解决AttributeError: ‘LearningRateScheduler‘ object has no attribute ‘_implements_train_batch_hooks‘问题

今天在搭建神经网络的时候,由于设置了学习率并且每50个epochs都会将学习率更改为原来的1/2,因此设置有如下代码:

from keras.callbacks import LearningRateScheduler

#学习率衰减
def step_decay(epoch):
    if epoch % 50 == 0 and epoch != 0:
        lr = K.get_value(model.optimizer.lr)
        K.set_value(model.optimizer.lr, lr * 0.5)
        print("lr changed to {}".format(lr * 0.5))
    return K.get_value(model.optimizer.lr)

lrate = LearningRateScheduler(step_decay)

model.fit(train_x, train_y, batch_size=32, epochs=300, callbacks=[lrate])

结果发现在运行的过程中一直报错如下:

AttributeError: 'LearningRateScheduler' object has no attribute '_implements_train_batch_hooks'

后来发现是tensorflow与keras版本的问题:

把之前的代码更改为:

#注意在这里进行了更改
from tensorflow.keras.callbacks import LearningRateScheduler

#学习率衰减
def step_decay(epoch):
    if epoch % 50 == 0 and epoch != 0:
        lr = K.get_value(model.optimizer.lr)
        K.set_value(model.optimizer.lr, lr * 0.5)
        print("lr changed to {}".format(lr * 0.5))
    return K.get_value(model.optimizer.lr)

lrate = LearningRateScheduler(step_decay)

model.fit(train_x, train_y, batch_size=32, epochs=300, callbacks=[lrate])

然后错误得到了解决,如有不足,欢迎指正