MLLM-as-a-Judge: Assessing Multimodal LLM-as-a-Judge with Vision-Language Benchmark
This addresses the problem of evaluating MLLMs for researchers and practitioners by providing a benchmark, though it is incremental as it extends the LLM-as-a-Judge concept to multimodal settings.
The paper tackles the challenge of assessing Multimodal Large Language Models (MLLMs) by introducing a new benchmark called MLLM-as-a-Judge to evaluate their judgment abilities across tasks like Scoring Evaluation, Pair Comparison, and Batch Ranking. It finds that MLLMs show human-like discernment in Pair Comparison but significant divergence from human preferences in other tasks, highlighting issues like biases and inconsistencies even in advanced models like GPT-4V.
Multimodal Large Language Models (MLLMs) have gained significant attention recently, showing remarkable potential in artificial general intelligence. However, assessing the utility of MLLMs presents considerable challenges, primarily due to the absence of multimodal benchmarks that align with human preferences. Drawing inspiration from the concept of LLM-as-a-Judge within LLMs, this paper introduces a novel benchmark, termed MLLM-as-a-Judge, to assess the ability of MLLMs in assisting judges across diverse modalities, encompassing three distinct tasks: Scoring Evaluation, Pair Comparison, and Batch Ranking. Our study reveals that, while MLLMs demonstrate remarkable human-like discernment in Pair Comparison, there is a significant divergence from human preferences in Scoring Evaluation and Batch Ranking. Furthermore, a closer examination reveals persistent challenges in the judgment capacities of LLMs, including diverse biases, hallucinatory responses, and inconsistencies in judgment, even in advanced models such as GPT-4V. These findings emphasize the pressing need for enhancements and further research efforts to be undertaken before regarding MLLMs as fully reliable evaluators. In light of this, we advocate for additional efforts dedicated to supporting the continuous development within the domain of MLLM functioning as judges. The code and dataset are publicly available at our project homepage: \url{https://mllm-judge.github.io/}.