Will there be a PyTorch equivalent of Tensorflow's MCMC module?

Hi All,

I was wondering if PyTorch will ever make a Markov Chain Monte Carlo module like Tensorflow’s mcmc module?

I’ve made a basic version of a Metropolis-Hastings sampler myself, and want to have a looking at doing other (more complicated) samplers in the future. However, I was just curious about if this is something that will be implemented in a later version of PyTorch?

Thank you in advance!

I think Pyro has an implementation of Markov Chain Monte Carlo.

1 Like