chapter11_part02_sequence-models.ipynb

{
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0A49wbWHJvID"
      },
      "source": [
        "이 노트북은 [케라스 창시자에게 배우는 딥러닝 2판](https://tensorflow.blog/kerasdl2/)의 예제 코드를 담고 있습니다.\n",
        "\n",
        "<table align=\"left\">\n",
        "    <tr>\n",
        "        <td>\n",
        "            <a href=\"https://colab.research.google.com/github/rickiepark/deep-learning-with-python-2nd/blob/main/chapter11_part02_sequence-models.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>\n",
        "        </td>\n",
        "    </tr>\n",
        "</table>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "5Xd_jva3JvIF"
      },
      "source": [
        "### 단어를 시퀀스로 처리하기: 시퀀스 모델 방식"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "gJDW7N7JJvIG"
      },
      "source": [
        "#### 첫 번째 예제"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "oBcqxVV3JvIG"
      },
      "source": [
        "**데이터 다운로드**"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 1,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "vM4VpJ7KJvIH",
        "outputId": "cfab1e01-31b0-48e0-bc52-1cea355fc05d"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "rm: cannot remove 'aclImdb': No such file or directory\n",
            "  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current\n",
            "                                 Dload  Upload   Total   Spent    Left  Speed\n",
            "100 80.2M  100 80.2M    0     0  8265k      0  0:00:09  0:00:09 --:--:-- 16.2M\n"
          ]
        }
      ],
      "source": [
        "!rm -r aclImdb\n",
        "!curl -O https://ai.stanford.edu/~amaas/data/sentiment/aclImdb_v1.tar.gz\n",
        "!tar -xf aclImdb_v1.tar.gz\n",
        "!rm -r aclImdb/train/unsup"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ztDHYG4gJvII"
      },
      "source": [
        "**데이터 준비**"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 2,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "azorKPW0JvII",
        "outputId": "4dce51bf-47a0-47c7-ffad-5ad49abff445"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Found 20000 files belonging to 2 classes.\n",
            "Found 5000 files belonging to 2 classes.\n",
            "Found 25000 files belonging to 2 classes.\n"
          ]
        }
      ],
      "source": [
        "import os, pathlib, shutil, random\n",
        "from tensorflow import keras\n",
        "batch_size = 32\n",
        "base_dir = pathlib.Path(\"aclImdb\")\n",
        "val_dir = base_dir / \"val\"\n",
        "train_dir = base_dir / \"train\"\n",
        "\n",
        "for category in (\"neg\", \"pos\"):\n",
        "    os.makedirs(val_dir / category)\n",
        "    files = os.listdir(train_dir / category)\n",
        "    random.Random(1337).shuffle(files)\n",
        "    num_val_samples = int(0.2 * len(files))\n",
        "    val_files = files[-num_val_samples:]\n",
        "    for fname in val_files:\n",
        "        shutil.move(train_dir / category / fname,\n",
        "                    val_dir / category / fname)\n",
        "\n",
        "train_ds = keras.utils.text_dataset_from_directory(\n",
        "    \"aclImdb/train\", batch_size=batch_size\n",
        ")\n",
        "val_ds = keras.utils.text_dataset_from_directory(\n",
        "    \"aclImdb/val\", batch_size=batch_size\n",
        ")\n",
        "test_ds = keras.utils.text_dataset_from_directory(\n",
        "    \"aclImdb/test\", batch_size=batch_size\n",
        ")\n",
        "text_only_train_ds = train_ds.map(lambda x, y: x)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Zz4tz-n4JvIJ"
      },
      "source": [
        "**정수 시퀀스 데이터셋 준비하기**"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 3,
      "metadata": {
        "id": "iuPqw9cqJvIJ"
      },
      "outputs": [],
      "source": [
        "from tensorflow.keras import layers\n",
        "\n",
        "max_length = 600\n",
        "max_tokens = 20000\n",
        "text_vectorization = layers.TextVectorization(\n",
