Intuitive, easy CLIs based on python type hints.
Set of tools to assess and improve LLM security.
Mamba-Chat is the first chat language model based on a state-space model architecture, not a transformer.
pytest daemon: 10X Local Test Iteration Speed
At Discord, their Python test suite suffered from slow local test runs, taking 13 seconds per test. They built a pytest daemon that improves local test iteration speed by 10x, significantly reducing development time. This solution involves offloading heavy work to a background process and caching results, bypassing slow imports and fixtures.
Django: Sanitize incoming HTML fragments with nh3
Let’s look at how to use nh3 for HTML sanitisation in Django forms. You can adapt this approach to other situations, such as in DRF serializers.
NVIDIA / Megatron-LM
Ongoing research training transformer models at scale
imoneoi / openchat
OpenChat: Advancing Open-source Language Models with Imperfect Data
netease-youdao / EmotiVoice
EmotiVoice 😊: a Multi-Voice and Prompt-Controlled TTS Engine
Python in Visual Studio Code – December 2023 Release
This release includes the following announcements:
Annotated At Runtime
PEP 593 is a bit vague on how you’re supposed to actually consume arguments to Annotated; here is my proposal.
LouisShark / chatgpt_system_prompt
collect agent's system prompt and share some prompt inject knowledge
llmware-ai / llmware
Providing enterprise-grade LLM-based development framework, tools, and fine-tuned models.
Arrest is a small utility to easily structure and validate your REST api calls using pydantic and httpx.
µHTTP emerged from the need of a simple web framework. It's great for micro-services, single page applications, AND monolithic monsters.
Build and iterate on your UI 10x faster with AI - right from your own IDE.
An LLM Compiler for Parallel Function Calling.
A Production-ready Reinforcement Learning AI Agent Library brought by the Applied Reinforcement Learning team at Meta.
Netchecks is a set of tools for testing network conditions and asserting that they are as expected.
A library for generative social simulation.
Single source of truth with requirements for pip and conda.
Project by Ruslan Keba. Since 2012. Powered by Python. Made in 🇺🇦Ukraine.