English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Causal explanations of the predictions of NLP systems are essential to ensuresafety and establish trust. Yet, existing methods often fall short ofexplaining model predictions effectively or efficiently and are oftenmodel-specific. In this paper, we address model-agnostic explanations,proposing two approaches for counterfactual (CF) approximation. The firstapproach is CF generation, where a large language model (LLM) is prompted tochange a specific text concept while keeping confounding concepts unchanged.While this approach is demonstrated to be very effective, applying LLM atinference-time is costly. We hence present a second approach based on matching,and propose a method that is guided by an LLM at training-time and learns adedicated embedding space. This space is faithful to a given causal graph andeffectively serves to identify matches that approximate CFs. After showingtheoretically that approximating CFs is required in order to construct faithfulexplanations, we benchmark our approaches and explain several models, includingLLMs with billions of parameters. Our empirical results demonstrate theexcellent performance of CF generation models as model-agnostic explainers.Moreover, our matching approach, which requires far less test-time resources,also provides effective explanations, surpassing many baselines. We also findthat Top-K techniques universally improve every tested method. Finally, weshowcase the potential of LLMs in constructing new benchmarks for modelexplanation and subsequently validate our conclusions. Our work illuminates newpathways for efficient and accurate approaches to interpreting NLP systems.

Faithful Explanations of Black-box NLP Models Using LLM-generated Counterfactuals