Ok but in my case the 1st pipeline has no output because pytorch lightning handles alone the model checkpoint save, and the 2nd pipeline has also no output because onnx handles the model conversion alone. They both use a variable written in

parameters

, Maybe they could output something useful.