分享

Recurrent Drafter for Fast Speculative Decoding in Large Language Models

28
查看论文
热度