I know e.g. ALBERTA, a variant of BERT, has been recreated in Julia, and AlphaZero, but all the (recent) papers I read, I see are still based on Python code, if the language is known to me (e.g. GPT-2 and updated GPT-3). [SciML, merging neural networks and PDEs, maybe the the exception.] What (other…

I downloaded yesterday some scripts of how taxes are computed that the French Tax ministry released (it is so complicated in France that you can’t compute or check the tax by yourself…) It ended up it was a bunch of COBOL scripts (2020)…

I think the answer is very simple: Python is a very simple language, with a lot of consolidated libraries in the area (Scikit-learn, Numpy, PyTorch, Keras, OpenCV .API …). The majority of pre-processing algorithms are also available in Python (as sci. The Julia equivalents are, obviously, not so co…

[image] ToucheSir: Distributed and multi-gpu training: […] CliMA is the only public project I know of that has distributed GPU support, but it’s not DL FYI: I see multi-GPU done (while I guess not Distributed, then) in 2018, with Flux: [image] Multiple GPUs with Julia GPU …

[image] lmiq: There are countries that still do not use the metric system, you figure it out. Incidentally, one of those countries is also a birthplace of Julia…

Nice finds! Yup, I think this is possible on current CUDA.jl as well. However, multi-gpu training of the same model (whether on the same machine or across machines) requires functionality that isn’t implemented anywhere in the Flux ecosystem yet. Seems like all of these benchmarks are prior to t…

From your link, some research using Julia (and I also edited my top post, with state-of-the-art research from one of the main Knet/Julia guy): [image] Where does Julia provide the biggest benefits over other ML frameworks for research? Personal Examples: Although this paper I’m going to show yo…

Will Julia be as fast as optimized cpp + python in machine learning? Most machine learning core parts in cpp are written by experts so maybe they do not have the two-language problem?

If well written julia, then yes. In some cases it can be faster (If python-c interop creates bottlenecks). Also, even just using tensorflow or most other python deep learning libraries has a 2 language problem even if you’re just using python. python+tensorflow is 1000% uglier than pure python.

PyTorch has had more development and resources. Flux.jl is still quite early days. Any opinions on Knet vs Pytorch?

Are there any companies using Flux and Knet?

Why is Python, not Julia, still used for most state-of-the-art AI research?

Offtopic

Palli September 1, 2020, 10:05pm 28

From your link, some research using Julia (and I also edited my top post, with state-of-the-art research from one of the main Knet/Julia guy):

Topic		Replies	Views
Where does Julia provide the biggest benefits over other ML frameworks for research? Machine Learning	34	10783	September 16, 2019
Has Julia met your need for AI and ML? New to Julia	26	2120	October 7, 2024
Is Julia Falling Behind in Relevance? (because it's not used in LLM research?) Offtopic	69	8513	July 16, 2025
On Machine Learning and Programming Languages Machine Learning	48	9006	January 25, 2018
State of machine learning in Julia Machine Learning	60	66867	August 26, 2022

Why is Python, not Julia, still used for most state-of-the-art AI research?

Related topics