Explore Projects
Discover 133 open source projects
Showing 41-60 of 133 projects
gpustack/gpustack
Optimize AI inference performance on GPUs with this Python library for selecting and tuning inference engines.
NVIDIA/nccl
An optimized library for efficient multi-GPU communication in deep learning applications.
NVlabs/tiny-cuda-nn
A fast C++/CUDA neural network framework for high-performance deep learning and rendering.
OpenNMT/CTranslate2
Fast C++ inference engine for Transformer models, supporting CUDA, MKL, and other optimizations.
ROCm/hip
A C++ interface for portability across different heterogeneous computing platforms like CUDA and HIP.
iree-org/iree
A retargetable MLIR-based machine learning compiler and runtime toolkit for AI/ML developers.
ashvardanian/StringZilla
Accelerate string operations in C, C++, Python, Rust, and more with SIMD and GPU-powered algorithms.
Infatoshi/cuda-course
A CUDA course for developers to learn about GPU computing and parallel processing.
patriciogonzalezvivo/lygia
LYGIA is a flexible, multi-language shader library designed for performance, supporting GLSL, HLSL, Metal, WGSL, WEGL, and CUDA.
Jittor/jittor
Jittor is a high-performance deep learning framework based on JIT compiling and meta-operators.
HazyResearch/ThunderKittens
A library of tile primitives for building speedy CUDA kernels, useful for vibe coders working on AI tools.
thu-ml/SageAttention
Quantized attention that achieves 2-5x speedup over FlashAttention for language, image, and video models.
NVIDIA/TransformerEngine
A high-performance Transformer library for accelerating AI models on NVIDIA GPUs, including low-precision support.
NVIDIA/cuda-python
CUDA Python: A library that brings the power of CUDA to Python, enabling high-performance GPU acceleration.
LeelaChessZero/lc0
Open-source neural network chess engine with GPU acceleration and broad hardware support.
pytorch/TensorRT
PyTorch compiler for NVIDIA GPUs using TensorRT, enabling efficient deep learning inference on CUDA hardware.
NVIDIA/MinkowskiEngine
A high-performance, auto-diff neural network library for 3D and 4D sparse tensor computations.
BBuf/how-to-optim-algorithm-in-cuda
This repository provides guidance on optimizing algorithms for CUDA, a framework for parallel computing on NVIDIA GPUs.
containers/ramalama
RamaLama simplifies local serving of AI models and enables their use for inference in production via containers.
enpeizhao/CVprojects
A collection of computer vision and AI projects in Python, C++, and embedded systems for developers.
Stay in the loop
Get weekly updates on trending AI coding tools and projects.