FLUKE: A Linguistically-Driven and Task-Agnostic Framework for Robustness Evaluation
This work addresses the need for systematic robustness testing in NLP to understand model behaviors, though it is incremental as it builds on existing evaluation methods.
The authors tackled the problem of evaluating model robustness in NLP by introducing FLUKE, a framework for systematic minimal variations across linguistic levels, and found that models are more brittle to natural modifications like syntax changes and that robustness varies significantly by task, with LLMs showing surprising brittleness in some cases.
We present FLUKE (Framework for LingUistically-driven and tasK-agnostic robustness Evaluation), a framework for assessing model robustness through systematic minimal variations of test data. FLUKE introduces controlled variations across linguistic levels -- from orthography to dialect and style -- and leverages large language models (LLMs) with human validation to generate modifications. We demonstrate FLUKE's utility by evaluating both fine-tuned models and LLMs across six diverse NLP tasks (four classification and two generation tasks), and reveal that (1) the impact of linguistic variations is highly task-dependent, with some tests being critical for certain tasks but irrelevant for others; (2) LLMs still exhibit significant brittleness to certain linguistic variations, with reasoning LLMs surprisingly showing less robustness on some tasks compared to base models; (3) models are overall more brittle to natural, fluent modifications such as syntax or style changes (and especially to negation), compared to corruption-style tests such as letter flipping; (4) the ability of a model to use a linguistic feature in generation does not correlate to its robustness to this feature on downstream tasks. These findings highlight the importance of systematic robustness testing for understanding model behaviors.