org.deeplearning4j.earlystopping.listener

Interface EarlyStoppingListener<T extends Model>

Type Parameters:

T - Type of model. For example, MultiLayerNetwork or ComputationGraph

public interface EarlyStoppingListener<T extends Model>

EarlyStoppingListener is a listener interface for conducting early stopping training. It provides onStart, onEpoch, and onCompletion methods, which are called as appropriate

Method Summary

Modifier and Type	Method and Description
void	onCompletion(EarlyStoppingResult<T> esResult) Method that is called at the end of early stopping training
void	onEpoch(int epochNum, double score, EarlyStoppingConfiguration<T> esConfig, T net) Method that is called at the end of each epoch completed during early stopping training
void	onStart(EarlyStoppingConfiguration<T> esConfig, T net) Method to be called when early stopping training is first started

Method Detail

onStart

void onStart(EarlyStoppingConfiguration<T> esConfig,
             T net)

Method to be called when early stopping training is first started

onEpoch

void onEpoch(int epochNum,
             double score,
             EarlyStoppingConfiguration<T> esConfig,
             T net)

Method that is called at the end of each epoch completed during early stopping training

Parameters:

epochNum - The number of the epoch just completed (starting at 0)

score - The score calculated

esConfig - Configuration

net - Network (current)

onCompletion

void onCompletion(EarlyStoppingResult<T> esResult)

Method that is called at the end of early stopping training

Parameters:

esResult - The early stopping result. Provides details of why early stopping training was terminated, etc