        "    max_tokens=max_tokens,\n",
        "    output_mode=\"int\",\n",
        "    output_sequence_length=max_length,\n",
        ")\n",
        "text_vectorization.adapt(text_only_train_ds)\n",
        "\n",
        "int_train_ds = train_ds.map(\n",
        "    lambda x, y: (text_vectorization(x), y),\n",
        "    num_parallel_calls=4)\n",
        "int_val_ds = val_ds.map(\n",
        "    lambda x, y: (text_vectorization(x), y),\n",
        "    num_parallel_calls=4)\n",
        "int_test_ds = test_ds.map(\n",
        "    lambda x, y: (text_vectorization(x), y),\n",
        "    num_parallel_calls=4)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "7rCBxfAFJvIK"
      },
      "source": [
        "**원-핫 인코딩된 벡터 시퀀스로 시퀀스 모델 만들기**"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 4,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "3rai-1TDJvIK",
        "outputId": "1979506e-1a87-41f1-d213-76f15283b424"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"model\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " input_1 (InputLayer)        [(None, None)]            0         \n",
            "                                                                 \n",
            " tf.one_hot (TFOpLambda)     (None, None, 20000)       0         \n",
            "                                                                 \n",
            " bidirectional (Bidirectiona  (None, 64)               5128448   \n",
            " l)                                                              \n",
            "                                                                 \n",
            " dropout (Dropout)           (None, 64)                0         \n",
            "                                                                 \n",
            " dense (Dense)               (None, 1)                 65        \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 5,128,513\n",
            "Trainable params: 5,128,513\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        }
      ],
      "source": [
        "import tensorflow as tf\n",
        "inputs = keras.Input(shape=(None,), dtype=\"int64\")\n",
        "embedded = tf.one_hot(inputs, depth=max_tokens)\n",
        "x = layers.Bidirectional(layers.LSTM(32))(embedded)\n",
        "x = layers.Dropout(0.5)(x)\n",
        "outputs = layers.Dense(1, activation=\"sigmoid\")(x)\n",
        "model = keras.Model(inputs, outputs)\n",
        "model.compile(optimizer=\"rmsprop\",\n",
        "              loss=\"binary_crossentropy\",\n",
        "              metrics=[\"accuracy\"])\n",
        "model.summary()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "TeqpjAQoJvIL"
      },
      "source": [
        "**첫 번째 시퀀스 모델 훈련하기**"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 5,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Bl6MtP3NJvIL",
        "outputId": "54cf1023-a8a1-41f9-aa09-7b0037c7eede"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/10\n",
            "625/625 [==============================] - 52s 68ms/step - loss: 0.5051 - accuracy: 0.7664 - val_loss: 0.3761 - val_accuracy: 0.8476\n",
            "Epoch 2/10\n",
            "625/625 [==============================] - 42s 67ms/step - loss: 0.3206 - accuracy: 0.8832 - val_loss: 0.2990 - val_accuracy: 0.8834\n",
            "Epoch 3/10\n",
            "625/625 [==============================] - 42s 66ms/step - loss: 0.2522 - accuracy: 0.9112 - val_loss: 0.3217 - val_accuracy: 0.8664\n",
            "Epoch 4/10\n",
            "625/625 [==============================] - 42s 66ms/step - loss: 0.2222 - accuracy: 0.9237 - val_loss: 0.3348 - val_accuracy: 0.8802\n",
            "Epoch 5/10\n",
            "625/625 [==============================] - 42s 66ms/step - loss: 0.1959 - accuracy: 0.9352 - val_loss: 0.3155 - val_accuracy: 0.8882\n",
            "Epoch 6/10\n",
            "625/625 [==============================] - 42s 66ms/step - loss: 0.1660 - accuracy: 0.9438 - val_loss: 0.3365 - val_accuracy: 0.8812\n",
