GlobEnc: Quantifying Global Token Attribution by Incorporating the Whole Encoder Layer in Transformers
This work addresses the need for more faithful explanations in Transformer models, which is important for researchers and practitioners in interpretable AI, though it is incremental as it builds on existing attribution methods.
The paper tackled the problem of interpreting Transformers by introducing a token attribution analysis method that incorporates all encoder components and aggregates them across layers, demonstrating that it produces more accurate global token attributions and significantly outperforms previous methods in correlation with gradient-based saliency scores.
There has been a growing interest in interpreting the underlying dynamics of Transformers. While self-attention patterns were initially deemed as the primary option, recent studies have shown that integrating other components can yield more accurate explanations. This paper introduces a novel token attribution analysis method that incorporates all the components in the encoder block and aggregates this throughout layers. Through extensive quantitative and qualitative experiments, we demonstrate that our method can produce faithful and meaningful global token attributions. Our experiments reveal that incorporating almost every encoder component results in increasingly more accurate analysis in both local (single layer) and global (the whole model) settings. Our global attribution analysis significantly outperforms previous methods on various tasks regarding correlation with gradient-based saliency scores. Our code is freely available at https://github.com/mohsenfayyaz/GlobEnc.