NLI Data Sanity Check: Assessing the Effect of Data Corruption on Model Performance
This work addresses the issue of dataset quality for researchers and practitioners in NLP, offering a method to evaluate and improve NLI benchmarks, though it is incremental as it builds on existing corruption techniques.
The authors tackled the problem of assessing whether natural language inference (NLI) datasets effectively test models' meaning understanding by proposing a diagnostics test suite that applies controlled corruption transformations to benchmarks like MNLI and ANLI, finding that high model accuracy on corrupted data indicates dataset biases, while a large decrease suggests proper reasoning challenges.
Pre-trained neural language models give high performance on natural language inference (NLI) tasks. But whether they actually understand the meaning of the processed sequences remains unclear. We propose a new diagnostics test suite which allows to assess whether a dataset constitutes a good testbed for evaluating the models' meaning understanding capabilities. We specifically apply controlled corruption transformations to widely used benchmarks (MNLI and ANLI), which involve removing entire word classes and often lead to non-sensical sentence pairs. If model accuracy on the corrupted data remains high, then the dataset is likely to contain statistical biases and artefacts that guide prediction. Inversely, a large decrease in model accuracy indicates that the original dataset provides a proper challenge to the models' reasoning capabilities. Hence, our proposed controls can serve as a crash test for developing high quality data for NLI tasks.