promptdojo_

The training and prediction loop — step 2 of 7

What makes the threshold approach "training" rather than a hardcoded rule?