Apply suggestion from @blainekasten

Author: blainekasten

src/together/lib/cli/api/fine_tuning/create.py

@@ -89,7 +89,7 @@ async def create(
         Parameter(converter=int_or_max_converter, name=["--batch-size", "-b"], help="Train batch size"),
     ] = "max",
     gradient_accumulation_steps: Annotated[
-        int,
+        Optional[int],
         Parameter(help="Number of gradient accumulation steps (increases effective batch size without more memory)"),
     ] = None,
     learning_rate: Annotated[