Details Model trained with train_size=0.8 or 80% of the Mozilla Common Voice Arabic training+validation dataset.