            "Epoch 7/10\n",
            "625/625 [==============================] - 42s 67ms/step - loss: 0.1515 - accuracy: 0.9487 - val_loss: 0.4601 - val_accuracy: 0.8840\n",
            "Epoch 8/10\n",
            "625/625 [==============================] - 42s 67ms/step - loss: 0.1335 - accuracy: 0.9560 - val_loss: 0.3583 - val_accuracy: 0.8786\n",
            "Epoch 9/10\n",
            "625/625 [==============================] - 42s 67ms/step - loss: 0.1212 - accuracy: 0.9602 - val_loss: 0.4242 - val_accuracy: 0.8868\n",
            "Epoch 10/10\n",
            "625/625 [==============================] - 42s 67ms/step - loss: 0.1075 - accuracy: 0.9671 - val_loss: 0.4790 - val_accuracy: 0.8790\n",
            "782/782 [==============================] - 28s 35ms/step - loss: 0.3039 - accuracy: 0.8774\n",
            "테스트 정확도: 0.877\n"
          ]
        }
      ],
      "source": [
        "callbacks = [\n",
        "    keras.callbacks.ModelCheckpoint(\"one_hot_bidir_lstm.keras\",\n",
        "                                    save_best_only=True)\n",
        "]\n",
        "model.fit(int_train_ds, validation_data=int_val_ds, epochs=10, callbacks=callbacks)\n",
        "model = keras.models.load_model(\"one_hot_bidir_lstm.keras\")\n",
        "print(f\"테스트 정확도: {model.evaluate(int_test_ds)[1]:.3f}\")"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "cVY_SwScJvIL"
      },
      "source": [
        "#### 단어 임베딩 이해하기"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "visueffQJvIL"
      },
      "source": [
        "#### 임베딩 층으로 단어 임베딩 학습하기"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "rH2EhPHUJvIM"
      },
      "source": [
        "**`Embedding` 층 만들기**"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 6,
      "metadata": {
        "id": "LNACHsDAJvIM"
      },
      "outputs": [],
      "source": [
        "embedding_layer = layers.Embedding(input_dim=max_tokens, output_dim=256)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "MfBr2Ld8JvIM"
      },
      "source": [
        "**밑바닥부터 훈련하는 `Embedding` 층을 사용한 모델**"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 7,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "NHbW8s3QJvIM",
        "outputId": "9e40e2e8-a039-40f9-d863-85a6099c1b97"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"model_1\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " input_2 (InputLayer)        [(None, None)]            0         \n",
            "                                                                 \n",
            " embedding_1 (Embedding)     (None, None, 256)         5120000   \n",
            "                                                                 \n",
            " bidirectional_1 (Bidirectio  (None, 64)               73984     \n",
            " nal)                                                            \n",
            "                                                                 \n",
            " dropout_1 (Dropout)         (None, 64)                0         \n",
            "                                                                 \n",
            " dense_1 (Dense)             (None, 1)                 65        \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 5,194,049\n",
            "Trainable params: 5,194,049\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n",
            "Epoch 1/10\n",
            "625/625 [==============================] - 37s 54ms/step - loss: 0.4805 - accuracy: 0.7814 - val_loss: 0.4047 - val_accuracy: 0.8448\n",
            "Epoch 2/10\n",
            "625/625 [==============================] - 33s 52ms/step - loss: 0.3092 - accuracy: 0.8865 - val_loss: 0.3327 - val_accuracy: 0.8704\n",
            "Epoch 3/10\n",
            "625/625 [==============================] - 32s 52ms/step - loss: 0.2399 - accuracy: 0.9142 - val_loss: 0.3343 - val_accuracy: 0.8656\n",
            "Epoch 4/10\n",
            "625/625 [==============================] - 32s 52ms/step - loss: 0.2014 - accuracy: 0.9313 - val_loss: 0.3448 - val_accuracy: 0.8748\n",
