Current position of Golang in machine learning?

Question

go

Greetings! https

://www.reddit.com/r/golang/comments/anbjin/is...
the question on reddit was asked 2 years ago, how are things going with the use of Golang in machine learning at the moment?

Answer 1

You can run libs written in C from Go, but why?
People choose Go mainly because of the smart scheduler that makes networking convenient. ML doesn't really fit here, especially since most DS know Python, which means that it's much easier to find the right people.
Perhaps the only reasonable reason to use Go ML libraries is when you already have a Go service that you need to add some ML to.
If you are writing a new service where you plan to actively use ML libraries, then Python is perfect for a low load, but for a high load it is better to immediately take C ++ or some other language that is transparently compatible in memory with C.
Therefore, it seems to me that the ML-libraries on Go will continue to lag behind Python - they are simply not needed there by almost anyone, so they are not being developed.

Answer 2

Current position of Golang in machine learning?

Pretty good. Gradually acquires sound software. It looks especially advantageous when the project becomes complex and loaded. When the cost of heating the air in the DC becomes more tangible than the profit from this ML (Python is single-threaded and at least you fill up with cores, you won’t get the same from multithreading as in Go), when you finally die from the mess with Python versions, their compatibility and incompatibility with libraries, when it becomes a question of simple and easy scaling (what are the difficulties with organizing run time for a Go application?), when a quick hot-fix can be another lottery (dynamic typing can screw up in the most unexpected place), then you start thinking and breaking your AI .
If you look at the state of affairs at the moment, then all sorts of companies like Ozon, or Avito (probably heard of such) are very actively recruiting developers even with a capped acquaintance of the Go language (for details, see hh.ru), including in the ML and antifraud departments.