Poor-Supervised Evaluation for SuperLLM via Mutual Consistency
This addresses the problem of evaluating LLMs for researchers and practitioners when human labels are limited, offering a novel evaluation paradigm that is incremental but practical.
The paper tackles the challenge of evaluating large language models (LLMs) on hard tasks where accurate labels are scarce, proposing the PoEM framework that assesses model capability via consistency with reference models. Experiments across 16 LLMs show PoEM achieves a 0.98 Pearson correlation with supervised evaluation results under poor supervision.
The guidance from capability evaluations has greatly propelled the progress of both human society and Artificial Intelligence. However, as LLMs evolve, it becomes challenging to construct evaluation benchmarks for them with accurate labels on hard tasks that approach the boundaries of human capabilities. To credibly conduct evaluation without accurate labels (denoted as poor-supervised evaluation), we propose the PoEM framework. We first prove that the capability of a model can be equivalently assessed by the consistency between it and certain reference model, when their prediction distributions are independent and the sample size is infinite. To alleviate the insufficiencies of the conditions in reality, we further introduce an algorithm that treats humans (when available) and the models under evaluation as reference models, alternately conducting model weights calibration and filtering during E-step and M-step. Comprehensive experiments across 3 types of tasks with 16 mainstream LLMs have shown that PoEM under poor supervision can achieve an average of 0.98 Pearson correlation coefficient with supervised evaluation results, demonstrating good effectiveness, efficiency and generalizability. More generally, PoEM has advanced the evaluation paradigm evolution from human-centric to human&model-centric by treating both of them as reference models, mitigating the limitations of human evaluation in the era of LLMs.