Python Weekly (Issue 647 April 25 2024)

Python Weekly - Issue 647

April 25, 2024

Python Weekly

Welcome to issue 647 of Python Weekly. Let's get straight to the links this week.

From Our Sponsor 

A weekly newsletter featuring the best hand curated news, articles, tutorials, talks, tools and libraries etc for programmers.

Articles, Tutorials and Talks

Learn how to implement RAG (Retrieval Augmented Generation) from scratch, straight from a LangChain software engineer. This Python course teaches you how to use RAG to combine your own custom data with the power of Large Language Models (LLMs).

py2wasm converts your Python programs to WebAssembly, running them at 3x faster speeds.

Kubernetes harbors capabilities that even seasoned developers might not be fully aware of. These hacks delve into the more esoteric, yet incredibly potent tricks that can significantly empower those who master them. These are not your everyday tips but profound insights into making Kubernetes do amazing things.

This article is primarily meant to act as a Python time complexity cheat sheet for those who already understand what time complexity is and how the time complexity of an operation might affect your code. 

Many people don't realize you can start a Django project with a single file. This series walks through the process of building a simple but non-trivial project by starting with a single file. The project only expands to additional files when it makes sense to move code out of the main file. By the end of the series, we'll have a project with a structure similar to what's generated by startproject and startapp.

Meta has unveiled Llama3, and now you can run it locally using Ollama. In this video, I explain how to use Ollama to operate various language models, specifically focusing on Llama2 and Llama3. I'll also guide you through the WebUI for the project, demonstrating how to serve models with Ollama and interact with them using Python.

In this post, we’ll cover Django’s branching structure, determining and searching through those commits, a worked example, and advanced behavioural searching with git bisect.

We'll build a voice notes app that uses OpenAI to perform speech to text. As a bonus, we'll use AlpineJS to manage state on the frontend.

By the end, you'll have built a multiplayer game using HTMX, using neat server-side logic and storing all results in your database. HTMX is a great way to use javascript without writing javascript.

When building RAG or agents, lots of LLM calls and non-LLM inputs feeds into the final output. The Langfuse decorator allows you to trace and evaluate holistically.

Everything you need to know about websockets to use them in your applications, with Django, channels, and HTMX.

Or: how to learn about clip/siglip and vector encoding images.

Get started building transformative AI-powered features within 5 minutes using Llama 3, Ollama, and Python.

Interesting Projects, Tools and Libraries

A library for training deep neural networks by Apple.

Cria is a library for programmatically running Large Language Models through Python. Cria is built so you need as little configuration as possible — even with more advanced features.

Automatic infrastructure for Django.

A JAX research toolkit for building, editing, and visualizing neural networks.

Build, evaluate and observe LLM apps.

A Python framework for defining and querying BI models in your data warehouse.

A Native-PyTorch Library for LLM Fine-tuning.

Command-line interface. Use this to chat with the model or train the model (training consumes the taxonomy data)

Open source ngrok alternative designed for teams

A collection of notebooks/recipes showcasing some fun and effective ways of using Claude.

New Releases

This release includes model weights and starting code for pre-trained and instruction tuned Llama 3 language models — including sizes of 8B to 70B parameters.

PyTorch 2.3 offers support for user-defined Triton kernels in torch.compile, allowing for users to migrate their own Triton kernels from eager without experiencing performance regressions or graph breaks. Tensor Parallelism improves the experience for training Large Language Models using native PyTorch functions, which has been validated on training runs for 100B parameter models. As well, semi-structured sparsity implements semi-structured sparsity as a Tensor subclass, with observed speedups of up to 1.6 over dense matrix multiplication.

Upcoming Events and Webinars

There will be following talks

  • How Generative AI will bring Schiphol's customer care to new heights

  • Unleashing Samurai: How Albert Heijn is Making Content Retrieval Hassle-free for Employees

There will be a talk, Accelerating Python Data Analysis with DuckDB.

There will be following talks

  • Introducing DBRX: A new SOTA open LLM by Databricks

  • Fidelius DBRXus: Build your own private Hogwarts

Our Other Newsletters

 - A free weekly newsletter for programmers.

- A free weekly newsletter for entrepreneurs featuring best curated content, must read articles, how to guides, tips and tricks, resources, events and more.