`PolyOpt` only accept functions without any extra inputs

[prbzrg]

As docs suggest for mini-batch, the loss function can be (u, p, xs_) -> vl and it works with OptimizationOptimisers but this line in OptimizationPolyalgorithms creates errors. Does using PolyOpt have an intended restriction on the loss function?

Optimization.jl/lib/OptimizationPolyalgorithms/src/OptimizationPolyalgorithms.jl

Line 14 in 8310517

loss, θ = x -> prob.f(x, prob.p), prob.u0

[ChrisRackauckas]

Oh that just needs to splat the args

[Vaibhavdixit02]

We have moved in the other direction, no solver accepts arbitrary arguments anymore so this won't be fixed