LLM training in simple, raw C/CUDA

DeepEP: an efficient expert-parallel communication library

DeepGEMM: clean and efficient FP8 GEMM kernels with fine-grained scaling

FlashInfer: Kernel Library for LLM Serving

[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.

Static suckless single batch CUDA-only qwen3-0.6B mini inference engine

Reference implementation of Megalodon 7B model

flash attention tutorial written in python, triton, cuda, cutlass

llama3.cuda is a pure C/CUDA implementation for Llama 3 model.

A quantization algorithm for LLM