Callbacks
ModelCheckpoint
keras.callbacks.ModelCheckpoint()
ModelCheckpoint is basically a function that saves the model every epoch (unless
otherwise directed via parameters). How it does so can be configured by the set of
parameters associated with ModelCheckpoint():
•
filepath: The path where you want to save the model file. Typing just
“model_name.h5” saves it in the same directory.
•
monitor: The quantity that you want the model to monitor. By
default, it’s set to “val_loss”.
•
verbose: Sets verbosity to 0 or 1. It’s set to 0 by default.
•
save_best_only: If set to true, then the model with the best
performance according to the quantity monitored will be saved.
•
Do'stlaringiz bilan baham: |