The most exciting news lately is Python 3.14's new interpreter, which uses technology from this article's author. The article reviews the application of tail call optimization technology in multiple projects over the four years since it was proposed, including Python 3.14, LuaJIT Remake, GCC and Clang support, and the ongoing push to include it in C standard proposals.
TL;DR
uv turns one with a frank look at pros, cons, and whether to migrate, PyPI adds iOS and Android wheels, a tail-calling interpreter stirs the performance pot, and LLaMA3 fits in 100 lines of Jax.
Articles & Tutorials
× 12uv has been out for a year now, and its popularity momentum is as strong as Ruff! This long article thoroughly analyzes uv's pros and cons, when you should use it, and when you shouldn't. Overall, the advantages far outweigh the disadvantages, and the conclusion is clear - it's time to migrate to uv!
From MongoDB's official blog, this introduces their journey of developing a NoSQL database backend for the Django framework, including discovering needs from survey data, developing the database backend based on open source projects, and implementing support for MongoDB features.
Great news! Through PEP-730 and PEP-738, Python now supports distributing binary packages (wheels) for iOS and Android platforms. This PyPI move will lower the barrier for Python mobile development.
The tenth article in the "Building a Python compiler and interpreter" series, implementing support for conditional statements elif and else, showing how to add complex control flow statements in a compiler.
Introduces how to develop a complete deck of playing cards using Python and SVG images, supporting dynamic card generation, sorting, displaying card faces, shuffling and dealing, demonstrating the application of data processing, file operations, and visualization.
How do you implement a simplified version of the LLaMA3 model? The article utilizes JAX's pure functions and features like JIT and vmap to implement all core components of the LLaMA3 model, suitable for learning Transformer architecture and JAX usage.
When Python's any() function takes generator expressions versus lists as parameters, which performs better? After testing, the author found that when the first True value appears early in the sequence, generator expressions perform better; when the entire sequence needs to be checked, lists might perform better.
Explores how to implement asynchronous push notifications through the Django framework using Web Push Notifications and Progressive Web Apps (PWA) technology without relying on mobile applications.
Introduces how to customize code decoding methods through Python's # coding: directive and codecs module, thereby changing Python's syntax. The article implements a decoder that can convert brace syntax to Python's indentation syntax.
The article compares the implementation of inheritance and subtyping in Java, Go, and Python, exploring how language design affects inheritance mechanisms: Java has explicit inheritance and subtyping, Go implements implicit inheritance through duck typing, while Python combines explicit inheritance with static duck typing through protocols.
The author participated in the Genuary activity, generating a visual art piece every day in January. In this challenge, he deeply explored the application of PyScript and WebGL in creative programming, discussing the performance and limitations of Pyodide and MicroPython in browsers. The article also shares some stunning 3D creative works.
Projects & Resources
× 12It allows you to perform jQuery queries on XML documents, using lxml for fast XML and HTML manipulation. (2.3K stars)
Automatically extracts subtitles from YouTube playlists, uses Gemini for text refinement, and outputs to formatted Markdown files.
Uses generative AI to transform multimodal content into engaging multilingual voice conversations, serving as an alternative to Google's NotebookLM. (3.2K stars)
A containerized LLM toolkit that allows you to run LLMs and other services. Includes a CLI and companion app for easy management and running of AI services. (1.3K stars)
A tool for analyzing NGINX configurations that prevents security misconfigurations and automates defect detection.
A dashboard project developed with Python + Vue.js that seamlessly integrates with various data sources, including databases, APIs, and business systems, enabling effective data analysis.
Open sourced by Microsoft, it can parse user interface screenshots into structured and easily understandable elements, enhancing GPT-4V's ability to accurately locate and generate actions. (15.8K stars)
A modern Python logging tool supporting multiple output formats and colorful console output. Uses Tailwind CSS color palette, integrates MongoDB, supports file rotation and email notifications.
Saves chain of thought as text, then converts it to embeddings using OpenAI API, and finally plots the embeddings using t-SNE.
A comprehensive repository containing predefined Pydantic schemas for extracting structured data from unstructured visual content such as images, videos, and documents. Leverages Pydantic for automatic data validation and type safety.
Utilizes large language models to extract and classify structured data from documents, functioning like an ORM to enable seamless document processing workflows. (1.1K stars)
Through an intuitive graph view powered by Neo4j database, it visualizes relationships between infrastructure assets, reveals hidden dependencies, and reduces security risks. (3.2K stars)