NLPositionality keeps track of how the datasets and models are aligned with demographics around the world. It does so by sending instances from existing socially-aligned NLP datasets to the Lab in the Wild Platform and obtaining responses from a diverse crowd. The responses from crowd is then used to measure alignment of the models with specific demographics. Check the following tasks to see their demographic alignment with people. For more information, check our upcoming paper.
Take our Lab in the Wild experiment if you are interested!