MagNET predicts NMR in seconds as an ML surrogate for DFT calculations

CatAstro_Piyush · x · 2026-07-23

A collaboration involving Merck and academic researchers launched MagNET, an ML surrogate for DFT-based NMR calculations.

The model predicts NMR with good accuracy in seconds, instead of running expensive quantum chemistry workflows.
The chart compares MagNET-Zero against older and newer baselines across geometry sources and basis sets, showing lower RMSE at much lower computation time.
The image frames the result as a practical speedup over conventional DFT NMR pipelines.

More from Research