1 paper
Minghui Chen, Ruinan Jin, Wenlong Deng +4
Recent studies highlight the promise of LLM-based prompt optimization, especially with TextGrad, which automates differentiation'' via texts and backpropagates textual feedback. Th…