Anti Python AI Club: AI for Python H8ers(github.com)
github.com
Anti Python AI Club: AI for Python H8ers
https://github.com/Fileforma/AntiPython-AI-Club
6 comments
I agree with your attitudes towards Python completely -- I'm a competent coder with a decade of experience, but find Python inscrutible for programs longer than 100 lines.
I would ask you to consider first: are there other efforts that you can join before deciding to strike out on your own? You are far from the first person to dislike Python, and there is very interesting work happening in the Elixir community around creating a viable alternative to Python for AI and Machine Learning. They are very capable folks, and I think spending some time looking at their work, what they're working on and why will do you well. There are also efforts in the Swift and Julia communities, but I'm personally less excited about those.
Check your LinkedIn messages.
I would ask you to consider first: are there other efforts that you can join before deciding to strike out on your own? You are far from the first person to dislike Python, and there is very interesting work happening in the Elixir community around creating a viable alternative to Python for AI and Machine Learning. They are very capable folks, and I think spending some time looking at their work, what they're working on and why will do you well. There are also efforts in the Swift and Julia communities, but I'm personally less excited about those.
Check your LinkedIn messages.
As someone who has been far down this road and built libraries in other languages, I would espouse that this is mostly not worth time.
The community for ML is built around Python and C++, and you will be fighting upstream the whole way instead of just producing things that are useful to people.
Cuda is written in C++ so most of everything builds up from that. Python has great C++ interop, not many languages do. The scale of tooling that has been built in Python is not easily replicated. I've watched libraries come and go over the years as they finally realize they can't keep up and it takes a big community to tackle these problems.
You can learn a lot by writing ML in other languages, and I get it, Python leaves a lot to be desired, but if you want to build things that are truly useful, I would stick with the community.
The community for ML is built around Python and C++, and you will be fighting upstream the whole way instead of just producing things that are useful to people.
Cuda is written in C++ so most of everything builds up from that. Python has great C++ interop, not many languages do. The scale of tooling that has been built in Python is not easily replicated. I've watched libraries come and go over the years as they finally realize they can't keep up and it takes a big community to tackle these problems.
You can learn a lot by writing ML in other languages, and I get it, Python leaves a lot to be desired, but if you want to build things that are truly useful, I would stick with the community.
I love the idea/concept. I think it could gain more traction if you didn’t use the term hate and instead went for something more like “an alternative to python” or something like that.
What on earth?
You don’t have to actively shit on people building things in a tool that works for them just because you don’t like it. Do whatever you like in whatever language you like, but don’t define yourself by being anti-something else, define yourself by building things people find useful.
You don’t have to actively shit on people building things in a tool that works for them just because you don’t like it. Do whatever you like in whatever language you like, but don’t define yourself by being anti-something else, define yourself by building things people find useful.
Your schedule is a bit on the ambitious side, especially when comparing to Python: that relies on libraries (e.g. tensorlfow, written in C++). Why don't you interface with those? Many have interfaces/bindings to other languages.
Although it is true that popular frameworks/libraries/ like Tensorflow, Keras, Pytorch..are written in C/C++, I find it very interesting what you are proposing.
Python is nice but I think it would be a good exercise to use only a compiled language like C/C++/Rust/Nim/other language...
We work together to build AI models in our favorite programming languages.
I feel that the only thing limiting my progress in AI is the Python programming language. I understand the math and the arxiv papers.
However, I experience something analogous to Dyslexia whenever I look at Python code. I wanted to build an open community for AI enthusiasts who want to work together to build models in our favorite programming languages.
I wanted to gauge interest before dedicating my time to this.