pymc4_prototypes
pymc4_prototypes copied to clipboard
Reproduce simple example
https://github.com/pymc-devs/pymc3/blob/master/pymc3/examples/simpletest.py
I think that can be a milestone for any backend. It requires API concept as well as stochastic parameters properly working
Hey, can I take this up?
I can start by trying to run the the simpletest.py
on code with pytorch, tensorflow, CNTK or/and mxnet as the computational backend?
Go for it ;-) Maybe try with using tensorflow.distribution first.