TensorFlow was initially designed as a library to implement deep nets. However not TensorFlow has evolved to become an eco-system that supports a model throught its all stages of life.
For example, with TensorFlow probability you can implement probabilistic models like Bayesian neural networks.
Typically when using TensorFlow in industry or to do most research you won't need to combine frequentist deep nets and bayesian deep nets. Bayesian deep nets are mostly used in specific research. For a newbie, it would be much safer to start with typical deep nets (not Bayesien ones).
Hope this helps.
PhD | Senior Data Scientist | AI/ML Educator
We find this kind of rampant individuality very disturbing. But not this tiny ad: