Among-us-X/

AmongUs-X

Dataset DOI: https://doi.org/10.57967/hf/8698 Companion code: https://github.com/among-us-X/Among-Us-X Croissant: machine-readable croissant.json (Croissant 1.1, with the eight NeurIPS RAI fields populated)

AmongUs-X is a corpus of 8,720 LLM-vs-LLM games of Among Us with full per-turn agent transcripts, voting beliefs, and game ground truth. It is released as the data companion to the paper "Among Us as a… See the full description on the dataset page: https://huggingface.co/datasets/Among-us-X/AmongUs-X.