We compare Transformer- and RNN-based language models' ability to account for measures of human reading effort. Our analysis shows Transformers to ...
確定! 回上一頁