Yeah, I was thinking about symmetries <=> conservation laws because of Noether's theorem. Think of regular NN training as not having any symmetries...since they

Yeah, I was thinking about symmetries <=> conservation laws because of Noether's theorem. Think of regular NN training as not having any symmetries...since they aren't baked into the model. But we can let the model learn symmetries <=> conservation laws by adding the Euler-Lagrange constraint to the forward pass of a NN.