12/2017: Tensorflow distributions

Collection of notes and related information on the discussed paper: TensorFlow Distributions

  1. The bijector class of tfb is quite a powerful concept. For example, two recent blog post [1] [2] from Eric Jang demonstrated how to use it to construct Flows to get expressive distribution.

  2. The implementation of HMC in tensorflow, the leapfrog step is done inside of Tensorflow (as a comparison, PyMC3 also has a version to perform leapfrog integration inside the computation graph, however, the current implementation performs leapfrog outside of theano). See alos the related discussion on Github about HMC in tensorflow.

  3. There are a few teams also in progress to implement something similar in PyTorch. You can find their discussion of the design here: https://goo.gl/9ccYsq.

1 Like

Hello Mate,

Thanks for sharing info on TensorFlow Distributions, I have been learning tensorflow for mindmajix.

1 Like