How to get libtorch support for XPU?

[JamasChuang94]

Describe the issue

How can I get libtorch to support XPU? The model.generate in python is too slow. When there are more than 20 words, it takes about 7 seconds. I am looking for a faster XPU version of libtorch.

[devpramod]

We don’t have plans to support xpu with libtorch
Could you please try using AOTInductor (https://docs.pytorch.org/docs/stable/torch.compiler_aot_inductor.html)