Pre-trained Language Models Learn Remarkably Accurate Representations of Numbers.

Marek Kadlck, Michal Stefnik, Timothee Mickus, Josef Kuchar, Michal Spiegel

Browse the full EMNLP paper archive.