DecompX: Explaining Transformers Decisions by Propagating Token Decomposition
This work addresses the problem of improving interpretability for Transformer-based models, which is crucial for researchers and practitioners in AI, though it appears incremental as it builds on existing vector-based analysis methods.
The paper tackles the challenge of providing faithful vector-based explanations for Transformer models by introducing DecompX, which propagates decomposed token representations without mixing between layers and includes all encoder components and the classification head, resulting in consistent outperformance of existing gradient-based and vector-based approaches on various datasets according to standard faithfulness evaluations.
An emerging solution for explaining Transformer-based models is to use vector-based analysis on how the representations are formed. However, providing a faithful vector-based explanation for a multi-layer model could be challenging in three aspects: (1) Incorporating all components into the analysis, (2) Aggregating the layer dynamics to determine the information flow and mixture throughout the entire model, and (3) Identifying the connection between the vector-based analysis and the model's predictions. In this paper, we present DecompX to tackle these challenges. DecompX is based on the construction of decomposed token representations and their successive propagation throughout the model without mixing them in between layers. Additionally, our proposal provides multiple advantages over existing solutions for its inclusion of all encoder components (especially nonlinear feed-forward networks) and the classification head. The former allows acquiring precise vectors while the latter transforms the decomposition into meaningful prediction-based values, eliminating the need for norm- or summation-based vector aggregation. According to the standard faithfulness evaluations, DecompX consistently outperforms existing gradient-based and vector-based approaches on various datasets. Our code is available at https://github.com/mohsenfayyaz/DecompX.