1,保存模型:
my_model = create_model_function( ...... )
my_model.compile( ...... )
my_model.fit( ...... )
model_name . save( filepath, overwrite: bool=True, include_optimizer: bool=True )
filepath:保存的路径
overwrite:如果存在源文件,是否覆盖
include_optimizer:是否保存优化器状态
ex : mymodel.save(filepath="p402/my_model.h5", includeoptimizer=False)
2, 载入模型:
my_model = keras . models . load_model( filepath )
载入后可以继续训练:
my_model . fit( X_train_2,Y_train_2 )
也可以直接评估:
preds = my_model . evaluate( X_test, Y_test )
print ( "Loss = " + str( preds[0] ) )
print ( "Test Accuracy = " + str( preds[1] ) )
————————————————
原文链接:https://blog.csdn.net/wslkd0123/article/details/80647041
原文:https://www.cnblogs.com/qingchen-forever/p/12922307.html