Обучение на основе учебного плана

AbstractVisual question answering (VQA) models, in particular modular ones, are commonly trained on large-scale datasets to achieve state-of-the-art performance. However, such datasets are sometimes not available. Further, it has been …...