Shayan Talaei@shayan-talaei · 1 worksInvestigates bias and preference steering in deployed language models.