Is your feature request related to a problem? Please describe.
It is not clear from the documentation and the sample code, if the forecast generation can be performed on a GPU, multiple GPUs, or multiple GPUs in multiple nodes. If this is the case, please add some descriptions on how to achieve this.
Describe the solution you'd like
Maybe a simple device_map="auto" argument support in the from_pretrained method that automatically loads the model and the subsequent data onto a GPU/s.
Additional context
I respect the great work you have done! Especially around enabling multivariate modelling.
Is your feature request related to a problem? Please describe. It is not clear from the documentation and the sample code, if the forecast generation can be performed on a GPU, multiple GPUs, or multiple GPUs in multiple nodes. If this is the case, please add some descriptions on how to achieve this.
Describe the solution you'd like Maybe a simple device_map="auto" argument support in the from_pretrained method that automatically loads the model and the subsequent data onto a GPU/s.
Additional context I respect the great work you have done! Especially around enabling multivariate modelling.