Language Model Preference Evaluation with Multiple Weak Evaluators

원문 초록

Despite the remarkable success of Large Language Models (LLMs), evaluating their outputs' quality regarding *preference* remains a critical challenge. Existing works usually leverage an LLM as the judge for comparing LLMs' output pairwisely, yet such model-based evaluator is *weak evaluator* due to *conflicting preference*, i.e., output A is better than B, B than C, but C than A, causing contradictory evaluation results. To address this, we introduce GED (Preference Graph Ensemble and Denoise), a novel approach that leverages multiple model-based evaluators to construct preference graphs, and then ensemble and denoise these graphs for better, non-contradictory evaluation results. In particular, our method consists of two primary stages: aggregating evaluations into a unified graph and applying a denoising process to eliminate cyclic inconsistencies, ensuring a directed acyclic graph (DAG) structure. We provide theoretical guarantees for our framework, demonstrating its efficacy in recovering the ground truth preference structure. Extensive experiments on ten benchmarks demonstrate GED's superiority in three applications: model ranking, response selection, and model alignment tasks. Notably, GED combines small LLM evaluators (e.g., Llama3-8B, Mistral-7B, Qwen2-7B) to outperform stronger ones (e.g., Qwen2-72B), showcasing its effectiveness in enhancing evaluation reliability and improving model performance.