https://hgpu.org/?p=7568
Efficient Parallelization of Natural Language Applications using GPUs