Attention Visualizer Package: Revealing Word Importance for Deeper Insight into Encoder-Only Transformer Models
This is an incremental contribution that addresses the need for better interpretability and explainability in neural networks, particularly for researchers and practitioners using transformer models.
The authors tackled the problem of interpreting encoder-only transformer models by developing the Attention Visualizer package, which visually illustrates word importance based on their impact on final embeddings, providing a tool for deeper insight into model mechanisms.
This report introduces the Attention Visualizer package, which is crafted to visually illustrate the significance of individual words in encoder-only transformer-based models. In contrast to other methods that center on tokens and self-attention scores, our approach will examine the words and their impact on the final embedding representation. Libraries like this play a crucial role in enhancing the interpretability and explainability of neural networks. They offer the opportunity to illuminate their internal mechanisms, providing a better understanding of how they operate and can be enhanced. You can access the code and review examples on the following GitHub repository: https://github.com/AlaFalaki/AttentionVisualizer.