This repository contains data and scripts to reproduce the results from our paper:
Aarne Talman, Marianna Apidianaki, Stergios Chatzikyriakidis, Jörg Tiedemann. 2022. How Does Data Corruption Affect Natural Language Understanding Models? A Study on GLUE datasets.
A central question in natural language understanding (NLU) research is whether high performance demonstrates the models' strong reasoning capabilities. We present an extensive series of controlled experiments where pre-trained language models are exposed to data that have undergone specific corruption transformations. The transformations involve removing instances of specific word classes and often lead to non-sensical sentences. Our results show that performance remains high for most GLUE tasks when the models are fine-tuned or tested on corrupted data, suggesting that the models leverage other cues for prediction even in non-sensical contexts. Our proposed data transformations can be used as a diagnostic tool for assessing the extent to which a specific dataset constitutes a proper testbed for evaluating models' language understanding capabilities.
Install the dependencies by running:
pip install -r requirements.txtRun the experiments using the following command:
bash run_experiment.shrun_experiment.sh starts a fine-tuning job for each configuration and only works in an environment where you have access to a lot of GPU instances managed with an orchestration system like SLURM. To run a single configuration, you can modify train.sh and tun it:
bash train.shThe Python script run_corrupt_glue.py is a modified version of the
run_glue.py script by Huggingface available in their Text classification examples.
@misc{talman_et_al2022,
title={How Does Data Corruption Affect Natural Language Understanding Models? A Study on GLUE datasets},
author={Aarne Talman and Marianna Apidianaki and Stergios Chatzikyriakidis and J\"org Tiedemann},
year={2022},
eprint={2201.04467},
archivePrefix={arXiv},
primaryClass={cs.CL}
}