A Unified and Flexible Inference Engine with Hybrid Cache Acceleration and Parallelism for 🤗Diffusers.

An open-source AI agent that brings the power of Gemini directly into your terminal.

A PyTorch native platform for training generative AI models

A general fine-tuning kit geared toward diffusion models.

CUDA Templates and Python DSLs for High-Performance Linear Algebra

A lightweight 2D graphics library for modern GPUs, delivering high-performance text, image, and vector rendering across major platforms.

JavaScript 3D Library.

DeepSpeed is a deep learning optimization library that makes distributed training and inference easy, efficient, and effective.

Context7 MCP Server -- Up-to-date code documentation for LLMs and AI code editors

CUDA Core Compute Libraries

Development repository for the Triton language and compiler

Pear 🍐 is extension for music player

ONNX Runtime: cross-platform, high performance ML inferencing and training accelerator

Qwen-Image is a powerful image generation foundation model capable of complex text rendering and precise image editing.

Efficient Triton Kernels for LLM Training

The most customisable and low-latency cross platform/shell prompt renderer

Making it easier to work with shaders

Claude Code is an agentic coding tool that lives in your terminal, understands your codebase, and helps you code faster by executing routine tasks, explaining complex code, and handling git workflo…

A generative world for general-purpose robotics & embodied AI learning.

The fundamental package for scientific computing with Python.

A collection of useful functions for 3D vision & graphics research in Python.

A part-based 3D generation framework & the largest and most comprehensively annotated 3D part dataset.

[SIGGRAPH Asia 2025] OmniPart: Part-Aware 3D Generation with Semantic Decoupling and Structural Cohesion

SANA: Efficient High-Resolution Image Synthesis with Linear Diffusion Transformer

GenAI Processors is a lightweight Python library that enables efficient, parallel content processing.

CUDA Marching Cube & Dual Marching Cube for Sparse Voxel SDFs

The public CGAL repository, see the README below

🚀 Efficient implementations of state-of-the-art linear attention models

[ICLR2025, ICML2025, NeurIPS2025 Spotlight] Quantized Attention achieves speedup of 2-5x compared to FlashAttention, without losing end-to-end metrics across language, image, and video models.

Import a 3D Model and automatically assign and export animations