Found 1 document(s)
Page 1 of 1
-
A fundamental challenge for explainable and accountable language models is revealing how external information influences their answers. This paper examines whether models are self-attributive: that is, whether they indic...