Why Swift for TensorFlow?

The core graph program extraction algorithm, automatic differentiation, and Python language interoperability features of Swift for TensorFlow can be implemented for other programming languages, and we are occasionally asked why we didn’t use some other one for this project. We defined goals around the properties that are important to maintain and improve in our system:

Expressiveness: We want a define-by-run model that feels like you’re directly programming against a numeric API and the host language (like NumPy), without an explicit graph abstraction in the way. Shallow learning curve: We want new users to spend their time learning TensorFlow, not struggling with a language that has a high learning curve.

Source: github.com