Splitting the pipeline for different models

Yeah, it’s possible to do that.
You can also refer Single source and multi inference models in parallel - #7 by chauncey.wang