tf调不到keras怎么 回事_tf.keras之回调函数

tf.keras的model在进行fit时可以传入各种回调函数，介绍几种常用的回调函数。

EarlyStopping

monitor='val_loss',monitor用来告知需要监听的变量，当该变量满足某些要求之后，就会停止训练，提前结束

min_delta=0, 判定变量不在提升的最小值，当变量的变动小于该值，就认为变量已经不再提升

patience=0, 该数字表示，经过多少轮之后，变量不再变动，则停止训练

verbose=0,这是个日志级别

mode='auto',由于监听的变量可能是准确率，所以变量提升的方向不确定，一般就设置为auto

baseline=None,要监控的变量的基准值。 如果模型没有显示baseline的改善，训练将停止。

restore_best_weights=False，模型是否根据监控变量的最佳值时的model存储下来

LearningRateScheduler

LearningRateScheduler是学习率调整策略，有以下两个参数

schedule 一个function，接受epoch参数，然后我们可以根据epoch参数，返回学习率

verbose，日志

ReduceLROnPlateau

monitor: 监听的变量

factor: 衰减因子，每次衰减都乘以这个数，factor by which the learning rate will be reduced. new_lr = lr *

factor

patience: 变量多少轮没有提升之后，减少学习率

verbose: int. 0: quiet, 1: update messages.

mode: one of {auto, min, max}. In min mode, lr will be reduced when the

quantity monitored has stopped decreasing; in max mode it will be

reduced when the quantity monitored has stopped increasing; in auto

mode, the direction is automatically inferred from the name of the

monitored quantity.

min_delta: 这是用来衡量监控值是否提升的阈值.

cooldown: 学习率衰减之后，会等几轮再重新判定是否衰减number of epochs to wait before resuming normal operation after

lr has been reduced.

min_lr:学习率的下届

LambdaCallback

接受如下六个参数，这几个参数都是匿名函数：

on_epoch_begin: called at the beginning of every epoch.

on_epoch_end: called at the end of every epoch.

on_batch_begin: called at the beginning of every batch.

on_batch_end: called at the end of every batch.

on_train_begin: called at the beginning of model training.

on_train_end: called at the end of model training.`

匿名函数的参数如下：

on_epoch_beginandon_epoch_endexpect two positional arguments:epoch,logs-on_batch_beginandon_batch_endexpect two positional arguments:batch,logs-on_train_beginandon_train_endexpect one positional argument:logs`

自定义CallBack

自定义callback继承父类即可，可以自己实现复杂的回调函数