Beyond Benchmark: LLMs Evaluation with an Anthropomorphic and Value-oriented Roadmap
This work addresses the problem of fragmented and incomplete evaluation frameworks for LLMs, offering a holistic approach for researchers and developers, though it is incremental as it builds on existing survey and taxonomy methods.
The paper tackles the disconnect between benchmark performance and real-world utility in Large Language Models (LLMs) by introducing an anthropomorphic evaluation paradigm with a three-dimensional taxonomy (IQ, EQ, PQ) and a Value-oriented Evaluation (VQ) framework, analyzing over 200 benchmarks to provide actionable guidance for developing technically proficient, contextually relevant, and ethically sound LLMs.
For Large Language Models (LLMs), a disconnect persists between benchmark performance and real-world utility. Current evaluation frameworks remain fragmented, prioritizing technical metrics while neglecting holistic assessment for deployment. This survey introduces an anthropomorphic evaluation paradigm through the lens of human intelligence, proposing a novel three-dimensional taxonomy: Intelligence Quotient (IQ)-General Intelligence for foundational capacity, Emotional Quotient (EQ)-Alignment Ability for value-based interactions, and Professional Quotient (PQ)-Professional Expertise for specialized proficiency. For practical value, we pioneer a Value-oriented Evaluation (VQ) framework assessing economic viability, social impact, ethical alignment, and environmental sustainability. Our modular architecture integrates six components with an implementation roadmap. Through analysis of 200+ benchmarks, we identify key challenges including dynamic assessment needs and interpretability gaps. It provides actionable guidance for developing LLMs that are technically proficient, contextually relevant, and ethically sound. We maintain a curated repository of open-source evaluation resources at: https://github.com/onejune2018/Awesome-LLM-Eval.