lstm python 代码实现

以下是一个简单的LSTM模型的Python代码实现：

import numpy as np
import tensorflow as tf

# 定义超参数
learning_rate = 0.001
training_steps = 10000
batch_size = 128
display_step = 200

# 定义LSTM模型
class LSTM(tf.keras.Model):
    def __init__(self, num_units):
        super(LSTM, self).__init__()
        self.lstm = tf.keras.layers.LSTM(num_units)
        self.fc = tf.keras.layers.Dense(1, activation='sigmoid')
        
    def call(self, x):
        x = self.lstm(x)
        x = self.fc(x)
        return x

# 加载数据
(x_train, y_train), (x_test, y_test) = tf.keras.datasets.mnist.load_data()
x_train = x_train.reshape(-1, 28, 28) / 255.0
x_test = x_test.reshape(-1, 28, 28) / 255.0

# 定义模型和优化器
model = LSTM(128)
optimizer = tf.keras.optimizers.Adam(lr=learning_rate)

# 定义损失函数
def loss_fn(logits, labels):
    return tf.reduce_mean(tf.nn.sigmoid_cross_entropy_with_logits(logits=logits, labels=labels))

# 定义准确率函数
def accuracy_fn(logits, labels):
    preds = tf.cast(tf.greater(tf.sigmoid(logits), 0.5), tf.float32)
    acc = tf.reduce_mean(tf.cast(tf.equal(preds, labels), tf.float32))
    return acc

# 定义训练函数
def train(model, optimizer, x_train, y_train):
    num_batches = x_train.shape[0] // batch_size
    
    for step in range(training_steps):
        indices = np.random.choice(x_train.shape[0], batch_size, replace=False)
        x_batch, y_batch = x_train[indices], y_train[indices]
        
        with tf.GradientTape() as tape:
            logits = model(x_batch)
            loss = loss_fn(logits, y_batch)
            acc = accuracy_fn(logits, y_batch)
            
        gradients = tape.gradient(loss, model.trainable_variables)
        optimizer.apply_gradients(zip(gradients, model.trainable_variables))
        
        if step % display_step == 0:
            print("Step:", step, "Loss:", loss.numpy(), "Accuracy:", acc.numpy())
            
# 开始训练
train(model, optimizer, x_train, y_train)