Plans to support AMD

[ljt019]

Are there any plans to support inference with AMD gpus? As far as I can tell currently both here and in Candle there is only support for NVIDIA gpus.

The closest repo I found working on support/bindings for rust was this crate here:
https://github.com/radudiaconu0/rocm-rs

[MarcusDunn]

If llama.cpp has support for AMD, we can realistically support it - PRs are always welcome.