Details Model trained with train_size=1.0 or 100% of the Mozilla Common Voice Arabic training+validation dataset.