flashinfer-ai / flashinfer
FlashInfer: Kernel Library for LLM Serving
See what the GitHub community is most excited about today.
FlashInfer: Kernel Library for LLM Serving
NCCL Tests
DeepEP: an efficient expert-parallel communication library
CUDA accelerated rasterization of gaussian splatting
Instant neural graphics primitives: lightning fast NeRF and more
NVIDIA cuOpt is an open-source GPU-accelerated optimization engine delivering near real-time solutions for complex decision-making challenges.
LLM training in simple, raw C/CUDA
CUDA Kernel Benchmarking Library
CUDA Library Samples
FlashMLA: Efficient MLA decoding kernels
Neighborhood Attention Extension. Bringing attention to a neighborhood near you!
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.
FSA/FST algorithms, differentiable, with PyTorch compatibility.
Causal depthwise conv1d in CUDA, with a PyTorch interface