Almieyar-Oryx-BloomBench: bilingual multimodal benchmark for VLM evaluation

Arxiv CS.CV (computer vision)

The benchmark is designed for cognitively informed evaluation of vision-language models (VLMs) in English and Arabic. It argues current benchmarks lack diagnostic rigor for reasoning abilities.

vision-language-modelsbenchmarks

Arxiv CS.CV (computer vision)·Mohammad Mahdi AbootorabiOmid GhahroodiAnas MadkoorMarzia NouriDoratossadat DastgheibMohamed HefeedaEhsaneddin Asgari

6 days ago