A Transformer Model Architecture Uses Self Attention