Check gpu usage in activity montior window GPU Usage ^{⌘ + 4}

he uses windows, explains how the model fits entirely on gpu but is nearly maxing out GPU

To see what happend if model is too large, increase 'Context', that will exceed model limit,

Note move on,my comp 32gb ram, do this later,
I tried,
Increasing contenxt didn't seem to do much

remember to eject model