MixEval: Deriving Wisdom of the Crowd from LLM Benchmark Mixtures

TL;DR

MixEval introduces a novel benchmark mixing approach that combines real-world user queries with existing benchmarks, providing a reliable, efficient, and dynamic evaluation method for large language models that correlates well with user-based assessments.

Contribution

The paper presents MixEval, a new paradigm for LLM evaluation that strategically combines benchmark data with web-mined queries, improving reliability, efficiency, and dynamic updating capabilities.

Findings

Achieves 0.96 correlation with Chatbot Arena rankings.

Runs 94% faster and cheaper than MMLU.

Enables dynamic and reproducible evaluation updates.

Abstract

Evaluating large language models (LLMs) is challenging. Traditional ground-truth-based benchmarks fail to capture the comprehensiveness and nuance of real-world queries, while LLM-as-judge benchmarks suffer from grading biases and limited query quantity. Both of them may also become contaminated over time. User-facing evaluation, such as Chatbot Arena, provides reliable signals but is costly and slow. In this work, we propose MixEval, a new paradigm for establishing efficient, gold-standard LLM evaluation by strategically mixing off-the-shelf benchmarks. It bridges (1) comprehensive and well-distributed real-world user queries and (2) efficient and fairly-graded ground-truth-based benchmarks, by matching queries mined from the web with similar queries from existing benchmarks. Based on MixEval, we further build MixEval-Hard, which offers more room for model improvement. Our benchmarks' advantages lie in (1) a 0.96 model ranking correlation with Chatbot Arena arising from the highly impartial query distribution and grading mechanism, (2) fast, cheap, and reproducible execution (6% of the time and cost of MMLU), and (3) dynamic evaluation enabled by the rapid and stable data update pipeline. We provide extensive meta-evaluation and analysis for our and existing LLM benchmarks to deepen the community's understanding of LLM evaluation and guide future research directions.