Keras: How directly get BestModel, not by saving?

[shlpu]

As the title say, thank you!

Checkpoint the weights for best model

from keras.callbacks import ModelCheckpoint

checkpoint

filepath="weights.best.hdf5"
checkpoint = ModelCheckpoint(filepath, monitor='val_acc', verbose=1, save_best_only=True, mode='max')
callbacks = [checkpoint]

Fit the model

model.fit(X, Y, validation_split, epochs, batch_size, callbacks=callbacks, verbose=0)

[shlpu]

Are there anybody could HELP me? Really thanks!

[shlpu]

Of course, we could save and load timely, the question is for the custom layer/model (not built-in), somwhat inconvenient, really expect your reply.

[shlpu]

Any kind advices would be grateful.