3 ms·Run High-Performance LLM Inference Kernels from Nvidia Using FlashInfer1 points by mfiguiere 1y ago