{
"cells": [
{
"cell_type": "markdown",
"id": "02eca690",
"metadata": {},
"source": [
"# 机器学习与社会科学应用"
]
},
{
"cell_type": "markdown",
"id": "c22dc224",
"metadata": {},
"source": [
"# 第七章 深度学习"
]
},
{
"cell_type": "markdown",
"id": "03a90d78",
"metadata": {},
"source": [
"# 第三节 循环神经网络"
]
},
{
"cell_type": "markdown",
"id": "97e7c62f-6e60-4fc4-9229-c886cd2a596c",
"metadata": {},
"source": [
"郭峰 \n",
" 教授、博士生导师 \n",
"上海财经大学公共经济与管理学院 \n",
"上海财经大学数实融合与智能治理实验室 \n",
" 邮箱：guofengsfi@163.com "
]
},
{
"cell_type": "markdown",
"id": "843ca6fb",
"metadata": {},
"source": [
"本节目录： \n",
"3.1.数据预处理 \n",
"3.2.创建RNN模型 \n",
"3.3.定义损失函数和优化器 \n",
"3.4.配置检查点（模型保存） \n",
"3.5.训练模型 \n",
"3.6.文本生成 "
]
},
{
"cell_type": "markdown",
"id": "d85a10c0",
"metadata": {},
"source": [
"### 3.1. 数据预处理"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "b894f73d",
"metadata": {},
"outputs": [],
"source": [
"import tensorflow as tf\n",
"import numpy as np\n",
"import os\n",
"import random\n",
"import time\n",
"\n",
"\n",
"# 首先，我们需要加载并预处理文本数据\n",
"path_to_file = tf.keras.utils.get_file('shakespeare.txt', 'https://storage.googleapis.com/download.tensorflow.org/data/shakespeare.txt')\n",
"\n",
"text = open(path_to_file, 'rb').read().decode(encoding='utf-8')\n",
"\n",
"# 去除文本中的换行符\n",
"text = text.replace('\\n', ' ')\n",
"\n",
"# 创建文本字典，将字符映射到整数并反向映射\n",
"vocab = sorted(set(text))\n",
"char2idx = {u: i for i, u in enumerate(vocab)}\n",
"idx2char = np.array(vocab)\n",
"text_as_int = np.array([char2idx[c] for c in text])\n",
"\n",
"# 创建训练样本\n",
"seq_length = 100\n",
"examples_per_epoch = len(text) // (seq_length + 1)\n",
"\n",
"char_dataset = tf.data.Dataset.from_tensor_slices(text_as_int)\n",
"\n",
"sequences = char_dataset.batch(seq_length + 1, drop_remainder=True)\n",
"\n",
"def split_input_target(chunk):\n",
" input_text = chunk[:-1]\n",
" target_text = chunk[1:]\n",
" return input_text, target_text\n",
"\n",
"dataset = sequences.map(split_input_target)\n",
"\n",
"BATCH_SIZE = 64\n",
"BUFFER_SIZE = 10000\n",
"\n",
"dataset = dataset.shuffle(BUFFER_SIZE).batch(BATCH_SIZE, drop_remainder=True)"
]
},
{
"cell_type": "markdown",
"id": "3539a145",
"metadata": {},
"source": [
"### 3.2. 创建RNN模型"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "ca0df0e1-49d3-4490-a434-f142f45e92c8",
"metadata": {},
"outputs": [],
"source": [
"\n",
"# 下面我们将创建一个简单的RNN模型\n",
"\n",
"vocab_size = len(vocab)\n",
"embedding_dim = 256\n",
"rnn_units = 1024\n",
"\n",
"def build_model(vocab_size, embedding_dim, rnn_units, batch_size):\n",
" model = tf.keras.Sequential([\n",
" tf.keras.layers.Embedding(vocab_size, embedding_dim, batch_input_shape=[batch_size, None]),\n",
" tf.keras.layers.LSTM(rnn_units, return_sequences=True, stateful=True, recurrent_initializer='glorot_uniform'),\n",
" tf.keras.layers.Dense(vocab_size)\n",
" ])\n",
" return model\n",
"\n",
"model = build_model(\n",
" vocab_size=vocab_size,\n",
" embedding_dim=embedding_dim,\n",
" rnn_units=rnn_units,\n",
" batch_size=BATCH_SIZE\n",
")"
]
},
{
"cell_type": "markdown",
"id": "ee4783df",
"metadata": {},
"source": [
"### 3.3. 定义损失函数和优化器"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "19cdf100",
"metadata": {},
"outputs": [],
"source": [
"\n",
"def loss(labels, logits):\n",
" return tf.keras.losses.sparse_categorical_crossentropy(labels, logits, from_logits=True)\n",
"\n",
"model.compile(optimizer='adam', loss=loss)"
]
},
{
"cell_type": "markdown",
"id": "4f9d8fb3",
"metadata": {},
"source": [
"### 3.4. 配置检查点（模型保存）"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "893c8462",
"metadata": {},
"outputs": [],
"source": [
"\n",
"checkpoint_dir = './training_checkpoints'\n",
"checkpoint_prefix = os.path.join(checkpoint_dir, \"ckpt_{epoch}\")\n",
"\n",
"checkpoint_callback = tf.keras.callbacks.ModelCheckpoint(\n",
" filepath=checkpoint_prefix,\n",
" save_weights_only=True\n",
")"
]
},
{
"cell_type": "markdown",
"id": "ddfab44b",
"metadata": {},
"source": [
"### 3.5. 训练模型"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "a0c0e53f",
"metadata": {},
"outputs": [],
"source": [
"\n",
"EPOCHS = 5\n",
"\n",
"history = model.fit(dataset, epochs=EPOCHS, callbacks=[checkpoint_callback])"
]
},
{
"cell_type": "markdown",
"id": "9c404895",
"metadata": {},
"source": [
"### 3.6. 文本生成"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "9bc65b09",
"metadata": {},
"outputs": [],
"source": [
"\n",
"# 下面是生成文本的代码\n",
"\n",
"model = build_model(vocab_size, embedding_dim, rnn_units, batch_size=1)\n",
"\n",
"model.load_weights(tf.train.latest_checkpoint(checkpoint_dir))\n",
"model.build(tf.TensorShape([1, None]))\n",
"\n",
"def generate_text(model, start_string):\n",
" num_generate = 1000\n",
" input_eval = [char2idx[s] for s in start_string]\n",
" input_eval = tf.expand_dims(input_eval, 0)\n",
" text_generated = []\n",
"\n",
" temperature = 0.5\n",
"\n",
" model.reset_states()\n",
" for i in range(num_generate):\n",
" predictions = model(input_eval)\n",
" predictions = tf.squeeze(predictions, 0)\n",
"\n",
" predictions = predictions / temperature\n",
" predicted_id = tf.random.categorical(predictions, num_samples=1)[-1,0].numpy()\n",
"\n",
" input_eval = tf.expand_dims([predicted_id], 0)\n",
"\n",
" text_generated.append(idx2char[predicted_id])\n",
"\n",
" return start_string + ''.join(text_generated)\n",
"\n",
"# 使用模型生成文本\n",
"generated_text = generate_text(model, start_string=u\"ROMEO: \")\n",
"print(generated_text)"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "52e9ec2c",
"metadata": {},
"outputs": [],
"source": []
}
],
"metadata": {
"kernelspec": {
"display_name": "Python 3 (ipykernel)",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.9.12"
}
},
"nbformat": 4,
"nbformat_minor": 5
}