Large language model (LLM) routing reduces serving cost by assigning each query to an appropriate model while preserving response quality. Learning such a ro…
机构:南大
来源:arXiv 2609.37402 | AI4Papers 论文推荐平台