Diversify Your Datasets: Analyzing Generalization via Controlled Variance in Adversarial Datasets

Abstract

Phenomenon-specific "adversarial" datasets have been recently designed to perform targeted stress-tests for particular inference types. Recent work (Liu et al., 2019a) proposed that such datasets can be utilized for training NLI and other types of models, often allowing to learn the phenomenon in focus and improve on the challenge dataset, indicating a "blind spot" in the original training data. Yet, although a model can improve in such a training process, it might still be vulnerable to other challenge datasets targeting the same phenomenon but drawn from a different distribution, such as having a different syntactic complexity level. In this work, we extend this method to drive conclusions about a model's ability to learn and generalize a target phenomenon rather than to "learn" a dataset, by controlling additional aspects in the adversarial datasets. We demonstrate our approach on two inference phenomena -dative alternation and numerical reasoning, elaborating, and in some cases contradicting, the results of Liu et al.. Our methodology enables building better challenge datasets for creating more robust models, and may yield better model understanding and subsequent overarching improvements.

Injecting Numerical Reasoning Skills into Language Models

“…A related line of work has been analyzing the mathematical reasoning abilities of neural models over text (Wallace et al, 2019;Rozen et al, 2019;Ravichander et al, 2019), and on arithmetic problems (Saxton et al, 2019;Amini et al, 2019;Lample and Charton, 2020).…”

Do Neural Models Learn Systematicity of Monotonicity Inference in Natural Language?

“…Regarding assessment of the behavior of modern language models, Linzen et al (2016), Goldberg (2019) investigated their syntactic capabilities by testing such models on subject-verb agreement tasks. Many studies of NLI tasks (Liu et al, 2019;Glockner et al, 2018;Poliak et al, 2018;Tsuchiya, 2018;McCoy et al, 2019;Rozen et al, 2019;Ross and Pavlick, 2019) have provided evaluation methodologies and found that current NLI models often fail on particular inference types, or that they learn undesired heuristics from the training set. In particular, recent works (Yanaka et al, 2019a,b;Richardson et al, 2020) have evaluated models on monotonicity, but did not focus on the ability to generalize to unseen combinations of patterns.…”

The Curse of Performance Instability in Analysis Datasets: Consequences, Source, and Suggestions

“…To mitigate these problems, Liu et al (2019a) introduced a systematic task-agnostic method to analyze datasets. Rozen et al (2019) further explain how to improve challenging datasets and why diversity matters. Geva et al (2019) suggest that the training and test data should be from exclusive annotators to avoid annotator bias.…”