Mixture of Complementary Agents for Robust LLM Ensemble

In three linesStudy on optimal model selection in LLM ensembles. Authors reframe proposer selection as a combinatorial problem based on complementarity rather than accuracy alone. Greedy algorithms tested on small labeled sets to balance performance and computational cost.

Summary generated by Claude — human-verified