This paper was co-authored by Hon Professor Haitham Bou Ammar, Sanome’s Lead Reinforcement Learning Advisor.
nb. This paper is in no way affiliated with/to Sanome.
“Our results show end to end training for BO yielding new sota results on HPO, antibody design, EDA and MIP tuning. We use a small model and less data than optformer but arrive at identical regrets while needing way less compute and memory!” – Prof Bou Ammar