Incorrect student answers can become valuable learning opportunities, provided that the student understands where they went wrong and why. To this end, rather than being given the correct answer, students should receive elaborated feedback on how to correct a mistake on their own. Highlighting the complex demands that the generation of such feedback places on a model’s input utilization abilities, we propose two extensions to the training pipeline. Firstly, we employ a KL regularization term between a standard and enriched input format to achieve more targeted input representations. Secondly, we add a preference optimization step to encourage student answer-adaptive feedback generation. The effectiveness of those extensions is underlined by a significant increase in model performance of 3.3 METEOR points. We go beyond traditional surface form-based metrics to assess two important dimensions of feedback quality, i.e., faithfulness and informativeness. Hereby, we are the first to propose an automatic metric measuring the degree to which feedback divulges the correct answer, that we call Informativeness Index I². We verify in how far each metric captures feedback quality.

The paradigm of leveraging large pre-trained language models has made significant progress on benchmarks on task-oriented dialogue (TOD) systems. In this paper, we combine this paradigm with multi-task learning framework for end-to-end TOD modeling by adopting span prediction as an auxiliary task. In end-to-end setting, our model achieves new state-of-the-art results with combined scores of 108.3 and 107.5 on MultiWOZ 2.0 and MultiWOZ 2.1, respectively. Furthermore, we demonstrate that multi-task learning improves not only the performance of model but its generalization capability through domain adaptation experiments in the few-shot setting. The code is available at github.com/bepoetree/MTTOD.