The BAbI benchmark presents a difficult set of tasks designed to evaluate the skills of AI systems in interpreting commonsense knowledge. It comprises a wide range of cases that require logic about everyday notions. By evaluating how well AI models can resolve these problems, researchers strive to better understand the character of commonsense reas… Read More