How can we align foundation models with populations of diverse users with different preferences? We are excited to share our work on Personalizing RLHF using Variational Preference Learning! 🧵 📜: arxiv.org/abs/2408.10075 🌎: weirdlabuw.github.io/vpl/