Read the original at HF Daily Papers
Researchers introduce a formal framework and benchmark for multi-agent vision-and-language navigation, featuring a dataset of 11,724 episodes and a new coordination system.
Carried by: HF Daily Papers. First seen: .