[torch.logit] how to use one input version interface ?

According the torch.logit.html, the eps is optional

torch.logit(input, eps=None, *, out=None) → Tensor

but I get [error info ] (error: custom op 'torch.aten.logit' expected 2 operands) if don't input the 2nd input

func.func @torch_aten_logit_with_one_input(%arg0: !torch.vtensor<[2, 2, 8],f32>) -> !torch.vtensor<[2, 2, 8],f32>  {
  %0 = torch.aten.logit %arg0 : !torch.vtensor<[2, 2, 8],f32> -> !torch.vtensor<[2, 2, 8],f32>
  return %0 : !torch.vtensor<[2, 2, 8],f32>
} 

• I can work around it to use explicit %none for the 2nd optional input, but indeed here input 2 inputs.

func.func @torch_aten_logit_with_one_input(%arg0: !torch.vtensor<[2, 2, 8],f32>, %arg1: !torch.none) -> !torch.vtensor<[2, 2, 8],f32> attributes {hacc.entry} {
  %none = torch.constant.none
  %0 = torch.aten.logit %arg0, %none : !torch.vtensor<[2, 2, 8],f32>, !torch.none -> !torch.vtensor<[2, 2, 8],f32>
  return %0 : !torch.vtensor<[2, 2, 8],f32>
} 

Start by tracing the torch.aten.logit operation from the supplied IR examples and inspect how its optional eps operand is represented. Verify the documented one-input form and the explicit !torch.none form against the operation's existing tests or validation path; done means both forms are handled consistently.