Joint Prompt Optimization of Stacked LLMs using Variational Inference | Read Paper on Bytez