A natively parallel dataloader for Python, written in Rust. Serving data at GB/s speeds, while covering aspect ratio bucketing, crop and resize for image ML workloads.

Framework based on a vector dabase to store, manage and curate large image datasets

H-Net: Hierarchical Network with Dynamic Chunking

tiny vision language model

Fast and flexible reference benchmarks

Burn - A Flexible and Comprehensive Deep Learning Framework in Rust

Hackable and optimized Transformers building blocks, supporting a composable construction.

Generate text captions for images from their embeddings.

PyTorch extensions for high performance and large scale training.

A PyTorch implementation of Connected Components Labeling

Development repository for the Triton language and compiler

Plenoxels: Radiance Fields without Neural Networks, Code release WIP

a Lightweight library for sequential learning agents, including reinforcement learning

PyTorch implementation of diffusion models.

Tensors and Dynamic neural networks in Python with strong GPU acceleration