Use of CUDA for the Continuous Space Language Model

Abstract

Abstract-The training phase of the Continuous Space Language Model (CSLM) was implemented in the NVIDIA hardware/software architecture Compute Unified Device Architecture (CUDA). Implementation was accomplished using a combination of CUBLAS library routines and CUDA kernel calls on three different CUDA enabled devices of varying compute capability and a time savings over the traditional CPU approach demonstrated.