[D] Swift for TensorFlow is currently the best System for ML. CMV
After watching the talks and reading some of the docs it seems that Swfit for TensorFlow addresses most, if not all of the complaints people have about TensorFlow right now. The Swift tooling appears especially helpful with TF’s learning curve, which people often cite as one of the major downsides of TF.
What are your thoughts? Why do/don’t you use it and what would make you change your mind?