Using Pytorch neural network as the custom “black box” likelihood function

Questions v5
4

It works. Thank you.
BTW, why the Pytorch model is compatible here but not in other posts, e.g., How to combine pymc with neural network - #2 by jessegrabowski

The worked code is as belows,

        processed_params = [param.squeeze() if param.shape == (1,) else param 
for param in inputs]
        awid, bdep, cf, cr, eta, data = processed_params