Pre-trained models

Hi, first of all thanks for your great work!
I'm wondering if you can kindly share the pre-trained STEVE model that can (at least approximately) reproduce the results reported in the paper.
Thank you in advance for considering this request.