Fetching the paper…

Recurrent Drafter for Fast Speculative Decoding in Large Language Models · Around