[Query] TensorFlow Probability - Why not part of core TensorFlow?

Special Interest Groups
, ,
#1

As the title implies, is there any particular reason for not having tensorflow-probability (tfp) in the core tensorflow (tf)?

The reason I’m asking is because

  1. I’ve faced several times compatibility issue from both. For example, currently the latest version of tfp supports tf 2.9+ (unlike the latest tensorflow-addons (tfa), which I can use even with tf 2.4).

  2. Emerging library like KerasCV (and probabely KerasNLP too) doesn’t want tfp dependencies. That means, all the probability distributions and sampling functions needs to be written. For example, function like tfp.distributions.Dirichlet, tfp.distributions.MarkovChain, etc.

  3. I don’t know if it has been discussed before and the probability (tfp) package started maintaining its own repo (from tf.compat.v1.distributions!). It’s a bit confusing and sometimes documentation is hard to follow in tfp. In contrast, torch serves both in single API torch.distributions while following the design pattern of TensorFlow Distributions.

1 Like