MILE-RefHumEval: A Reference-Free, Multi-Independent LLM Framework for Human-Aligned Evaluation

TL;DR

MILE-RefHumEval is a novel reference-free framework that uses independent evaluators guided by human-aligned schemas to assess LLMs, achieving high alignment with human judgments and improved efficiency.

Contribution

It introduces a flexible, scalable, and human-aligned evaluation framework for LLMs that does not rely on ground-truth data or evaluator coordination.

Findings

Aligns closely with human judgments

Outperforms prior evaluation methods

Reduces computational overhead

Abstract

We introduce MILE-RefHumEval, a reference-free framework for evaluating Large Language Models (LLMs) without ground-truth annotations or evaluator coordination. It leverages an ensemble of independently prompted evaluators guided by a human-aligned schema, supporting both discrete and continuous scoring judgement. With task-specific prompts from best candidate selection, summarization and image captioning to dialogue, MILE-RefHumEval provides flexible, interpretable, and scalable assessments. Experiments show it aligns closely with human judgments, outperforms prior methods, and reduces computational overhead, offering an efficient, robust, and human-aligned solution for real-world LLM evaluation.