metadata
license: mit
task_categories:
- video-classification
- multiple-choice
tags:
- video
- video_understanding
- NLP
- computational_social_science
To use this dataset, please cite our paper "VIBE: Can a VLM Read the Room?": https://aclanthology.org/2025.findings-emnlp.1252/
To access the data, please go to "Files and versions", then download and extract vibe.zip. The videos for the task can be found in the vibe/videos folder. The associated vibe_task.jsonl file has the task instances.