Pruna is a model optimization framework built for developers, enabling you to deliver faster, more efficient models with minimal overhead.

The Rust web framework for lazy developers.

This Python-based dashboard-like tool designed to seamlessly interact with the Todoist API. It allows users to fetch project and task data, and generate insightful reports and statistics.

A new type of sorting algorithm. Use large language model (llm like gpt, chat-gpt or others) to sort collections.

A modern commutative diagram editor for the web.

Official implementation of the BRO algorithm

Adds rich IDE support for Hydra config files

The simplest implementation of recent Sparse Attention patterns for efficient LLM inference.

Plugin for Submitit that allows launching jobs on *remote* SLURM clusters

Online Goal-Conditioned Reinforcement Learning in JAX. ICLR 2025 Spotlight.

Supercharge Your Model Training

LLM101n: Let's build a Storyteller

Code to reproduce "Transformers Can Do Arithmetic with the Right Embeddings", McLeish et al (NeurIPS 2024)

Anthropic's educational courses

Training Sparse Autoencoders on Language Models

llama3 implementation one matrix multiplication at a time

A community-maintained Python framework for creating mathematical animations.

Ongoing research training transformer models at scale

Grok open release

Automate browser based workflows with AI

IDEAS scientific achievements

Language models scale reliably with over-training and on downstream tasks