Summary of An Investigation Of Structures Responsible For Gender Bias in Bert and Distilbert, by Thibaud Leteno et al.
An investigation of structures responsible for gender bias in BERT and DistilBERT
by Thibaud Leteno, Antoine Gourru, Charlotte Laclau, Christophe Gravier
First submitted to arxiv on: 12 Jan 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Computers and Society (cs.CY); Machine Learning (cs.LG)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The paper presents an empirical exploration of gender bias in large Transformer-based Pre-trained Language Models (PLMs) and their compressed counterparts. Specifically, it investigates whether distillation accentuates or mitigates gender bias in BERT and DistilBERT models. The study finds that every attention head uniformly encodes bias, except for underrepresented classes with high imbalance of the sensitive attribute. The distilled model is found to produce bias more homogeneously than the original BERT model. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper looks at how language models like BERT and its smaller version, DistilBERT, treat men and women differently. It wants to know if making these models smaller makes them more or less likely to have biases against certain groups. The study found that every part of the model is biased in some way, but it’s only really a problem when it’s dealing with groups that are already underrepresented. The smaller DistilBERT model seems to make things worse by spreading the bias out more evenly. |
Keywords
* Artificial intelligence * Attention * Bert * Distillation * Transformer
