1 paper
Harikrishna Narasimhan, Wittawat Jitkrittum, Ankit Singh Rawat +4
Cascades and speculative decoding are two common approaches to improving language models' inference efficiency. Both approaches involve interleaving models of different sizes, but…