Tile primitives for speedy kernels

Style your webpage like Edward Tufte’s handouts.

Fast CUDA matrix multiplication from scratch

a minimal cache manager for PagedAttention, on top of llama3.

Renders music in HTML.

3D framework for Svelte

A native PyTorch Library for large model training

A modern SaaS template/boilerplate built with SvelteKit, Tailwind, and Supabase. Includes marketing page, blog, subscriptions, auth, user dashboard, user settings, pricing page, and more.

Make PyTorch models up to 40% faster! Thunder is a source to source compiler for PyTorch. It enables using different hardware executors at once; across one or thousands of GPUs.

Solve puzzles. Improve your pytorch.

This repository contains the experimental PyTorch native float8 training UX

Minimal (400 LOC) implementation Maximum (multi-node, FSDP) GPT training

Read Japanese manga inside browser with selectable text.

Localized watermarking for AI-generated speech audios, with SOTA on robustness and very fast detector

Universal Tensor Operations in Einstein-Inspired Notation for Python.

A swift and unified toolkit for symbolic music processing

Development repository for the Triton language and compiler

Particles simulation with WebGPU

3D Visualization of an GPT-style LLM

Python launcher of animated MIDI player by @cifkao & @magenta

Simple and efficient pytorch-native transformer text generation in <1000 LOC of python.

Fast and memory-efficient exact attention

Contrastive Language-Audio Pretraining

Fast, collaborative live terminal sharing over the web

unofficial MaskGIT reproduction in PyTorch

A minimal PyTorch implementation of probabilistic diffusion models for 2D datasets.

MIDI / symbolic music tokenizers for Deep Learning models 🎶

Official inference library for Mistral models