Reference PyTorch implementation and models for DINOv3

LLM training in simple, raw C/CUDA

Fast Hadamard transform in CUDA, with a PyTorch interface

Full-gradient saliency maps

Pytorch library for fast transformer implementations

a language for fast, portable data-parallel computation

higher is a pytorch library allowing users to obtain higher order gradients over losses spanning training loops rather than individual training steps.

A domain specific language to express machine learning workloads.

xv6 OS

Tensors and Dynamic neural networks in Python with strong GPU acceleration