AutoJudger: An Agent-Driven Framework for Efficient Benchmarking of MLLMs

TL;DR

AutoJudger is an adaptive, agent-driven framework that significantly reduces the cost of benchmarking multimodal large language models by selectively choosing questions based on difficulty and performance, achieving high accuracy with minimal data.

Contribution

This work introduces AutoJudger, a novel framework combining IRT and autonomous agents for efficient, adaptive benchmarking of MLLMs, reducing evaluation costs substantially.

Findings

AutoJudger uses only 4% of data for 90% accuracy.

It effectively covers diverse and challenging scenarios.

Reduces evaluation expenses dramatically.

Abstract

Evaluating multimodal large language models (MLLMs) is increasingly expensive, as the growing size and cross-modality complexity of benchmarks demand significant scoring efforts. To tackle with this difficulty, we introduce AutoJudger, an agent-driven framework for efficient and adaptive benchmarking of MLLMs that tackles this escalating cost. AutoJudger employs the Item Response Theory (IRT) to estimate the question difficulty and an autonomous evaluation agent to dynamically select the most informative test questions based on the model's real-time performance. Specifically, AutoJudger incorporates two pivotal components: a semantic-aware retrieval mechanism to ensure that selected questions cover diverse and challenging scenarios across both vision and language modalities, and a dynamic memory that maintains contextual statistics of previously evaluated questions to guide coherent and globally informed question selection throughout the evaluation process. Extensive experiments on four representative multimodal benchmarks demonstrate that our adaptive framework dramatically reduces evaluation expenses, i.e. AutoJudger uses only 4% of the data to achieve over 90% ranking accuracy with the full benchmark evaluation on MMT-Bench.