NeurIPS 2022 | The First Multi-Modal Science Question Answering (Science QA) Dataset with Detailed Explanations Deep Learning Model Reasoning has a Chain of Thoughts

Learning and completing complex tasks efficiently like humans is one of the long-term goals pursued by artificial intelligence. Human beings can follow a complete chain of thought (CoT) reasoning process in the decision-making process, so as to make reasonable explanations for the given answers.

However, most of the existing machine learning models rely on a large number of input-output sample training to complete specific tasks. These black-box models often directly generate the final answer without revealing the specific reasoning process.

The Science Question Answering task (Science Question Answering) can well diagnose whether the artificial intelligence model has multi-step reasoning ability and interpretability. To answer scientific questions, a model not only needs to understand multimodal content, but also needs to extract external knowledge to arrive at the correct answer. At the same time, a reliable model should also give explanations that reveal its reasoning process. However, most current science question answering datasets lack detailed explanations for answers, or are limited to textual modalities.