LLM-to-Flow Distillation aims to steer scientific generators with natural-language goals

A post introducing LLM-to-Flow Distillation (LFD) presents it as a way to adapt scientific generators to goals such as designing peptides with promising anticancer activity. The author says it distills LLM knowledge into generators through iterative DPO on LLM-generated sub-goals, works with any frontier LLM via API access, and outperforms FlowGRPO and DiffusionNFT in held-out computational evaluations.