datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
RoomReader-Full
RoomReader-Full
RoomReader-Full is a multimodal benchmark for fine-grained understanding of how a
speaker manages information during a conversation. Given a short target video clip
and its audio/dialogue, a model must determine whether the response is consistent
with the relevant interactional expectation and assign one of six information
strategy labels.
The benchmark is designed for cases where the observable response alone is not
enough to determine the label. Formal gold… See the full description on the dataset page: https://huggingface.co/datasets/xilinghuiye/RoomReader-Full.roomnumber