karpathy / llm.c
LLM training in simple, raw C/CUDA
See what the GitHub community is most excited about this month.
LLM training in simple, raw C/CUDA
GPU accelerated decision optimization
DeepEP: an efficient expert-parallel communication library
CUDA accelerated rasterization of gaussian splatting
FlashInfer: Kernel Library for LLM Serving
Instant neural graphics primitives: lightning fast NeRF and more
DeepGEMM: clean and efficient FP8 GEMM kernels with fine-grained scaling
RCCL Performance Benchmark Tests
Causal depthwise conv1d in CUDA, with a PyTorch interface
NCCL Tests
Distributed multigrid linear solver library on GPU
Fast CUDA matrix multiplication from scratch
Graphics Processing Units Molecular Dynamics
[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.