@inproceedings{barkan2021grad-sam,
author = {Barkan, Oren and Hauon, Edan and Caciularu, Avi and Katz, Ori and Malkiel, Itzik and Armstrong, Omri and Koenigstein, Noam},
title = {Grad-SAM: Explaining Transformers via Gradient Self-Attention Maps},
booktitle = {30th ACM International Conference on Information & Knowledge Management},
year = {2021},
month = {October},
abstract = {Transformer-based language models significantly advanced the state-of-the-art in many linguistic tasks. As this revolution continues, the ability to explain model predictions has become a major area of interest for the NLP community. In this work, we present Gradient Self-Attention Maps (Grad-SAM) - a novel gradient-based method that analyzes self-attention units and identifies the input elements that explain the model's prediction the best. Extensive evaluations on various benchmarks show that Grad-SAM obtains significant improvements over state-of-the-art alternatives.},
url = {http://approjects.co.za/?big=en-us/research/publication/grad-sam-explaining-transformers-via-gradient-self-attention-maps/},
}