Gumbel-Softmax version of Bernoulli and Categorical distributions

Hamster_on_wheels · May 31, 2019, 1:24pm

Does pymc have gumbel-softmax reparametrization of Bernoulli and Categorical?
Does that reparametrization actually improve fitting of Bernoulli and Categorical?
Is it possible to use a random variable for the temperature?

Thanks again.

Topic		Replies	Views
Why doesn't pymc.Multinomial support logit_p parameterization? v5 modeling	3	43	May 15, 2025
Problem with pymc softmax transformation producing unexpected results v5 modeling	3	24	April 28, 2025
Implementing softmax regression with a latent Gaussian variable Questions	4	1849	May 26, 2019
Multi-Class Classifier in PyMC3 Questions	9	2699	February 22, 2023
Categorical latent parameter v5 modeling	4	109	March 29, 2024

Gumbel-Softmax version of Bernoulli and Categorical distributions

Related topics