MEGA-Bench: Scaling Multimodal Evaluation to over 500 Real-World Tasks
This addresses the need for scalable and diverse evaluation of multimodal models for end users, though it is incremental as it builds on existing benchmarks by expanding task variety and metrics.
The authors tackled the problem of evaluating multimodal AI models by creating MEGA-Bench, a suite with over 500 real-world tasks and 8,000 samples, which provides fine-grained capability reports across multiple dimensions.
We present MEGA-Bench, an evaluation suite that scales multimodal evaluation to over 500 real-world tasks, to address the highly heterogeneous daily use cases of end users. Our objective is to optimize for a set of high-quality data samples that cover a highly diverse and rich set of multimodal tasks, while enabling cost-effective and accurate model evaluation. In particular, we collected 505 realistic tasks encompassing over 8,000 samples from 16 expert annotators to extensively cover the multimodal task space. Instead of unifying these problems into standard multi-choice questions (like MMMU, MMBench, and MMT-Bench), we embrace a wide range of output formats like numbers, phrases, code, \LaTeX, coordinates, JSON, free-form, etc. To accommodate these formats, we developed over 40 metrics to evaluate these tasks. Unlike existing benchmarks, MEGA-Bench offers a fine-grained capability report across multiple dimensions (e.g., application, input type, output format, skill), allowing users to interact with and visualize model capabilities in depth. We evaluate a wide variety of frontier vision-language models on MEGA-Bench to understand their capabilities across these dimensions.