mackelab/snl

Python 3 port of SNL

Sequential Neural Likelihood

This is a convenience package wrapper of the code written by George Papamakarios found here

for reproducing the experiments in the paper:

G. Papamakarios, D. C. Sterratt, I. Murray. Sequential Neural Likelihood: Fast Likelihood-free Inference with Autoregressive Flows. arXiv:1805.07226. 2018. [pdf] [bibtex]

Content

This repo contains a python 3 port of the original python 2 code obtained from a simple application of 2to3 (with a few subsequent manual fixes, see commits). A setup.py allows installing and calling the functionality (e.g. fitting MAFs to data with SNL) from outside. Scripts and modules for reproducing the original experiments of Papamakarios et al. (2018) were identified and removed.