Final graph through `tensor.long()`

Exported graph: graph(%d1 : Int(requires_grad=0, device=cpu),
      %d2 : Int(requires_grad=0, device=cpu),
      %d3 : Int(requires_grad=0, device=cpu)):
  %/Cast_output_0 : Long(requires_grad=0, device=cpu) = onnx::Cast[to=7, onnx_name="/Cast"](%d1), scope: __main__.DummyModel::
  %/Cast_1_output_0 : Long(requires_grad=0, device=cpu) = onnx::Cast[to=7, onnx_name="/Cast_1"](%d2), scope: __main__.DummyModel::
  %/Cast_2_output_0 : Long(requires_grad=0, device=cpu) = onnx::Cast[to=7, onnx_name="/Cast_2"](%d3), scope: __main__.DummyModel::
  %onnx::Unsqueeze_6 : Long(1, strides=[1], device=cpu) = onnx::Constant[value={0}]()
  %/Unsqueeze_output_0 : Long(1, strides=[1], device=cpu) = onnx::Unsqueeze[onnx_name="/Unsqueeze"](%/Cast_output_0, %onnx::Unsqueeze_6), scope: __main__.DummyModel::
  %onnx::Unsqueeze_8 : Long(1, strides=[1], device=cpu) = onnx::Constant[value={0}]()
  %/Unsqueeze_1_output_0 : Long(1, strides=[1], device=cpu) = onnx::Unsqueeze[onnx_name="/Unsqueeze_1"](%/Cast_1_output_0, %onnx::Unsqueeze_8), scope: __main__.DummyModel::
  %onnx::Unsqueeze_10 : Long(1, strides=[1], device=cpu) = onnx::Constant[value={0}]()
  %/Unsqueeze_2_output_0 : Long(1, strides=[1], device=cpu) = onnx::Unsqueeze[onnx_name="/Unsqueeze_2"](%/Cast_2_output_0, %onnx::Unsqueeze_10), scope: __main__.DummyModel::
  %/Concat_output_0 : Long(3, strides=[1], device=cpu) = onnx::Concat[axis=0, onnx_name="/Concat"](%/Unsqueeze_output_0, %/Unsqueeze_1_output_0, %/Unsqueeze_2_output_0), scope: __main__.DummyModel:: # /home/nbeneitez/Documentos/work/repos/returnn_pytorch/testtorchonnx.py:25:0
  %casted_data : Float(*, *, *, strides=[4, 2, 1], requires_grad=0, device=cpu) = onnx::ConstantOfShape[value={2}, onnx_name="/ConstantOfShape"](%/Concat_output_0), scope: __main__.DummyModel:: # /home/nbeneitez/Documentos/work/repos/returnn_pytorch/testtorchonnx.py:25:0
  return (%casted_data)