            "Epoch 5/10\n",
            "625/625 [==============================] - 32s 52ms/step - loss: 0.1693 - accuracy: 0.9424 - val_loss: 0.3367 - val_accuracy: 0.8706\n",
            "Epoch 6/10\n",
            "625/625 [==============================] - 33s 52ms/step - loss: 0.1397 - accuracy: 0.9563 - val_loss: 0.3859 - val_accuracy: 0.8732\n",
            "Epoch 7/10\n",
            "625/625 [==============================] - 32s 52ms/step - loss: 0.1222 - accuracy: 0.9613 - val_loss: 0.3923 - val_accuracy: 0.8638\n",
            "Epoch 8/10\n",
            "625/625 [==============================] - 32s 52ms/step - loss: 0.0963 - accuracy: 0.9700 - val_loss: 0.5025 - val_accuracy: 0.8768\n",
            "Epoch 9/10\n",
            "625/625 [==============================] - 32s 52ms/step - loss: 0.0853 - accuracy: 0.9746 - val_loss: 0.4973 - val_accuracy: 0.8736\n",
            "Epoch 10/10\n",
            "625/625 [==============================] - 32s 52ms/step - loss: 0.0689 - accuracy: 0.9784 - val_loss: 0.4690 - val_accuracy: 0.8792\n",
            "782/782 [==============================] - 21s 25ms/step - loss: 0.3490 - accuracy: 0.8672\n",
            "테스트 정확도: 0.867\n"
          ]
        }
      ],
      "source": [
        "inputs = keras.Input(shape=(None,), dtype=\"int64\")\n",
        "embedded = layers.Embedding(input_dim=max_tokens, output_dim=256)(inputs)\n",
        "x = layers.Bidirectional(layers.LSTM(32))(embedded)\n",
        "x = layers.Dropout(0.5)(x)\n",
        "outputs = layers.Dense(1, activation=\"sigmoid\")(x)\n",
        "model = keras.Model(inputs, outputs)\n",
        "model.compile(optimizer=\"rmsprop\",\n",
        "              loss=\"binary_crossentropy\",\n",
        "              metrics=[\"accuracy\"])\n",
        "model.summary()\n",
        "\n",
        "callbacks = [\n",
        "    keras.callbacks.ModelCheckpoint(\"embeddings_bidir_lstm.keras\",\n",
        "                                    save_best_only=True)\n",
        "]\n",
        "model.fit(int_train_ds, validation_data=int_val_ds, epochs=10, callbacks=callbacks)\n",
        "model = keras.models.load_model(\"embeddings_bidir_lstm.keras\")\n",
        "print(f\"테스트 정확도: {model.evaluate(int_test_ds)[1]:.3f}\")"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "W32FVDl_JvIM"
      },
      "source": [
        "#### 패딩과 마스킹 이해하기"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "M37QJPqIJvIN"
      },
      "source": [
        "**마스킹을 활성화한 `Embedding` 층 사용하기**"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 8,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "VS7crehBJvIN",
        "outputId": "28664896-2074-4ae3-81ef-7d691d4073f3"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"model_2\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " input_3 (InputLayer)        [(None, None)]            0         \n",
            "                                                                 \n",
            " embedding_2 (Embedding)     (None, None, 256)         5120000   \n",
            "                                                                 \n",
            " bidirectional_2 (Bidirectio  (None, 64)               73984     \n",
            " nal)                                                            \n",
            "                                                                 \n",
            " dropout_2 (Dropout)         (None, 64)                0         \n",
            "                                                                 \n",
            " dense_2 (Dense)             (None, 1)                 65        \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 5,194,049\n",
            "Trainable params: 5,194,049\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n",
            "Epoch 1/10\n",
            "625/625 [==============================] - 42s 55ms/step - loss: 0.4013 - accuracy: 0.8180 - val_loss: 0.3657 - val_accuracy: 0.8612\n",
