Authors: Nat Mathews (NASA GSFC), Barbara Thompson (NASA GSFC)

We present a Physics-Informed Neural Net architecture for full-vector-field construction of magnetohydrostatic models of coronal active regions. We first leverage a state-of-the-art extrapolation method to develop a public dataset of over five thousand data cubes based on the SHARP library, which are presented for public availability. We then provide an example use for this data by training a graph-network-based PINN to reproduce coronal magnetic fields.