How are machines able to answer human questions?

Explore Neural Question Answering (QA) with this article. QA is a system's ability to accurately answer questions using information from provided sources. Learn how such systems are implemented, how QA datasets are created and collected, and take a look at the results of modeling experiments using these datasets. Download

Do machines learn more from images when humans describe it?

Learn how contrastive learning works in this reproducibility paper, where we replicate portions of ConVIRT. ConVIRT boosts machines’ ability to “understand” images while reducing costs by combining image datasets with pre-existing text. Find out about the datasets, the methods of how we reproduced this research, and our results. Download