meta-llama / codellama

Inference code for CodeLlama models

https://github.com/meta-llama/codellama

jxnl / instructor

structured outputs for llms

https://github.com/jxnl/instructor

BeyondLLM

Build, evaluate and observe LLM apps.

https://github.com/aiplanethub/beyondllm

Pydantic 2.7 released: the fastest version yet

It finally integrates jiter, a high speed JSON parser

https://github.com/pydantic/pydantic/releases/tag/v2.7.0

I accidentally built a meme search engine

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

https://harper.blog/2024/04/12/i-accidentally-built-a-meme-search-engine

CoreNet

A library for training deep neural networks by Apple.

https://github.com/apple/corenet

Learn to use Websockets with Django by building your own ChatGPT

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

https://www.saaspegasus.com/guides/django-websockets-chatgpt-channels-htmx/

Django from first principles

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 sta...

https://www.mostlypython.com/django-from-first-principles-2/

Asyncio Coroutine Object Methods in Python

We can define coroutine methods on custom Python objects. This allows methods on custom Python objects to use async/await syntax, such as awaiting other coroutines and tasks and allows the custom coroutine methods themselves to be awaited within our asyncio programs. In this tutorial, you will discover how to define object methods as coroutines.

https://superfastpython.com/asyncio-coroutine-methods/

Local LLM-as-judge evaluation with lm-buddy, Prometheus and llamafile

This post examines how different software components came together to allow LLM-as-judge evaluation without the need for expensive GPUs. All the components were built with and chosen for their user control, open source nature, and interoperability.

https://blog.mozilla.ai/local-llm-as-judge-evaluation-with-lm-buddy-prometheus-and-llamafile

cohere-toolkit

Toolkit is a collection of prebuilt components enabling users to quickly build and deploy RAG applications.

https://github.com/cohere-ai/cohere-toolkit

Patching requests for fun and (concurrent) profit

Because life is too short to spam calls to SSLCTXloadverifylocations().

https://blog.borrego.dev/entries/patching-requests-for-fun-and-concurrent-profit.html

WebLlama

Llama-3 agents that can browse the web by following instructions and talking to you.

https://github.com/McGill-NLP/webllama

TypeIs does what I thought TypeGuard would do in Python

The article discusses the differences between the TypeGuard and TypeIs type annotations in Python. It explains that TypeIs has more intuitive behavior compared to TypeGuard, especially when dealing with union types, and recommends using TypeIs over TypeGuard in most cases.

https://rednafi.com/python/typeguard_vs_typeis/

Doriandarko / maestro

A framework for Claude Opus to intelligently orchestrate subagents.

https://github.com/Doriandarko/maestro

Sleepsort: Sorting while sleeping

The post introduces the "sleepsort" algorithm, a quirky sorting method that utilizes thread sleep times to sort integers in ascending order. It provides a playful yet insightful exploration of the algorithm's functionality and limitations.

https://animeshchouhan.com/posts/sleepsort