            "Epoch 2/10\n",
            "625/625 [==============================] - 32s 51ms/step - loss: 0.2283 - accuracy: 0.9115 - val_loss: 0.2796 - val_accuracy: 0.8866\n",
            "Epoch 3/10\n",
            "625/625 [==============================] - 32s 51ms/step - loss: 0.1616 - accuracy: 0.9396 - val_loss: 0.3395 - val_accuracy: 0.8804\n",
            "Epoch 4/10\n",
            "625/625 [==============================] - 32s 51ms/step - loss: 0.1214 - accuracy: 0.9570 - val_loss: 0.4593 - val_accuracy: 0.8678\n",
            "Epoch 5/10\n",
            "625/625 [==============================] - 32s 51ms/step - loss: 0.0903 - accuracy: 0.9681 - val_loss: 0.4086 - val_accuracy: 0.8804\n",
            "Epoch 6/10\n",
            "625/625 [==============================] - 32s 51ms/step - loss: 0.0655 - accuracy: 0.9768 - val_loss: 0.4162 - val_accuracy: 0.8750\n",
            "Epoch 7/10\n",
            "625/625 [==============================] - 32s 51ms/step - loss: 0.0487 - accuracy: 0.9837 - val_loss: 0.4813 - val_accuracy: 0.8700\n",
            "Epoch 8/10\n",
            "625/625 [==============================] - 32s 51ms/step - loss: 0.0351 - accuracy: 0.9882 - val_loss: 0.5398 - val_accuracy: 0.8720\n",
            "Epoch 9/10\n",
            "625/625 [==============================] - 32s 52ms/step - loss: 0.0223 - accuracy: 0.9922 - val_loss: 0.5528 - val_accuracy: 0.8688\n",
            "Epoch 10/10\n",
            "625/625 [==============================] - 33s 52ms/step - loss: 0.0158 - accuracy: 0.9951 - val_loss: 0.6536 - val_accuracy: 0.8692\n",
            "782/782 [==============================] - 21s 24ms/step - loss: 0.2886 - accuracy: 0.8807\n",
            "테스트 정확도: 0.881\n"
          ]
        }
      ],
      "source": [
        "inputs = keras.Input(shape=(None,), dtype=\"int64\")\n",
        "embedded = layers.Embedding(\n",
        "    input_dim=max_tokens, output_dim=256, mask_zero=True)(inputs)\n",
        "x = layers.Bidirectional(layers.LSTM(32))(embedded)\n",
        "x = layers.Dropout(0.5)(x)\n",
        "outputs = layers.Dense(1, activation=\"sigmoid\")(x)\n",
        "model = keras.Model(inputs, outputs)\n",
        "model.compile(optimizer=\"rmsprop\",\n",
        "              loss=\"binary_crossentropy\",\n",
        "              metrics=[\"accuracy\"])\n",
        "model.summary()\n",
        "\n",
        "callbacks = [\n",
        "    keras.callbacks.ModelCheckpoint(\"embeddings_bidir_lstm_with_masking.keras\",\n",
        "                                    save_best_only=True)\n",
        "]\n",
        "model.fit(int_train_ds, validation_data=int_val_ds, epochs=10, callbacks=callbacks)\n",
        "model = keras.models.load_model(\"embeddings_bidir_lstm_with_masking.keras\")\n",
        "print(f\"테스트 정확도: {model.evaluate(int_test_ds)[1]:.3f}\")"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "WT_UyLakJvIN"
      },
      "source": [
        "#### 사전 훈련된 단어 임베딩 사용하기"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 9,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "QST9q6Q1JvIN",
        "outputId": "b5694535-885f-4e35-a6f9-3443242301b2"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "--2022-11-19 00:50:43--  http://nlp.stanford.edu/data/glove.6B.zip\n",
            "Resolving nlp.stanford.edu (nlp.stanford.edu)... 171.64.67.140\n",
            "Connecting to nlp.stanford.edu (nlp.stanford.edu)|171.64.67.140|:80... connected.\n",
            "HTTP request sent, awaiting response... 302 Found\n",
            "Location: https://nlp.stanford.edu/data/glove.6B.zip [following]\n",
            "--2022-11-19 00:50:43--  https://nlp.stanford.edu/data/glove.6B.zip\n",
            "Connecting to nlp.stanford.edu (nlp.stanford.edu)|171.64.67.140|:443... connected.\n",
            "HTTP request sent, awaiting response... 301 Moved Permanently\n",
            "Location: https://downloads.cs.stanford.edu/nlp/data/glove.6B.zip [following]\n",
