Learn to Explain: Multimodal Reasoning via Thought Chains for Science Question Answering

TL;DR

This paper introduces ScienceQA, a large multimodal science question benchmark with explanations, and demonstrates that chain-of-thought reasoning improves language model performance on science questions, especially with explanations.

Contribution

The paper presents ScienceQA, a new multimodal science question benchmark with annotations and explanations, and shows how chain-of-thought reasoning enhances model performance.

Findings

CoT improves GPT-3 few-shot accuracy by 1.20%.

Feeding explanations boosts GPT-3 performance by 18.96%.

Models benefit from explanations, achieving similar results with less data.

Abstract

When answering a question, humans utilize the information available across different modalities to synthesize a consistent and complete chain of thought (CoT). This process is normally a black box in the case of deep learning models like large-scale language models. Recently, science question benchmarks have been used to diagnose the multi-hop reasoning ability and interpretability of an AI system. However, existing datasets fail to provide annotations for the answers, or are restricted to the textual-only modality, small scales, and limited domain diversity. To this end, we present Science Question Answering (ScienceQA), a new benchmark that consists of ~21k multimodal multiple choice questions with a diverse set of science topics and annotations of their answers with corresponding lectures and explanations. We further design language models to learn to generate lectures and explanations as the chain of thought (CoT) to mimic the multi-hop reasoning process when answering ScienceQA questions. ScienceQA demonstrates the utility of CoT in language models, as CoT improves the question answering performance by 1.20% in few-shot GPT-3 and 3.99% in fine-tuned UnifiedQA. We also explore the upper bound for models to leverage explanations by feeding those in the input; we observe that it improves the few-shot performance of GPT-3 by 18.96%. Our analysis further shows that language models, similar to humans, benefit from explanations to learn from fewer data and achieve the same performance with just 40% of the data. The data and code are available at https://scienceqa.github.io.