Model returned by the select_model_inputs_outputs() fails onnx.checker.check_model()

[TrigonaMinima]

The model created after the use of select_model_inputs_outputs as described on Select outputs fails the onnx.checker.check_model check.

I get the following error:

File .../site-packages/onnx/checker.py:179, in check_model(model, full_check, skip_opset_compatibility_check, check_custom_domain)
    175 if sys.getsizeof(protobuf_string) > MAXIMUM_PROTOBUF:
    176     raise ValueError(
    177         "This protobuf of onnx model is too large (>2GB). Call check_model with model path instead."
    178     )
--> 179 C.check_model(
    180     protobuf_string,
    181     full_check,
    182     skip_opset_compatibility_check,
    183     check_custom_domain,
    184 )

ValidationError: Field 'type' of 'value_info' is required but missing.

Looking at the graph, before-after of the relevant section is as follows:

before

  output {
    name: "probabilities"
    type {
      tensor_type {
        elem_type: 1
        shape {
          dim {
          }
          dim {
            dim_value: 3
          }
        }
      }
    }
  }

after

  output {
    name: "probabilities"
  }

Will this have any impact in production when used in a runtime?

[xadupre]

output type should not be needed by the runtime. I can be inferred. I probably changed it to avoid setting the wrong type. You should try this function https://onnx.ai/onnx/api/utils.html#extract-model which was introduced in onnx.