Fig. 1From: Solving multiphysics-based inverse problems with learned surrogates and constraintsPermeability models. First row shows the realistic permeability samples for FNO and NF training. Second row shows the generative samples from the trained NFBack to article page