1
Beyond Majority Voting: LLM Aggregation by Leveraging Higher-Order Information
顶会ICML2026新研究，利用高阶信息提升LLM聚合效果，超越传统多数投票方法。
arXiv:2510.01499v2 Announce Type: replace Abstract: With the rapid progress of multi-agent large language model (LLM) reasoning, how to effectively ag…