[skohan]

Swift for Tensorflow also isn't at 1.0 AFAIK and hasn't been deemed "ready for primetime".

I think it's going to be hard to get data-scientists using anything other than Python any time soon, but I'm hopeful it will take off at some point, since it's really painful to work with scripting languages once you get used to a decent type system.

It's also a super interesting project to follow. The stuff they are doing with automatic differentiation, and the stuff they are talking about with metaprogramming is truly fascinating stuff, with all kinds of applications outside of machine learning. I really hope they manage to get it off the ground.

[6700417]

Yes it’s not at 1.0, but my point when I mention to people actually doing deep learning that Swift is coming and has support for native differentiation and static typing, the response is underwhelming to the point that I wonder if either of those features provide enough pull to get people to ever drop Python for Swift.

[skohan]

Yeah I think it will be a really hard sell since my perception is that programming is python for a lot of data scientists. Also static typing is something you have to use for a bit before you can understand that the little bit of extra formality is actually saving you a lot of trouble in the medium and long term.

Still given the growth of that industry, it's hard to imagine that Python can bear the burden of the increasingly complex work there forever.

[m_ke]

Static typing is useless for exploratory data analysis and data wrangling.

Once you get to the modelling step all of the data is in typed arrays.

Static types start to shine once the project grows and the new type annotation syntax works great for that.

I've tried doing machine learning in C++, Java, Scala and JavaScript and all of them are a pain compared to python.

Python would be perfect if I could annotate parts of my code and compile them. (Pretty much what's commonly done with cython or numba jit.

[skohan]

> Static typing is useless for exploratory data analysis and data wrangling.

In my experience, doing this kind of task in an weak-typed or duck-typed language is already tedious. Function signatures don't tell me enough detail to know exactly what I'm working with, so I end up spending a lot of time digging through documentation or print debugging. Why not let the compiler, or the function signatures do this work for you?

[pjmlp]

Julia is the only sound alternative to Python.

If they want to make an Apple only version of Tensorflow, so be it, but don't expect it to take off on Linux and Windows research labs.

[skohan]

What’s with the FUD? You can already run S4TF notebooks on Colab, and it’s trivial to spin up S4TF in any environment using Docker.

[pjmlp]

How do you spin S4TF on Windows Containers or any kind of Linux distribution in-house?

[skohan]

Using Docker

Edit: or on Ubuntu just download the toolchain.

[pjmlp]

So instead of using the native tooling provided by Julia, ML.NET, PyTorch, the solution is to somehow either emulate Linux on Windows, or being stuck with Ubuntu as distribution.