This dataset provides the complete data needed to run and evaluate models on 3MDBench (Medical Multimodal Multi-agent Dialogue Benchmark) — a large-scale, multimodal benchmark designed for assessing AI-driven medical dialogue systems. It simulates realistic telemedicine consultations between a Doctor Agent and a temperament-driven Patient Agent using image and text inputs. An Assessor Agent, aligned with human… See the full description on the dataset page:
https://huggingface.co/datasets/univanxx/3mdbench.
