That would be very simple to implement and a reason why I didn’t think of it is probably because llama runs so slowly on my machine!

Thank you for the feedback and the suggestion.