Explore the Falconer framework, which leverages LLMs as planners to train lightweight proxies, achieving a 90% reduction in inference costs and 20x speedup f...
Level: advanced
By Unknown
Category: research