Packages

Production ML model serving on the BEAM. Serve PyTorch models faster than Python with pre-compiled graph execution, AOTI compiled inference, and OTP fault tolerance.

Current section

Files

Jump to
extorch lib extorch native aoti.ex
Raw

lib/extorch/native/aoti.ex

defmodule ExTorch.Native.AOTI do
@moduledoc false
defmacro __using__(_opts) do
quote do
@doc false
def aoti_is_available(), do: :erlang.nif_error(:nif_not_loaded)
@doc false
def aoti_load(_path, _model_name, _device_index), do: :erlang.nif_error(:nif_not_loaded)
@doc false
def aoti_forward(_model, _inputs), do: :erlang.nif_error(:nif_not_loaded)
@doc false
def aoti_get_metadata_keys(_model), do: :erlang.nif_error(:nif_not_loaded)
@doc false
def aoti_get_metadata_value(_model, _key), do: :erlang.nif_error(:nif_not_loaded)
@doc false
def aoti_get_constant_fqns(_model), do: :erlang.nif_error(:nif_not_loaded)
end
end
end