Language model identifies PDE terms more accurately from summaries
A 44-sample preprint benchmark found higher PDE-term recovery scores from physical summaries than raw slices, using noise-free simulations.
Developing Light ยท https://developinglight.com/editorial/developing-light