Mmmu: Massive Multi-Discipline Multimodal Understanding And Reasoning Benchmark
Massive Multi-discipline Multimodal Understanding and Reasoning Benchmark
Areas of application
- Education
- Research
- Industry
- Healthcare
- Finance
- Marketing
- Government
Example
The MMMU benchmark is a new evaluation tool for multimodal models that can assess their ability to understand and reason across multiple disciplines, such as Art & Design, Business, Health & Medicine, Science, Humanities & Social Science, and Technology & Engineering. The benchmark includes over 183 subfields and covers a variety of image formats, such as diagrams, tables, charts, chemical structures, photographs, paintings, geometric shapes, and musical scores.