UNIREX: A Unified Learning Framework for Language Model Rationale Extraction

TL;DR

UNIREX is a flexible learning framework that jointly trains language models and rationale extractors to improve faithfulness, plausibility, and task performance, outperforming baselines across multiple datasets.

Contribution

It introduces a unified framework for rationale extraction that replaces heuristics with learned extractors optimized for multiple desiderata.

Findings

UNIREX outperforms baselines by 32.9% NRG on average.

UNIREX-trained extractors generalize to unseen datasets and tasks.

The Normalized Relative Gain metric facilitates multi-objective comparison.

Abstract

An extractive rationale explains a language model's (LM's) prediction on a given task instance by highlighting the text inputs that most influenced the prediction. Ideally, rationale extraction should be faithful (reflective of LM's actual behavior) and plausible (convincing to humans), without compromising the LM's (i.e., task model's) task performance. Although attribution algorithms and select-predict pipelines are commonly used in rationale extraction, they both rely on certain heuristics that hinder them from satisfying all three desiderata. In light of this, we propose UNIREX, a flexible learning framework that generalizes rationale extractor optimization as follows: (1) specify architecture for a learned rationale extractor; (2) select explainability objectives (i.e., faithfulness and plausibility criteria); and (3) jointly the train task model and rationale extractor on the task using the selected objectives. UNIREX enables replacing prior works' heuristic design choices with a generic learned rationale extractor in (1) and optimizing it for all three desiderata in (2)-(3). To facilitate comparison between methods with respect to multiple desiderata, we introduce the Normalized Relative Gain (NRG) metric. Across five text classification datasets, our best UNIREX configuration outperforms baselines by an average of 32.9% NRG. Plus, we find that UNIREX-trained rationale extractors can even generalize to unseen datasets and tasks.