Towards Acyclic Preference Evaluation of Language Models via Multiple Evaluators

Despite the remarkable success of Large Language Models (LLMs), evaluating their outputs' quality regarding preference remains a critical challenge. While existing works usually leverage a strong LLM as the judge for comparing LLMs' response pairwisely, such a single-evaluator approach is vulnerable to cyclic preference, i.e., output A is better than B, B than C, but C is better than A, causing contradictory evaluation results. To address this, we introduce PGED (Preference Graph Ensemble and Denoising), a novel approach that leverages multiple model-based evaluators to construct preference graphs, and then ensembles and denoises these graphs for acyclic, non-contradictory evaluation results. We provide theoretical guarantees for our framework, demonstrating its efficacy in recovering the ground truth preference structure. Extensive experiments on ten benchmarks demonstrate PGED's superiority in three applications: 1) model ranking for evaluation, 2) response selection for test-time scaling, and 3) data selection for model fine-tuning. Notably, PGED combines small LLM evaluators (e.g., Llama3-8B, Mistral-7B, Qwen2-7B) to outperform strong ones (e.g., Qwen2-72B), showcasing its effectiveness in enhancing evaluation reliability and improving model performance.