Currently any deep learning model in SMILER expects a CUDA-compatible GPU to be available, and SMILER will fail if that is unavailable. In order to more flexibly support a wider range of operating systems and machines, it would be helpful to have the option to run models in CPU mode.
We may want to have this as a user-specified parameter, or base it off of a systems check to see if a CUDA-compatible GPU is available. Some research will likely need to be done to see which is the most feasible to implement.
Currently any deep learning model in SMILER expects a CUDA-compatible GPU to be available, and SMILER will fail if that is unavailable. In order to more flexibly support a wider range of operating systems and machines, it would be helpful to have the option to run models in CPU mode.
We may want to have this as a user-specified parameter, or base it off of a systems check to see if a CUDA-compatible GPU is available. Some research will likely need to be done to see which is the most feasible to implement.