Use Nmbrs (SE) with Forefront

Deploy standard and fine-tuned GPT-J instances with the best cost-efficiency and throughput available.