In the pursuit of a deeper understanding of a model's behaviour, there is recent impetus for developing suites of probes aimed at diagnosing models beyond simple metrics like accuracy or BLEU. This paper takes a step back and asks an important and timely question: how reliable are these diagnostics in providing insight into models and training setups? We critically examine three recent diagnostic tests for pre-trained language models, and find that likelihood-based and representation-based model diagnostics are not yet as reliable as previously assumed. Based on our empirical findings, we also formulate recommendations for practitioners and researchers.
@article{arxiv.2105.05641,
title = {How Reliable are Model Diagnostics?},
author = {Vamsi Aribandi and Yi Tay and Donald Metzler},
journal= {arXiv preprint arXiv:2105.05641},
year = {2022}
}