Can Prompt Probe Pretrained Language Models? Understanding the Invisible Risks from a Causal View

TL;DR

This paper examines the risks and biases in prompt-based probing of pretrained language models from a causal perspective, proposing causal interventions to improve evaluation reliability and guide better model assessment.

Contribution

It introduces a causal view to identify biases in prompt probing and proposes debiasing methods, enhancing the reliability of evaluating pretrained language models.

Findings

Identified three critical biases in prompt-based probing.

Proposed causal intervention methods for debiasing.

Provided insights for designing unbiased evaluation datasets.

Abstract

Prompt-based probing has been widely used in evaluating the abilities of pretrained language models (PLMs). Unfortunately, recent studies have discovered such an evaluation may be inaccurate, inconsistent and unreliable. Furthermore, the lack of understanding its inner workings, combined with its wide applicability, has the potential to lead to unforeseen risks for evaluating and applying PLMs in real-world applications. To discover, understand and quantify the risks, this paper investigates the prompt-based probing from a causal view, highlights three critical biases which could induce biased results and conclusions, and proposes to conduct debiasing via causal intervention. This paper provides valuable insights for the design of unbiased datasets, better probing frameworks and more reliable evaluations of pretrained language models. Furthermore, our conclusions also echo that we need to rethink the criteria for identifying better pretrained language models. We openly released the source code and data at https://github.com/c-box/causalEval.