mlx-bun library API
    Preparing search index...
    ConvertDtype: "float16" | "bfloat16" | "float32"

    The floating dtypes mlx_lm.convert --dtype accepts.