TabPert: An Effective Platform for Tabular Perturbation
This provides a tool for researchers to improve model evaluation in tabular reasoning, but it is incremental as it builds on existing counterfactual data generation methods.
The authors tackled the problem of evaluating Natural Language Inference models on counterfactual data by developing TabPert, a platform that assists in generating counterfactual tables and hypotheses for assessing tabular reasoning issues, enabling systematic and quantitative exploration of model shortcomings.
To truly grasp reasoning ability, a Natural Language Inference model should be evaluated on counterfactual data. TabPert facilitates this by assisting in the generation of such counterfactual data for assessing model tabular reasoning issues. TabPert allows a user to update a table, change its associated hypotheses, change their labels, and highlight rows that are important for hypothesis classification. TabPert also captures information about the techniques used to automatically produce the table, as well as the strategies employed to generate the challenging hypotheses. These counterfactual tables and hypotheses, as well as the metadata, can then be used to explore an existing model's shortcomings methodically and quantitatively.