Fast implementation of BERT inference directly on NVIDIA (CUDA, CUBLAS) and Intel MKL
No resources for this project.