            "--2022-11-19 00:50:44--  https://downloads.cs.stanford.edu/nlp/data/glove.6B.zip\n",
            "Resolving downloads.cs.stanford.edu (downloads.cs.stanford.edu)... 171.64.64.22\n",
            "Connecting to downloads.cs.stanford.edu (downloads.cs.stanford.edu)|171.64.64.22|:443... connected.\n",
            "HTTP request sent, awaiting response... 200 OK\n",
            "Length: 862182613 (822M) [application/zip]\n",
            "Saving to: ‘glove.6B.zip’\n",
            "\n",
            "glove.6B.zip        100%[===================>] 822.24M  5.13MB/s    in 2m 40s  \n",
            "\n",
            "2022-11-19 00:53:24 (5.15 MB/s) - ‘glove.6B.zip’ saved [862182613/862182613]\n",
            "\n"
          ]
        }
      ],
      "source": [
        "!wget http://nlp.stanford.edu/data/glove.6B.zip\n",
        "!unzip -q glove.6B.zip"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "t0aUgmK2JvIN"
      },
      "source": [
        "**GloVe 단어 임베딩 파일 파싱하기**"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 10,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "xZwkvoGYJvIN",
        "outputId": "25b0ce38-664c-4c3b-c503-142608b1fe3c"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "단어 벡터 개수: 400000\n"
          ]
        }
      ],
      "source": [
        "import numpy as np\n",
        "path_to_glove_file = \"glove.6B.100d.txt\"\n",
        "\n",
        "embeddings_index = {}\n",
        "with open(path_to_glove_file) as f:\n",
        "    for line in f:\n",
        "        word, coefs = line.split(maxsplit=1)\n",
        "        coefs = np.fromstring(coefs, \"f\", sep=\" \")\n",
        "        embeddings_index[word] = coefs\n",
        "\n",
        "print(f\"단어 벡터 개수: {len(embeddings_index)}\")"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "hCdliBn7JvIO"
      },
      "source": [
        "**GloVe 단어 임베딩 행렬 준비하기**"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 11,
      "metadata": {
        "id": "HTEMsvf2JvIO"
      },
      "outputs": [],
      "source": [
        "embedding_dim = 100\n",
        "\n",
        "vocabulary = text_vectorization.get_vocabulary()\n",
        "word_index = dict(zip(vocabulary, range(len(vocabulary))))\n",
        "\n",
        "embedding_matrix = np.zeros((max_tokens, embedding_dim))\n",
        "for word, i in word_index.items():\n",
        "    if i < max_tokens:\n",
        "        embedding_vector = embeddings_index.get(word)\n",
        "    if embedding_vector is not None:\n",
        "        embedding_matrix[i] = embedding_vector"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 12,
      "metadata": {
        "id": "PmEvrFkYJvIO"
      },
      "outputs": [],
      "source": [
        "embedding_layer = layers.Embedding(\n",
        "    max_tokens,\n",
        "    embedding_dim,\n",
        "    embeddings_initializer=keras.initializers.Constant(embedding_matrix),\n",
        "    trainable=False,\n",
        "    mask_zero=True,\n",
        ")"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "m24hC78pJvIO"
      },
      "source": [
        "**사전 훈련된 임베딩을 사용하는 모델**"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 13,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "CAdQbaClJvIO",
        "outputId": "bfe9027d-df32-4161-8984-29e5ed843098"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"model_3\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " input_4 (InputLayer)        [(None, None)]            0         \n",
            "                                                                 \n",
            " embedding_3 (Embedding)     (None, None, 100)         2000000   \n",
            "                                                                 \n",
            " bidirectional_3 (Bidirectio  (None, 64)               34048     \n",
            " nal)                                                            \n",
            "                                                                 \n",
            " dropout_3 (Dropout)         (None, 64)                0         \n",
