Current agent benchmarks usually assume the clearance of given tasks and exclude user intention understanding as an important aspect for evaluation. Given this ignorance in assessment, we formulate Intention-in-Interaction… See the full description on the dataset page:
https://huggingface.co/datasets/hbx/IN3.