Neural Information Processing Systems http://nips.cc/

Our work makes one of the first attempts at highlighting the potential of post hoc explanations as valuable tools for enhancing the effectiveness of LLMs.

In this work, we analyze stability as a property for reliable feature attribution methods.

We qualitatively and quantitatively analyze FFAM with multiple detectors on several datasets. Experimental results validate the high-quality visual explanations produced by FFAM.

Equally contribute to this research.

Through extensive experiments we show that some existing saliency methods are independent both of the model and of the data generating process.