CounterEval: Towards Unifying Evaluation of Counterfactual Explanations
Overview
CounterEval offers a human-evaluated dataset of 30 counterfactual scenarios, engineered to serve as a benchmark for evaluating a wide range of counterfactual explanation generation frameworks. Each scenario has been strategically designed to exhibit varying levels across multiple explanatory quality metrics, such as Feasibility, Consistency, Trust, Completeness, Fairness, and Complexity… See the full description on the dataset page: https://huggingface.co/datasets/anitera/CounterEval.