            "                                                                 \n",
            " dense_3 (Dense)             (None, 1)                 65        \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 2,034,113\n",
            "Trainable params: 34,113\n",
            "Non-trainable params: 2,000,000\n",
            "_________________________________________________________________\n",
            "Epoch 1/10\n",
            "625/625 [==============================] - 42s 54ms/step - loss: 0.5644 - accuracy: 0.7024 - val_loss: 0.4789 - val_accuracy: 0.7626\n",
            "Epoch 2/10\n",
            "625/625 [==============================] - 31s 50ms/step - loss: 0.4517 - accuracy: 0.7908 - val_loss: 0.4066 - val_accuracy: 0.8162\n",
            "Epoch 3/10\n",
            "625/625 [==============================] - 30s 47ms/step - loss: 0.4017 - accuracy: 0.8248 - val_loss: 0.5004 - val_accuracy: 0.7774\n",
            "Epoch 4/10\n",
            "625/625 [==============================] - 31s 50ms/step - loss: 0.3730 - accuracy: 0.8392 - val_loss: 0.3753 - val_accuracy: 0.8312\n",
            "Epoch 5/10\n",
            "625/625 [==============================] - 31s 50ms/step - loss: 0.3437 - accuracy: 0.8576 - val_loss: 0.3374 - val_accuracy: 0.8572\n",
            "Epoch 6/10\n",
            "625/625 [==============================] - 31s 50ms/step - loss: 0.3248 - accuracy: 0.8648 - val_loss: 0.3324 - val_accuracy: 0.8544\n",
            "Epoch 7/10\n",
            "625/625 [==============================] - 31s 50ms/step - loss: 0.3072 - accuracy: 0.8702 - val_loss: 0.3197 - val_accuracy: 0.8642\n",
            "Epoch 8/10\n",
            "625/625 [==============================] - 30s 47ms/step - loss: 0.2876 - accuracy: 0.8805 - val_loss: 0.3845 - val_accuracy: 0.8464\n",
            "Epoch 9/10\n",
            "625/625 [==============================] - 29s 46ms/step - loss: 0.2753 - accuracy: 0.8892 - val_loss: 0.3308 - val_accuracy: 0.8672\n",
            "Epoch 10/10\n",
            "625/625 [==============================] - 29s 46ms/step - loss: 0.2642 - accuracy: 0.8956 - val_loss: 0.3412 - val_accuracy: 0.8628\n",
            "782/782 [==============================] - 18s 21ms/step - loss: 0.3110 - accuracy: 0.8661\n",
            "테스트 정확도: 0.866\n"
          ]
        }
      ],
      "source": [
        "inputs = keras.Input(shape=(None,), dtype=\"int64\")\n",
        "embedded = embedding_layer(inputs)\n",
        "x = layers.Bidirectional(layers.LSTM(32))(embedded)\n",
        "x = layers.Dropout(0.5)(x)\n",
        "outputs = layers.Dense(1, activation=\"sigmoid\")(x)\n",
        "model = keras.Model(inputs, outputs)\n",
        "model.compile(optimizer=\"rmsprop\",\n",
        "              loss=\"binary_crossentropy\",\n",
        "              metrics=[\"accuracy\"])\n",
        "model.summary()\n",
        "\n",
        "callbacks = [\n",
        "    keras.callbacks.ModelCheckpoint(\"glove_embeddings_sequence_model.keras\",\n",
        "                                    save_best_only=True)\n",
        "]\n",
        "model.fit(int_train_ds, validation_data=int_val_ds, epochs=10, callbacks=callbacks)\n",
        "model = keras.models.load_model(\"glove_embeddings_sequence_model.keras\")\n",
        "print(f\"테스트 정확도: {model.evaluate(int_test_ds)[1]:.3f}\")"
      ]
    }
  ],
  "metadata": {
    "accelerator": "GPU",
    "colab": {
      "name": "chapter11_part02_sequence-models.i",
      "provenance": []
    },
    "kernelspec": {
      "display_name": "default:Python",
      "language": "python",
      "name": "conda-env-default-py"
    },
    "language_info": {
      "codemirror_mode": {
        "name": "ipython",
        "version": 3
      },
      "file_extension": ".py",
      "mimetype": "text/x-python",
      "name": "python",
      "nbconvert_exporter": "python",
      "pygments_lexer": "ipython3",
      "version": "3.9.10"
    },
    "gpuClass": "premium"
  },
  "nbformat": 4,
  "nbformat_minor": 0
}