Benchmark RadarAI BENCHMARK PROFILE
MetaphorVU-Bench
MetaphorVU-Bench evaluates metaphorical video understanding in multimodal LLMs through tasks requiring cross-domain mapping, with a benchmark dataset and evaluation code publicly available.
- Released
- 2026-05-25
- Readiness
- Runnable
- Primary field
- General AI
Why it matters
It fills the gap of systematic evaluation of high-order cognitive capabilities in video understanding, revealing defects in cross-domain mapping and enabling future improvements.
Motivation
Metaphorical videos are prevalent across various real-world scenarios to convey complex ideas, and understanding them typically requires high-order cognitive capabilities.
Primary resources
Benchmark Radar records only publicly supported details and links back to primary sources for verification.