This TI was trained using a new method called Dynamic LR adaptation where there is no need to specify a LR. The result is pretty good for a 1st test.