1 paper · 1 filter
Weishun Zhong, Doron Sivan, Tankut Can +2
Humans and large language models can predict next letter or word from its prior context much better than random guessing, indicating strong redundancy of language viewed as a stoch…