Under the premise that surprisal theory holds, we would expect that higher quality language models provide more powerful predictors of human reading behavior—a ...

Dec 10, 2023 · Under the premise that surprisal theory holds, we would expect that higher quality language models, whose predictions are more accurate, provide ...

We find correlations between LM quality and power in eleven of these thirteen languages, suggesting that, within the range of model classes and sizes tested, ...

Dec 10, 2023 · Some studies in English have found correlations between LM quality and psychometric predictive power, but other studies using Japanese data, as ...

Our results align with the studies that have found language model quality to positively correlate with the model's psychometric predictive power (22) . ...

This repo contains code and data associated with "Language Model Quality Correlates with Psychometric Predictive Power in Multiple Languages", ...

Dec 7, 2023 · Language Model Quality Correlates with Psychometric Predictive Power in Multiple Languages ... Multilingual Translation Models with Fisher ...

Sep 10, 2024 · Deriving estimates from language models trained on monolin- gual and multilingual corpora, we test three predictions associated with surprisal ...

Language Model Quality Correlates with Psychometric Predictive Power in Multiple Languages · On the Optimality of Word Lengths · Revisiting the Optimality of Word ...

We quantify the psychometric predictive power of surprisal and contextual entropy (i.e., how well each predicts reading times) by including them as variables in ...