Libra-Bench & PIE_bench

dataset
Meituan
Benchmarks for evaluating visual and logical reasoning.
Paper (arXiv)HuggingFace
benchmarkreasoningvision