Function supported_backends

pub fn supported_backends() -> Backends

We target Vulkan for native and WebGL for wasm. We only use Vulkan for native because that allows us to use the PyTorch and wgpu interoperability