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

6 pointsposted 5 months ago
by emschwartz
(research.google)

No comments yet