From a Large-Scale Teacher to Efficient Multimodal Students for Search Relevance Prediction
This work proposes a framework that bridges the gap in large language models at scale by distilling a high-performing LLM-based relevance judgment system into efficient multimodal student models, and achieves performance close to the teacher system while significantly improving latency, cost, and scalability.
Tomasz Gniazdowski, Mohammadreza Momenifar, Nadia Vase et al.
· 0 citations