Quantized Attention that achieves speedups of 2.1x and 2.7x compared to FlashAttention2 and xformers, respectively, without lossing end-to-end metrics across various models.

Entropy Based Sampling and Parallel CoT Decoding

🤗 LeRobot: Making AI for Robotics more accessible with end-to-end learning

A golang-based data loader which can be used from Python. Focused on a VectorDB stack at the moment, fetching and processing data per sample at GB/s speeds.

extensible collectives library in triton

Official software repository of S. Bruch, F. M. Nardini, C. Rulli, and R. Venturini, "Efficient Inverted Indexes for Approximate Retrieval over Learned Sparse Representations". Long Paper @ ACM SIG…

Late Interaction Models Training & Retrieval

Materials for the Ultimate Hybrid Search Workshop

A Rust HTTP server for Python applications

Intro to leetcodes. Basic techniques, quicksort and hash structures implementation, space and time complexities.

MLX: An array framework for Apple silicon

Einsum-like high-level array sharding API for JAX

Sparse autoencoders

Fast lexical search implementing BM25 in Python using Numpy, Numba and Scipy

SPLADE: sparse neural search (SIGIR21, SIGIR22)

Rapidly build AI apps in Python

NLP with Rust for Python 🦀🐍

Neural Search

A self-paced course to learn Rust, one exercise at a time.

depyf is a tool to help you understand and adapt to PyTorch compiler torch.compile.

A modern model graph visualizer and debugger

3D Gaussian Splatting in JAX

FastKAN: Very Fast Implementation of Kolmogorov-Arnold Networks (KAN)

An efficient pure-PyTorch implementation of Kolmogorov-Arnold Network (KAN).

Tile primitives for speedy kernels

A multi-level tensor algebra superoptimizer

Code for our EMNLP 2023 Paper: "LLM-Adapters: An Adapter Family for Parameter-Efficient Fine-Tuning of Large Language Models"

seqax = sequence modeling + JAX