shumai
Nuitka
shumai | Nuitka | |
---|---|---|
15 | 94 | |
1,122 | 10,835 | |
0.2% | 2.0% | |
2.2 | 10.0 | |
10 months ago | 7 days ago | |
TypeScript | Python | |
MIT License | Apache License 2.0 |
Stars - the number of stars that a project has on GitHub. Growth - month over month growth in stars.
Activity is a relative number indicating how actively a project is being developed. Recent commits have higher weight than older ones.
For example, an activity of 9.0 indicates that a project is amongst the top 10% of the most actively developed projects that we are tracking.
shumai
-
PyTorch Primitives in WebGPU for the Browser
https://github.com/tensorflow/tfjs/tree/master/tfjs-backend-...
([...], tflite-support, tflite-micro)
From facebookresearch/shumai (a JS tensor library) https://github.com/facebookresearch/shumai/issues/122 :
> It doesn't make sense to support anything besides WebGPU at this point. WASM + SIMD is around 15-20x slower on my machine[1]. Although WebGL is more widely supported today, it doesn't have the compute features needed for efficient modern ML (transformers etc) and will likely be a deprecated backend for other frameworks when WebGPU comes online.
tensorflow rust has a struct.Tensor:
-
Why do people curse JS so much, but also say it's better than Python
JS for ML actually does exist https://github.com/facebookresearch/shumai
-
Breaking Up with Python
> It's really a shame that data science, ML, and notebooks are so wrapped up in it. Otherwise we could jettison the whole thing into space
Although I personally feel Python has its place, I contribute to a project that hopes to diversify the ML/scientific computing space with a TypeScript tensor lib called Shumai: https://github.com/facebookresearch/shumai
-
Tinygrad: A simple and powerful neural network framework
Doesn’t really matter for large batch/large model training on GPUs that don’t need much coordination.
But Python speed is one of the main motivations for a JS/TS based ML lib I’m working on: https://github.com/facebookresearch/shumai
-
[D] Using JavaScript for ML Training/Research (not in the browser)
As a hedge against CPython never becoming fast, we're creating a project called Shumai that attempts to deeply integrate with a new JavaScript runtime (Bun[3]).
-
Python 3.11 is much faster than 3.8
You can expose objects. Here's how it is done in Bun: https://github.com/facebookresearch/shumai/blob/main/shumai/...
We've been using this feature heavily in Shumai.
I think you are vastly overestimating the complexity associated with this (user exposed ref-counting/garbage collection) and may not be totally up to date on what's implemented.
- Shumai: Fast Differentiable Tensor Library in TypeScript with Bun and Flashlight
- Shumai: A fast differentiable tensor library for research in TypeScript and JavaScript
-
7% Speedup from Switch to and
This thought is pretty much the exact motivation behind a recent effort I’m helping out with https://github.com/facebookresearch/shumai
Nuitka
-
Py2wasm – A Python to WASM Compiler
Thanks for the feedback! I'm Syrus, main author of the work on py2wasm.
We already opened a PR into Nuitka to bring the relevant changes upstream: https://github.com/Nuitka/Nuitka/pull/2814
We envision py2wasm being a thin layer on top of Nuitka, as also commented in the article.
From what we gathered, we believe that there's usefulness on having py2wasm as a separate package, as py2wasm would also need to ship the precompiled Python distribution (3.11) for WASI (which will not be needed for the other Nuitka use cases), apart of also shipping other tools that are not directly relevant for Nuitka
-
Python Is Portable
This is a good place to mention https://nuitka.net/ which aims to compile python programs into standalone binaries.
-
We are under DDoS attack and we do nothing
For Python, you could make a proper deployment binary using Nuitka (in standalone mode – avoid onefile mode for this). I'm not pretending it's as easy as building a Go executable: you may have to do some manual hacking for more unusual unusual packages, and I don't think you can cross compile. I think a key element you're getting at is that Go executables have very few dependencies on OS packages, but with Python (once you've sorted the actual Python dependencies) you only need the packages used for manylinux [2], which is not too onerous.
[1] https://nuitka.net/
[2] https://peps.python.org/pep-0599/#the-manylinux2014-policy
-
Faster Blogging: A Developer's Dream Setup
glee is rich in blogging features but has some drawbacks. One of the main drawbacks is its compatibility with multiple operating systems and system architectures. We lost one potential customer due to glee incompatibility in macOS. Another major issue is the deployment time. We built the first version of glee entirely in Python and used nuitka, nuitka compiles Python programs into a single executable binary file. We need to create three separate stages for creating executable binaries for Windows, Mac, and Linux in deployment, and it takes around 20 minutes to complete.
-
Python 3.13 Gets a JIT
There is already an AOT compiler for Python: Nuitka[0]. But I don't think it's much faster.
And then there is mypyc[1] which uses mypy's static type annotations but is only slightly faster.
And various other compilers like Numba and Cython that work with specialized dialects of Python to achieve better results, but then it's not quite Python anymore.
[0] https://nuitka.net/
[1] https://github.com/python/mypy/tree/master/mypyc
-
Briefcase: Convert a Python project into a standalone native application
Nuitka deals pretty well with those in general: https://nuitka.net/
-
Ask HN: How does Nuitka (Python compiler) work?
Hi HN,
Has anyone explored Nuitka [1] and developed understanding from a blank slate?
Is there any toy version of this, so that one can start playing with the language translation concepts?
Is there any underlying theory/inspiration upon which this project is built?
Are there any similar projects, in say other languages?
[1] https://github.com/Nuitka/Nuitka
-
Why not tell people to “simply” use pyenv, poetry or anaconda
That's more of cultural problem in the Python community.
If I provide an end user software to my client written an Python (so not a backend, not a lib...), I will compile it with nuitka (https://github.com/Nuitka/Nuitka) and hide the stack trace (https://www.bitecode.dev/p/why-and-how-to-hide-the-python-st...) to provide a stand alone executable.
This means the users don't have to know it's made with Python or install anything, and it just works.
However, Python is not like Go or Rust, and providing such an installer requires more than work, so a huge part of the user base (which have a lot of non professional coders) don't have the skill, time or resources to do it.
And few people make the promotion of it.
I should write an article on that because really, nobody wants to setup python just to use a tool.
- Python cruising on back of c++
-
Is cython a safe option for obfuscate a python project?
As for a simpler option, you could use a "compiler": https://github.com/Nuitka/Nuitka
What are some alternatives?
rosettaboy - A gameboy emulator in several different languages
PyInstaller - Freeze (package) Python programs into stand-alone executables
jittor - Jittor is a high-performance deep learning framework based on JIT compiling and meta-operators.
pyarmor - A tool used to obfuscate python scripts, bind obfuscated scripts to fixed machine or expire obfuscated scripts.
openpilot - openpilot is an open source driver assistance system. openpilot performs the functions of Automated Lane Centering and Adaptive Cruise Control for 250+ supported car makes and models.
PyOxidizer - A modern Python application packaging and distribution tool
devdocs - API Documentation Browser
py2exe - modified py2exe to support unicode paths
FrameworkBenchmarks - Source for the TechEmpower Framework Benchmarks project
false-positive-malware-reporting - Trying to release your software sucks, mostly because of antivirus false positives. I don't have an answer, but I do have a list of links to help get your code whitelisted.
Pytorch - Tensors and Dynamic neural networks in Python with strong GPU acceleration
py2app