English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

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

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