The simplest way to try out your finetuned model for inference is to use it in a [pipeline].