How to use the visualdl.server.model.onnx.onnx_pb2.ModelProto function in visualdl

def load_from_string(s):
    '''
    Loads a binary string that stores onnx model

    @params
    Takes a string object containing protobuf
    @return ONNX ModelProto object
    '''
    model = ModelProto()
    decoded = model.ParseFromString(s)
    # in python implementation ParseFromString returns None
    if decoded is not None and decoded != len(s):
        return None

#         raise google.protobuf.message.DecodeError(
#             "Protobuf decoding consumed too few bytes: {} out of {}".format(
#                 decoded, len(s)))
    return model