PythonHub
2.64K subscribers
2.35K photos
50.5K links
Human-curated Python news, projects, articles & tools.

https://pythonhub.dev/
Download Telegram
The last mile of a long road: faster NumPy in the browser

NumPy in the browser now uses OpenBLAS through Emscripten-forge, accelerating matrix multiplication by up to 30.92× for float32 and 14.90× for float64 in the authors’ 1024×1024 benchmarks. The article details the WebAssembly and Fortran toolchain work behind the improvement, plus upcoming SIMD optimizations that push performance further, reaching up to 64.89× with optional Relaxed SIMD i...

https://notebook.link/blog/the-last-mile-faster-numpy/
Yokan: a compiler from a statically typed subset of Python to native desktop apps

Develop on real CPython with hot reload, ship one native binary, and check that both runs behave the same.

https://github.com/i2y/yokan
🔥1
Soft-deprecating re.match()

The article explains when to use re.prefixmatch(), re.search(), and re.fullmatch(), what soft deprecation means for existing code, and how to flag re.match() usage with Ruff.

https://hugovk.dev/blog/2026/soft-deprecating-re.match/
FastHTML - Build Web Apps in Pure Python!

The video demonstrates building a polling application with FastHTML, covering form creation, POST request handling, and storing polls in a database. It also shows how to define database models using Python data classes, create a voting page, and process submitted votes.

https://www.youtube.com/watch?v=Ck0w7zqshjU
Fighting for #1 in the Ultimate Tic-Tac-Toe Arena

Tom Alard details how he built a highly competitive Ultimate Tic-Tac-Toe bot using a neural network trained on over 300 million self-play positions, a custom search algorithm, and SIMD-optimized C code. He also explains how he compressed the engine and neural network into a Python submission using UTF-16 encoding to bypass CodinGame’s 100,000-character limit, reaching second place on the...

https://tomalard.github.io/posts/fighting-for-1-in-the-ultimate-tic-tac-toe-arena/
Hardware-Agnostic Models in vLLM

The article explains how vLLM is introducing hardware-agnostic layers so it can keep supporting diverse models and accelerators even as frontier models increasingly rely on hardware-specific “flat” implementations. The new path remains compatible with torch.compile and, in tests on NVIDIA H100s, delivered total token throughput within 3.4% of the native implementation across three recent...

https://pytorch.org/blog/hardware-agnostic-models-in-vllm/
👌1
PyPy v8.0.0

PyPy 8.0.0 introduces its first Python 3.12 interpreter as a beta, alongside Python 2.7 and 3.11 releases, and raises the minimum glibc requirement for Linux binaries to 2.28. The release also advances compatibility with CPython’s limited C API and abi3 wheels, improves RPython code generation, and drops HPy as a default backend, though abi3 wheel installation support is not yet complete.

https://pypy.org/posts/2026/09/pypy-v800-release.html
🤯1