1 paper
Risto Luukkonen, Jonathan Burdge, Elaine Zosa +5
The pretraining of state-of-the-art large language models now requires trillions of words of text, which is orders of magnitude more than available for the vast majority of languag…