Closed gyeongin closed 5 years ago
This PR adds functions for transferring tensors between CPU and CUDA devices. I commented this line to avoid build error. We need to extend this transfer function later to support multiple CUDA devices (e.g., "cuda:3").
This PR adds functions for transferring tensors between CPU and CUDA devices. I commented this line to avoid build error. We need to extend this transfer function later to support multiple CUDA devices (e.g., "cuda:3").