keras_回调函数总结

keras回调函数总结了实际项目过程中可能会出现的种种问题,并作出了针对性的措施,十分的使用,故记录了下来,日后如果官方有新的API,我也会定期更新.

class BaseLogger: 累计评估值的回调函数

Callback that accumulates epoch averages of metrics.

class CSVLogger: 将每次迭代的流数据保存到csv文件的回调函数

Callback that streams epoch results to a csv file.

class Callback: 用于生成新回调的抽象基类

Abstract base class used to build new callbacks.

class EarlyStopping: 当监控量停止改善时停止培训

Stop training when a monitored quantity has stopped improving.

class History: 记录日志到History的回调函数

Callback that records events into a History object.

class LambdaCallback: 用于动态创建简单的自定义回调的回调函数

Callback for creating simple, custom callbacks on-the-fly.

class LearningRateScheduler: 学习率调度程序

Learning rate scheduler.

class ModelCheckpoint: 在每个epoch之后保存模型

Save the model after every epoch.

class ProgbarLogger: 将评估值打印到stdout的回调函数

Callback that prints metrics to stdout.

class ReduceLROnPlateau: 当一个评估停止改进时降低学习率

Reduce learning rate when a metric has stopped improving.

class RemoteMonitor: 用于将事件流式传输到服务器的回调

Callback used to stream events to a server.

class TensorBoard: 启用TensorBoard的可视化

Enable visualizations for TensorBoard.

class TerminateOnNaN: 当遇到NaN丢失时终止培训的回调

Callback that terminates training when a NaN loss is encountered.