Python Hub Weekly Digest for 2025-08-03
This week in Python, a tutorial on PyTorch was popular, offering a quick introduction to key topics like tensors, GPU training, and backpropagation. Toad, a universal terminal UI for AI coding agents, was announced, promising a flicker-free and interactive experience. Qwen3-Coder, the code version of Qwen3 large language model series, was developed by the Qwen team at Alibaba Cloud. Other highlights include a Python dict that reports unused keys, a tutorial on Python's type hints, and an article on Fully Homomorphic Encryption's potential for a truly private internet. Interesting projects include passage-of-time-mcp, UQLM, and Sniffly. Have a great week and happy coding!
đź’– Most Popular
PyTorch in One Hour: From Tensors to Training Neural Networks on Multiple GPUs
This tutorial offers a fast-paced introduction to PyTorch, covering key topics like tensors, GPU training, and backpropagation in about one hour. Its goal is to help you quickly start building and training deep neural networks, including large language models.
Announcing Toad - a universal UI for agentic coding in the terminal
Will McGugan announces Toad, a new universal terminal UI for AI coding agents built with Textual, offering a flicker-free and interactive experience compared to existing tools. Toad will be open source, supports any backend language, and uses JSON for communication between frontend and backend.
QwenLM / Qwen3-Coder
Qwen3-Coder is the code version of Qwen3, the large language model series developed by Qwen team, Alibaba Cloud.
A Python dict that can report which keys you did not use
Python Tutorial: Type Hints - From Basic Annotations to Advanced Generics
Corey Schafer explains Python’s type hints, from basic function annotations to advanced features like generics, with practical examples showing their benefits for code clarity, early bug detection, and IDE support. He highlights that type hints are optional and flexible, letting you gradually adopt them in your projects.
đź“– Articles
Fully Homomorphic Encryption and the Dawn of A Truly Private Internet
Fully Homomorphic Encryption (FHE) allows computation on encrypted data and is becoming ~8Ă— faster each year. It could enable a privacy-by-default internet, powering secure cloud and AI services without exposing user data.
Python and Make in 2025
The post recommends using GNU Make with Python projects for automation entrypoints, preferring uv for fast interpreter/venv setup and delegating complex logic to Python task runners, keeping the Makefile minimal and focused.
Superfunctions: A universal solution against sync/async fragmentation in Python
Tinyio: A tiny (~200 line) event loop for Python
Python Gotcha: Reusing Generators Returns Nothing
Reusing a Python generator after it’s been consumed yields no results, because generators maintain state and don’t reset once iterated. The fix is straightforward: call the generator function again to produce a fresh iterator before reusing it.
You might not need a Python class
Python AI Agents: Overview to Generative AI and Intro to Agents
Glenn Mossy, an AI Solutions Architect, presents an overview of generative AI and introduces Python AI agents, discussing how these technologies are automating tasks and transforming industries, with specific examples including an advanced calculator agent and a file system organizer. The session aims to be accessible for both beginners and experienced coders, highlighting the ongoing ev...
3 pandas Workflows That Slowed to a Crawl on Large Datasets—Until We Turned on GPUs
NVIDIA shows how switching from pandas to GPU-accelerated cuDF made slow data workflows run up to 30Ă— faster. Common tasks like time-series analysis and dashboard filtering became near-instant with minimal code changes.
Python Hub Weekly Digest for 2025-07-27
⚙️ Projects
passage-of-time-mcp
An MCP server that gives language models temporal awareness and time calculation abilities. Teaching AI the significance of the passage of time through collaborative tool development.
UQLM
UQLM is a Python library for Large Language Model (LLM) hallucination detection using state-of-the-art uncertainty quantification techniques.
Sniffly
Claude Code dashboard with usage stats, error analysis, and sharable feature.
AmazingHand
Code and model to control the AH!
higgs-audio
Text-audio foundation model from Boson AI.
đź‘ľ Reddits
UV is helping me slowly get rid of bad practices and improve company’s internal tooling.
Python 3.14: time for a release name?
Be careful on suspicious projects like this
Project by Ruslan Keba. Since 2012. Powered by Python. Made in 🇺🇦Ukraine.