Re-Examining Calibration: The Case of Question Answering
Chenglei Si, Chen Yang Zhao, Sewon Min, Jordan Lee Boyd-Graber · 2022
For users to trust model predictions, they need to understand model outputs, particularly their confidence-calibration aims to adjust (calibrate) models' confidence to match expected accuracy.We argue that the traditional calibration evaluation does not promote effective calibrations: for example, it can encourage always assigning a mediocre confidence score to all predictions, which does not help users distinguish correct predictions from wrong ones.Building on those observations, we propose a new calibration metric, MACROCE, that better captures whether the model assigns low confidence to wrong predictions and high confidence to correct predictions.Focusing on the practical application of open-domain question answering, we examine conventional calibration methods applied on the widely-used retrieverreader pipeline, all of which do not bring significant gains under our new MACROCE metric.Toward better calibration, we propose a new calibration method (CONSCAL) that uses not just final model predictions but whether multiple model checkpoints make consistent predictions.Altogether, we provide an alternative view of calibration along with a new metric, re-evaluation of existing calibration methods on our metric, and proposal for a more effective calibration method. 1