Robin Yadav

🎯We further improve diversity/feasibility of generated reactants with Feynman-Kac (FK) steering, an inference-time steering method that guides sampling of flow models. Using a forward-synthesis reward model, we achieve a 19% gain in top-5 round-trip accuracy over prior methods

🔬We propose the first flow matching framework for retrosynthesis: RetroSynFlow (RSF), which leverages intermediate molecular fragments called synthons. RSF achieves 60% top-1 accuracy on the USPTO-50k benchmark, outperforming state-of-the-art template-free methods.

Excited to announce RetroSynFlow, a discrete flow matching framework for diverse and accurate retrosynthesis! 📄Paper available here: arxiv.org/abs/2506.04439 Joint work w/ @qi_yan98 , Guy Wolf, @bose_joey and @lrjconan . 👇 A quick overview 🧵