The author runs Llama 3.3 70B model on their 64GB MacBook Pro M2. The article walks through the setup process and demonstrates impressive results for writing and coding tasks. (PS: Looking at my brand new MBP M4, I can't wait to try running local LLMs myself!)
TL;DR
A GPT-4-class model now runs on a laptop, PEP-768 proposes a safe external debugger interface for CPython, and the official MCP Python SDK is out.
Articles & Tutorials
× 12A new PEP proposing a zero-overhead debugging interface for CPython that allows debuggers/profilers to safely connect to running Python processes without stopping or restarting them. This feature is already implemented in PyPy and hopes to be merged into Python 3.14.
How do you handle various failures in Celery tasks and improve workflow resilience? The article provides solutions for three failure scenarios (transient failures, resource constraints, and race conditions), discussing how to handle process interruptions, memory overflows, task timeouts, and graceful shutdowns.
How to create a GitHub task list to check all functions in a file? The author wrote a script that parses all functions using regular expressions and creates a todo checklist.
A comprehensive guide on why and how to use uv. Covers installation methods, integrating it into existing workflows, managing virtual environments, package management, and uv-specific workflows and advanced usage.
10 years after PEP-484, what's the current state of Python's type system and what tools are developers using? JetBrains, Meta, and Microsoft collaborated on this comprehensive survey with over 1000 participants.
Python has recently become the most used language on GitHub, but what's the broader state of Python? Based on the latest community survey (the 2023 Python Developer Survey released in September), the article shares key trends such as 41% of developers having less than 2 years of experience, most web applications running on hyperscale clouds, and container technology surpassing virtual machines and hardware...
How to make Python scripts automatically create virtual environments and install dependencies on first run? The author shares how to use uv to simplify dependency management for Python scripts, avoid global package installations, and potentially replace pipx and pyenv for managing Python versions and CLI tools.
How to develop a simple chatbot using the Django framework and provide it with Wikipedia content for more meaningful Q&A interactions? The article covers Django project setup, loading encyclopedia data, generating and saving embeddings, and developing the chat interface.
The PyJokes library generates programming-related jokes and can add humor to your Python projects. But can other languages use it too? The article explains how to integrate PyJokes in Java, Ruby, and Node.js without requiring APIs or network access.
Python's match-case can't match iterators, but the author extends pattern matching behavior to support iterator matching, making pattern matching more precise and flexible for handling complex data structures and generators.
From December 1st to 24th, the author shared a favorite blog daily in 2021, wrote about technical writing daily in 2023, and this year's theme is software and design patterns.
Projects & Resources
× 12A keyboard-centric, vim-like browser built with Python and Qt. (9.9K stars)
A simple Python GUI framework that renders Python computation values through HTML in a webview.
A modern, stylish, and responsive admin dashboard built with DaisyUI and TailwindCSS. Features RTL support, multiple themes, and enhanced UX/UI.
An iCloud photo downloader supporting Linux, Windows, and macOS. Available as both command-line tool and executable, plus downloadable from package managers (Docker, PyPI, AUR, npm). (7.2K stars)
A Python-based static analysis tool that automatically identifies missing security patches in target systems by fetching the latest CVEs and their signatures from OSV. Currently supports C/C++ and Java source code.
An open-source RAG that integrates with ChatGPT and MCP, or runs completely locally. All neural networks (LLM, reranker, embedding) run on cloud or PC, ensuring data security.
An open-source database for vector search built with persistent storage, simplifying embedding retrieval, filtering, and management. Built with Rust + Lance, supports Python and JavaScript, featuring vector similarity search, full-text search, and SQL. (4.9K stars)
Validates DataFrame-like data objects, making data processing pipelines more readable and robust. (3.5K stars)
The official Python implementation of the Model Context Protocol (MCP), providing client and server functionality for LLM interactions.
A self-hosted PDF manager with Docker deployment support, offering seamless user experience across multiple devices. Features a clean, responsive UI with multiple themes, PDF sharing with password/view limits/expiration settings, and progress bars showing reading progress for each PDF.
An advanced cloud storage application built with Python + Vue, integrating secure authentication, encrypted transmission, and encrypted storage. Features file management, advanced encryption, end-to-end secure transmission, shared cloud storage, and multiple authentication methods.
An open-source AI search engine that not only searches the web but understands your questions, using advanced machine learning algorithms to optimize results. (17.4K stars)
Podcasts & Videos
× 2A dedicated website for Django events and videos: DjangoTV has released this year's DjangoCon US video series.
This video tutorial uses Replicate's AI service to fine-tune the open-source Flux image generation model, creating a Midjourney-like AI image generation tool using your own photos.