This repository contains the code to compute surprisal values from XGLM model. We used the MECO corpus and the XGLM model family to assess the relationship between the psychological accuracy of a language model (namely, the capability of a surprisal estimate to explain variance in human responses) and its linguistic accuracy (i.e., its ability to accurately predict the next token).
Andrea-de-Varda/surprisal-across-languages
Code to calculate surprisal values from multilingual XGLM models.
Python