Skip to content

Commit

Permalink
Update README.md
Browse files Browse the repository at this point in the history
  • Loading branch information
ARenanse authored May 7, 2020
1 parent 2b64dfe commit 94e6e96
Showing 1 changed file with 3 additions and 1 deletion.
4 changes: 3 additions & 1 deletion README.md
Original file line number Diff line number Diff line change
@@ -1,5 +1,7 @@
[![Build Status](https://travis-ci.org/WhenDustSettles/LGPT.svg?branch=master)](https://travis-ci.org/WhenDustSettles/LGPT)

# Langevin Gradient Parallel Tempering for Bayesian Neural Learning
This repository is a very tangible and general purpose implementation of the Langevin Gradient Parallel Tempering Algorithm as explained in Chandra et al in < arXiv:1811.04343v1 >.
This repository is a very tangible and general purpose implementation of the Langevin Gradient Parallel Tempering Algorithm as explained in Chandra et al in < arXiv:1811.04343v1 > using mutliprocessing.

Using this, one can sample from the Posterior Distribution in weights of any\* PyTorch based model by inheriting the *PTReplicaMetaBase* class from *ReplicaABC.py* and defining the other model parameters, like the Prior Distribution over weights, the Likelihood and other problem dependent parameters (look over to the following sections on how to declare them correctly).

Expand Down

0 comments on commit 94e6e96

Please sign in to comment.