TX-Ray: Quantifying and Explaining Model-Knowledge Transfer in (Un-)Supervised NLP
This work addresses the need for better explainability in NLP model training, particularly for understanding knowledge transfer, which is incremental as it adapts an existing computer vision principle to NLP.
The authors tackled the problem of explaining and quantifying knowledge transfer in NLP models during unsupervised or supervised training by developing TX-Ray, a method that visualizes neuron feature preferences to track and measure knowledge abstraction and transfer. They found that TX-Ray-based pruning can improve test set generalization and reveal how self-supervision learns linguistic abstractions like parts-of-speech.
While state-of-the-art NLP explainability (XAI) methods focus on explaining per-sample decisions in supervised end or probing tasks, this is insufficient to explain and quantify model knowledge transfer during (un-)supervised training. Thus, for TX-Ray, we modify the established computer vision explainability principle of 'visualizing preferred inputs of neurons' to make it usable transfer analysis and NLP. This allows one to analyze, track and quantify how self- or supervised NLP models first build knowledge abstractions in pretraining (1), and then transfer these abstractions to a new domain (2), or adapt them during supervised fine-tuning (3). TX-Ray expresses neurons as feature preference distributions to quantify fine-grained knowledge transfer or adaptation and guide human analysis. We find that, similar to Lottery Ticket based pruning, TX-Ray based pruning can improve test set generalization and that it can reveal how early stages of self-supervision automatically learn linguistic abstractions like parts-of-speech.