Speculative cascades – A hybrid approach for smarter, faster LLM inference

5 pointsposted 13 hours ago
by emschwartz
(research.google)

No comments yet