[2003.07892] Calibration of Pre-trained Transformers

arxiv.org · 4,533 words · saved by 1 readers

Abstract:Pre-trained Transformers are now ubiquitous in natural language processing, but despite their high end-task performance, little is known empirically about whether they are calibrated. Specifically, do these models' posterior probabilities provide an accurate empirical measure of how likely the model is to be correct on a given example? We focus on BERT and RoBERTa in this work, and analyze their calibration across three tasks: natural language inference, paraphrase detection, and commonsense reasoning. For each task, we consider in-domain as well as challenging out-of-domain settings, where models face more examples they should be uncertain about. We show that: (1) when used out-of-the-box, pre-trained models are calibrated in-domain, and compared to baselines, their calibration error out-of-domain can be as much as 3.5x lower; (2) temperature scaling is effective at further reducing calibration error in-domain, and using label smoothing to deliberately increase empirical uncertainty helps calibrate posteriors out-of-domain.

Calibration of Pre-trained Transformers Shrey Desai and Greg Durrett Department of Computer Science The University of Texas at Austin shreydesai@utexas.edu gdurrett@cs.utexas.edu Abstract 2019; Kovaleva et al., 2019). One step towards un-…

[2003.07892] Calibration of Pre-trained Transformers

related reading