- Python Weekly
- Posts
- Python Weekly (Issue 693 April 3 2025)
Python Weekly (Issue 693 April 3 2025)
Rahul Chaudhary
April 03, 2025
Welcome to issue 693 of Python Weekly. Let's get straight to the links this week.
You’ve heard the hype. It’s time for results.
After two years of siloed experiments, proofs of concept that fail to scale, and disappointing ROI, most enterprises are stuck. AI isn't transforming their organizations — it’s adding complexity, friction, and frustration.
But Writer customers are seeing positive impact across their companies. Our end-to-end approach is delivering adoption and ROI at scale. Now, we’re applying that same platform and technology to build agentic AI that actually works for every enterprise.
This isn’t just another hype train that overpromises and underdelivers. It’s the AI you’ve been waiting for — and it’s going to change the way enterprises operate. Be among the first to see end-to-end agentic AI in action. Join us for a live product release on April 10 at 2pm ET (11am PT).
Can't make it live? No worries — register anyway and we'll send you the recording!
Articles, Tutorials and Talks
This video provides a comprehensive, step-by-step coding guide to understanding and implementing DeepSeek V3, a cutting-edge deep learning model. It covers key concepts like the attention mechanism, multihead latent attention (MLA), rotary positional embeddings (RoPE), and the mixture of experts (MoE) architecture, explaining the science behind it all.
Researchers at LandH won the Most Valuable Hacker award by finding a vulnerability in Gemini's Python sandbox, which allowed them to extract internal Google source code and access Google tools via crafted RPC messages. They discovered that the sandbox communicates with external Google servers and could be manipulated by exploiting file descriptors and protobuf messages, potentially bypassing security measures.
Sharing single-file Python scripts with external dependencies is now easy thanks to uv and PEP 723, which enable embedding dependency metadata directly within scripts. This approach eliminates the need for complex setup tools like requirements.txt or package managers, making script distribution and execution seamless and simplifying deployment while maintaining flexibility and efficiency.
This post examines multi-threaded parallel inference on PyTorch models using the new No-GIL, free-threaded version of Python. Using a simple 124M parameter GPT2 model that we train from scratch, we explore the novel new territory unlocked by free-threaded Python: parallel PyTorch model inference, where multiple threads, unimpeded by the Python GIL, attempt to generate text from a transformer-based model in parallel.
Have you ever wanted to try an Async Python migration, but were unsure where to start? Read this post to find out!
This video explains principles for writing less brittle Python code by avoiding type constraints, using value constraints inside functions, and carefully considering optional values. It also advocates for raising errors instead of returning None when a function expects to find an object by ID but cannot.
This video provides a tutorial on deploying a Django project to Railway using modern deployment practices and covers setting up Python environments, Git, and configuring Django for production. It demonstrates how to use Railway's platform services to host and deploy Django applications with proper configurations, using Nixpacks or custom Dockerfiles.
This video demonstrates using a robotic hand to perform gestures like rock, paper, and scissors, programmed with the help of language models, highlighting the model's ability to understand and execute these gestures even if they aren't explicitly defined in the hand's code. The creator explores how language models can abstract and translate general knowledge into specific robotic actions, while also facing challenges with code execution and module usage.
This article discusses a common Python pitfall where shadowing a variable within a function can lead to an UnboundLocalError due to Python's scoping rules. It explains that if a variable is bound anywhere in a function, it's considered local to the entire function, even before it's initialized, which can cause unexpected errors.
This article introduces testing in Flask, emphasizing its importance for code quality and streamlined development. It covers key testing areas like routes, business logic, database interactions, and provides practical examples using pytest, fixtures, mocking, and a focus on edge cases.
Build your first Streamlit app and explore some basic features.
This post explains how to train and fine-tune reranker models using Sentence Transformers v4, detailing components like datasets, loss functions, and evaluators. It demonstrates that fine tuning models on specific domains outperforms general-purpose models, providing practical examples for improving relevance in information retrieval.
Interesting Projects, Tools, and Libraries
The fast, Pythonic way to build Model Context Protocol servers.
Your all-in-one for beautiful, lightweight, prod-ready CLIs.
A Model Context Protocol server for searching and analyzing arXiv papers.
Tenacious tool calling built on LangGraph.
Agentic AI Development in Emacs.
A lightweight library for building Multimodal Agents. Use it to give LLMs superpowers like memory, knowledge, tools and reasoning.
This is a Model Context Protocol (MCP) server for WhatsApp.
Amazon Nova Act is a research preview of a new AI model for developers to build agents that take actions in web browsers.
Fast and Accurate Video Understanding Language Model.
Deferred computational framework for multi-engine pipelines.
A feature-rich Python text case conversion library.
Python applications to Docker, automatically.
An open file format for serializing stateful agents with persistent memory and behavior. Share, checkpoint, and version control agents across compatible frameworks.
AI Runner is a local-first tool that allows you to run open-source large language models (LLM) and AI image generators (Stable Diffusion) on your own hardware, without the need for a web server or cloud service.
New Releases
Django 5.2 is released with new features like automatic model imports, composite primary key support, and easier template overriding. With this release, Django 5.1 reaches the end of mainstream support, and Django 5.0 reaches the end of extended support, urging users to upgrade to newer versions.
Upcoming Events and Webinars
There will be following talks
PostgreSQL pgvector for Pythonistas
The hashy/cachy library
Remember to .close() file-like objects
There will be a talk, Python internals: tips to understand better the current and future state of Python.
There will be a talk, PySpark 101: Introduction to Big Data with Spark.
There will be a talk, Building RAG Solutions Using Ollama and LangChain.
There will be following talks
Real-time collaborative editors in JupyterLab
Handling imbalance data for client scoring
Choice modeling with the Python package choice-learn
There will be following talks
UV Got to Be Kidding: Making Dependency Management Fun Again
After the Hype, there’s Production: using AI Workflows Agents, and Agentic Workflows
There will be following talks
Anonymization: Why is it so hard?
LLMs, the do-it-yourself edition
There will be following talks
Automating open source LLM fine-tuning, monitoring and cloud deployment
The Power of Self-Learning (or, How I Built a Website With Zero Knowledge)
Our Other Newsletters |
Programmer Weekly - A free weekly newsletter for programmers.
Founder Weekly - A free weekly newsletter for entrepreneurs featuring best curated content, must read articles, how to guides, tips and tricks, resources, events and more.