Hi everyone!

I was trying to test the GPU ability of NeuralPDE.jl on the JuliaHub, my network is defined like this:

```
# Neural network
inner = 25
chain = Chain(Dense(3,inner,Lux.σ),
Dense(inner,inner,Lux.σ),
Dense(inner,inner,Lux.σ),
Dense(inner,inner,Lux.σ),
Dense(inner,6))
strategy = GridTraining(0.05)
ps = Lux.setup(Random.default_rng(), chain)[1]
ps = ps |> ComponentArray |> gpu .|> Float64
discretization = PhysicsInformedNN(chain,
strategy,
init_params = ps)
@named pdesystem = PDESystem(eqs, ic_bc, domains, [t,x,z], [u(t,x,z), w(t,x,z), P(t,x,z),
T(t,x,z), τ₁(t,x,z), τ₃(t,x,z) ])
prob = discretize(pdesystem,discretization)
```

But when I execute the last line I get:

```
ERROR: CuArray only supports element types that are allocated inline.
Real is not allocated inline
```

I was looking to see if this issue was mentioned somewhere and I found this

But I didn’t understand if this was solved yet.

This is my first time trying to use a GPU, so any hints would be highly appreciated!