Hi! I am trying to model a dataset where each object has a variable number of features. In particular, I am trying to build on a finite approximated beta process-bernoulli process type of model, where each object has a variable-length list of values drawn from a normal distribution, and where the …

If you update to the current master branch, it will give you a more informative error message showing you which RV is hitting the zero grad error. Looking at the code, it is likely that the RV errs is too little uncertainty - as @rlouf suggest, you can try setting it to a dummy variable: errs = np.…

Understanding error while sampling a model with conditional probabilities

Questions

ahmadsalim May 28, 2018, 8:25am 6

Great, thanks!

I was trying to experiment a bit myself, and it does seem that NUTS is more sensitive to low standard deviation than other MCMC methods.

Topic		Replies	Views
Sampling error when using complex tensor function as distribution parameter Questions	3	475	September 11, 2018
ValueError: Mass matrix contains zeros on the diagonal Questions	1	2225	March 16, 2020
Model with conditional parameters causing mass matrix to contain zeros on diagonal Questions	4	870	January 7, 2020
Reinfocement learning model - derivative of RV is zero Questions	5	1787	April 10, 2019
"ValueError: Mass matrix contains zeros on the diagonal" error, how to debug? Questions	7	1628	March 7, 2020

Understanding error while sampling a model with conditional probabilities

Related topics