Could not pickle model, samlping singlethreaded

Questions v5
1

Hi, my model runs succesfully ( a very simple ode model ) and my posterior is reasonable.

However, when I specify more cores in the sample method, I always get the error:

Could not pickle model, sampling singlethreaded.
Sequential sampling (4 chains in 1 job)

The sampling still completes, and for the case of this simple problem it takes 37 seconds for 1000 tunes and 1000 draws over 4 chains.

However, I would like to know if there is anything I can do to make sure multiple cores are used instead of sequential sampling. I could see from other forms that windows might be the problem but still would like to confirm.

I have tried using the if __name__ == '__main__': method, but it has no impact on my output.

The sampling code for those who want to know:

def run_model_slice(vars_list, model, tune, draws, chains=4, cores=4):
    with model:
        trace = pm.sample(step=[pm.Slice(vars_list)], tune=tune, draws=draws, chains=chains, cores=cores)
    return trace

trace_slice = run_model_slice(vars_list=vars_list, model=model, tune=tune, draws=draws, chains=4, cores=4)

Information on my setup:
Windows 11 version 23H2
VSCode with Jupyter Notebook
Anaconda environment with PyMC version 5.13.1

2

How is the model itself. Are you defining new Operators for it? If so you may need to move them to a separate file and import them

3

Only operators present are @as_op and @njit for ode specific functions, which are both present in separate files and imported (Same way that the Lotka-Volterra example does it, but in a separte file imported)