Thomas Tse

Results 1 comments of Thomas Tse

I noticed significant slow down using cuda (quadro p1000) too. Using model_size=medium.en, compute_type="in8" Downgrading to 1.0.2 resolved this issue.