NLI Data Sanity Check: Assessing the Effect of Data Corruption on Model Performance

Aarne Talman, Marianna Apidianaki, Stergios Chatzikyriakidis, Jörg Tiedemann


Abstract
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 is still 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.
Anthology ID:
2021.nodalida-main.28
Volume:
Proceedings of the 23rd Nordic Conference on Computational Linguistics (NoDaLiDa)
Month:
May 31--2 June
Year:
2021
Address:
Reykjavik, Iceland (Online)
Venue:
NoDaLiDa
SIG:
Publisher:
Linköping University Electronic Press, Sweden
Note:
Pages:
276–287
Language:
URL:
https://aclanthology.org/2021.nodalida-main.28
DOI:
Bibkey:
Cite (ACL):
Aarne Talman, Marianna Apidianaki, Stergios Chatzikyriakidis, and Jörg Tiedemann. 2021. NLI Data Sanity Check: Assessing the Effect of Data Corruption on Model Performance. In Proceedings of the 23rd Nordic Conference on Computational Linguistics (NoDaLiDa), pages 276–287, Reykjavik, Iceland (Online). Linköping University Electronic Press, Sweden.
Cite (Informal):
NLI Data Sanity Check: Assessing the Effect of Data Corruption on Model Performance (Talman et al., NoDaLiDa 2021)
Copy Citation:
PDF:
https://aclanthology.org/2021.nodalida-main.28.pdf
Code
 Helsinki-NLP/nli-data-sanity-check
Data
ANLIGLUEMultiNLISNLISuperGLUE