Alvaro Gonzalez-Jimenez

Basically you have installed torch without CUDA maybe because you do not have a GPU. Either you reinstall torch with CUDA or you change the code to run in CPU.

You have to change the variable `gpu_id` in the code to 'cpu'. Here an easy example so you can understand: ```{python} #To GPU with the device number 0 tensor =...

@samTheHacke The error says that could not found NVIDIA Driver which means that still the code is checking for GPU instead of CPU. Did you install the CPU version of...

Hello @Marcelo5444, We have not tested this loss in a multi-class setting. Since we have reparametrized the covariance matrix, I suspect that the loss might be unstable and not provide...

@Kishore4399 It's normal, please see #4