To who it may concern,
Here is a simple example showing how to change the architecture of the neural density estimators in MNLE. I found it very helpful. Specically, when creating the MNLE object, you can pass a pre-configured density estimator.
You can create this density estimator using the `likelihood_nn’ method from the toolbox and passing the desired hyperparameters there. Jan created a code example here: https://gist.github.com/janfb/8e6c52502b2470bc5ba1c50d8a51972b
Hope this helps!
Jiashun