Meta published OpenEQA, the Open-Vocabulary Embodied Question Answering Benchmark.

The new benchmark measures the vision language model’s understanding of physical environments by asking it questions. pic.twitter.com/yFw3O2PeXP

— Rowan Cheung (@rowancheung) 12 avril 2024

Meta published OpenEQA, the Open-Vocabulary Embodied Question Answering Benchmark. The new benchmark measures the vision language model’s understanding of physical environments by asking it questions.

→ View original post on X — @rowancheung