Hi, as suggested by the topic, I am using nerual network model as the custom likelihood function. My data is from FEM simulations and a ANN neural network is trained based on the simulation data as a surrogate model. In the PYMC model, the surrogate model is wrapped in the custom likelihood function…

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

jessegrabowski June 13, 2025, 12:16pm 6

You wrapped your NN in an Op, so it seems totally consistent with what I suggested in the post you linked. Am I missing something?

Topic		Replies	Views
A 'black box' likelihood function causes the error: Wrong number of dimensions: expected 0, got 1 with shape (10,) v3	2	609	April 29, 2022
Using pm.DensityDist and customized likelihood with a black-box function Questions	10	6830	August 29, 2018
Back-box function without specifying gradient Questions	0	482	February 23, 2021
Suggested workaround for blackbox likelihood example? Questions	6	763	February 8, 2021
How to calibrate the sigma of the likelihood within the black-box method? Questions	13	1409	June 30, 2019