Fetching the paper…

Confidence Scoring Using Whitebox Meta-models with Linear Classifier Probes · Around