karpathy / llm.c
LLM training in simple, raw C/CUDA
See what the GitHub community is most excited about today.
LLM training in simple, raw C/CUDA
Efficient GPU kernels for block-sparse matrix multiplication and convolution
[ARCHIVED] Cooperative primitives for CUDA C++. See https://github.com/NVIDIA/cccl
FlashInfer: Kernel Library for LLM Serving
Lightning fast differentiable SSIM.
CUDA Library Samples
Distributed multigrid linear solver library on GPU
CUDA Kernel Benchmarking Library
Instant neural graphics primitives: lightning fast NeRF and more
NCCL Tests
Quantized Attention achieves speedup of 2-5x and 3-11x compared to FlashAttention and xformers, without lossing end-to-end metrics across language, image, and video models.
FlashMLA: Efficient MLA decoding kernels
cuVS - a library for vector search and clustering on the GPU
Tile primitives for speedy kernels