About training strategy

I am trying to run the example of PINN soling PDE( GitHub - SciML/NeuralPDE.jl: Physics-Informed Neural Networks (PINN) and Deep BSDE Solvers of Differential Equations for Scientific Machine Learning (SciML) accelerated simulation )
But when I use the:
discretization = PhysicsInformedNN(chain, QuadratureTraining(),init_params =initθ)
in later part of:

it is having some error here.
But if I change the training strategy to GridTraining:
discretization = PhysicsInformedNN(chain, GridTraining(dx),init_params =initθ)
That error part is solved.
I am confused since the datatype of this : prob = discretize(pde_system,discretization)
is always OptimizationProblem for both training strategy why I cannot use the QuadratureTraining here

This is a failure on master due to type piracy in DistributionsAD.jl. See Neural adapter test is broken · Issue #412 · SciML/NeuralPDE.jl · GitHub. I want to get this fixed ASAP as well.

Thx