← all datasets

MMMR

Emerging
1papers using it
2025first seen

The MMMR is a benchmark that contains a high-difficulty dataset of 1,083 questions across six reasoning types and is used to evaluate multi-modal reasoning capabilities of models, focusing on reasoning quality through metrics beyond accuracy.

Papers using MMMR (1)

MMMR dataset β€” papers, benchmarks & downloads Β· Large Language Models