CreBench: Human-Aligned Creativity Evaluation from Idea to Process to Product

TL;DR

CreBench is a comprehensive benchmark and dataset designed to evaluate and enhance multimodal large language models' ability to understand and assess human-aligned creativity across ideas, processes, and products.

Contribution

The paper introduces CreBench and CreMIT, the first benchmark and dataset for multimodal creativity evaluation, and demonstrates how fine-tuning models on this data improves alignment with human creativity judgments.

Findings

CreExpert outperforms GPT-4V and Gemini-Pro-Vision in creativity evaluation.

The dataset includes 2.2K multimodal samples with extensive human feedback.

Fine-tuning on CreBench enhances models' creativity assessment capabilities.

Abstract

Human-defined creativity is highly abstract, posing a challenge for multimodal large language models (MLLMs) to comprehend and assess creativity that aligns with human judgments. The absence of an existing benchmark further exacerbates this dilemma. To this end, we propose CreBench, which consists of two key components: 1) an evaluation benchmark covering the multiple dimensions from creative idea to process to products; 2) CreMIT (Creativity Multimodal Instruction Tuning dataset), a multimodal creativity evaluation dataset, consisting of 2.2K diverse-sourced multimodal data, 79.2K human feedbacks and 4.7M multi-typed instructions. Specifically, to ensure MLLMs can handle diverse creativity-related queries, we prompt GPT to refine these human feedbacks to activate stronger creativity assessment capabilities. CreBench serves as a foundation for building MLLMs that understand human-aligned creativity. Based on the CreBench, we fine-tune open-source general MLLMs, resulting in CreExpert, a multimodal creativity evaluation expert model. Extensive experiments demonstrate that the proposed CreExpert models achieve significantly better alignment with human creativity evaluation compared to state-of-the-art MLLMs, including the most advanced GPT-4V and Gemini-Pro-Vision.