Function must be called in mode continuous time

[awecefil]

Hi, currently I encounter an error when calling a ME_NeuralFMU` in loss function

function lossSum(p)
    global neuralFMU, x₀, params, counter

    params = Dict(zip(Vector{String}(modified_Variables), Vector{Float64}(p.value)))
    solution = neuralFMU(x₀; parameters=params, p=p, showProgress=true, saveat=tSave, recordValues=vrs)  #** this asserts the error message**

    if !solution.success
        return Inf 
    end
    
    # id_Net, iq_Net = extractData(solution)
    id_Net = fmi2GetSolutionValue(solution, 1; isIndex=true)
    iq_Net = fmi2GetSolutionValue(solution, 2; isIndex=true)
    
    loss_value = FMIFlux.Losses.mse(id_Net, id_Data) + FMIFlux.Losses.mse(iq_Net, iq_Data)
    
    return loss_value
end

The error message is:

ERROR: AssertionError: condition!(...): 5 != 3
Function must be called in mode continuous time!
This is most probably because the FMU errored before. If no melected during export) and follow the message printing instru

After tracing the code, I found that this error caused by the assertion in https://github.com/ThummeTo/FMIFlux.jl/blob/main/src/neural.jl#L378, where c.state=5 but fmi2ComponentStateContinuousTimeMode=3

thus I trace the code to find when the c.state is been changed from 3 to 5, and I found that it occurs at https://github.com/ThummeTo/FMIImport.jl/blob/main/src/FMI2/c.jl#L1937 because the status returned by fmi2CompletedIntegratorStep!(c.fmu.cCompletedIntegratorStep, c.addr, noSetFMUStatePriorToCurrentPoint, enterEventMode, terminateSimulation,) is 3 and thus trigger the c.state = fmi2ComponentStateError in https://github.com/ThummeTo/FMIImport.jl/blob/main/src/FMI2/c.jl#L132 to assign 5 to c.state

I am not sure why the status returned by fmi2CompletedIntegratorStep! is 3, is there any suggestion to solve this issue? If more information is needed, please tell me, thanks!