A high-throughput and memory-efficient inference and serving engine for LLMs

My personal dotfiles

The math library of Lean 4

Disaster Prevention Information Platform (防災資訊平台)

SystemC Reference Implementation

LLM training in simple, raw C/CUDA

Tensors and Dynamic neural networks in Python with strong GPU acceleration

On-device AI across mobile, embedded and edge for PyTorch

MLX: An array framework for Apple silicon

Communication framework for RTL simulation and emulation.

Vector Acceleration IP core for RISC-V*

The PULP Ara is a 64-bit Vector Unit, compatible with the RISC-V Vector Extension Version 1.0, working as a coprocessor to CORE-V's CVA6 core

Bringup-Bench is a collection of standalone minimal library and system dependence benchmarks useful for bringing up newly designed CPUs, accelerators, compilers and operating systems. You probably …

AITemplate is a Python framework which renders neural network into high performance CUDA/HIP C++ code. Specialized for FP16 TensorCore (NVIDIA GPU) and MatrixCore (AMD GPU) inference.

Free Vue.js 3.x Tailwind 3.x admin dashboard template with dark mode. Vite builds. Pinia state. Laravel integration available

⚡ Dynamically generated stats for your github readmes

Semantic segmentation models with 500+ pretrained convolutional and transformer-based backbones.

CKIP Neural Chinese Word Segmentation, POS Tagging, and NER

"jf open-huninn" ("jf open 粉圓") is an open-source Traditional Chinese typeface based on Kosugi Maru and Varela Round, adjusted by the designers at justfont for everyday use in Taiwan.