How can one steer generative flow models (e.g., for geometric molecular design) not via explicit hand-designed reward functions but implicitly via high-level discovery goals expressed in natural language? LFD provides an approach to do so along with scientific case studies.
We’re bringing natural-language control to any scientific generator.
We are very excited to introduce LLM-to-Flow Distillation (LFD), a method to adapt scientific generators to high-level natural-language discovery goals.
Starting from textual goals like “design peptides with
