early_stop() is a helper function to be supplied on early_stopping arguments.
Usage
early_stop(
patience = 5L,
min_delta = 1e-04,
restore_best_weights = TRUE,
monitor = "val_loss"
)Arguments
- patience
Integer. Epochs to wait after last improvement. Default
5.- min_delta
Numeric. Minimum improvement to qualify as better. Default
1e-4.- restore_best_weights
Logical. Restore weights from best epoch. Default
TRUE.- monitor
Character. Metric to monitor. One of
"val_loss"(default) or"train_loss".