {
      "cell_type": "code",
      "source": [],
      "metadata": {
        "id": "zCtv_nN17b71"
      },
      "execution_count": null,
      "outputs": []
    }
  ],
  "metadata": {
    "kernelspec": {
      "display_name": "Python 3",
      "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": "3.9.0"
    },
    "colab": {
      "provenance": [],
      "gpuType": "T4",
      "include_colab_link": true
    },
    "accelerator": "GPU",
    "widgets": {
      "application/vnd.jupyter.widget-state+json": {
        "5e17e92f944c4baba8f833dc39536f8f": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HBoxModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HBoxModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HBoxView",
            "box_style": "",
            "children": [
              "IPY_MODEL_b118110b6c1a4e35a19e8e4f7e4aa66e",
              "IPY_MODEL_14905759328a47d0a32f94344c741a5a",
              "IPY_MODEL_23471a88434642798422fb3dd147e8ed"
            ],
            "layout": "IPY_MODEL_32f6f165951b433082b30d795f7d8c98"
          }
        },
        "b118110b6c1a4e35a19e8e4f7e4aa66e": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HTMLModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HTMLView",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_6902740078df4375bc31a6fc630cbd7e",
            "placeholder": "​",
            "style": "IPY_MODEL_34c652c518f84dc18a7444b4ed6c4fee",
            "value": "Map (num_proc=4): 100%"
          }
        },
        "14905759328a47d0a32f94344c741a5a": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "FloatProgressModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "ProgressView",
            "bar_style": "success",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_a9a9f8db1b804941b88ee1987bd6f4c6",
            "max": 2000,
            "min": 0,
            "orientation": "horizontal",
            "style": "IPY_MODEL_0a6e5118777348eea83ad7c38cf9fe29",
            "value": 2000
          }
        },
        "23471a88434642798422fb3dd147e8ed": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HTMLModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HTMLView",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_f923393b93554bc48360f415004f8560",
            "placeholder": "​",
            "style": "IPY_MODEL_3731ccc31e1a4349b52c2a6c4a91e0a8",
            "value": " 2000/2000 [00:03&lt;00:00, 3634.23 examples/s]"
          }
        },
        "32f6f165951b433082b30d795f7d8c98": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "6902740078df4375bc31a6fc630cbd7e": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "34c652c518f84dc18a7444b4ed6c4fee": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "DescriptionStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "description_width": ""
          }
        },
        "a9a9f8db1b804941b88ee1987bd6f4c6": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "0a6e5118777348eea83ad7c38cf9fe29": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "ProgressStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "ProgressStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "bar_color": null,
            "description_width": ""
          }
        },
        "f923393b93554bc48360f415004f8560": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "3731ccc31e1a4349b52c2a6c4a91e0a8": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "DescriptionStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "description_width": ""
          }
        }
      }
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}