A parallel version of GRU (Gated Recurrent Unit) model using CUDA, by Siping Wang and Junyan Pu. Final project for CMU 15-618.

https://wangsiping97.github.io/15618/

Note:

1. The NVIDIA CUDA C/C++ Compiler (NVCC) needs to be added to PATH.
2. The CUDA shared library must be loaded at runtime.

$ cd training
$ make
$ ./cudaGRU -g <1 for GPU or 0 for CPU> -i <number of iterations>

$ cd inference
$ make
$ ./cudaInference -g <1 for GPU or 0 for CPU>