The model is trained end-to-end to answer common-sense-reasoning questions after generating explanations for its answers, using sample human explanations as part of the training data.
This results in improved performance on the question-answering task.
This is fascinating, although in hindsight, not entirely surprising: inducing a machine to learn to model human explanations helps the machine perform better in testing.
A natural question follows:
Can we find ways to induce much larger models to learn to generate human explanations about a growing number of subjects of increasing complexity?
> Social Stories are a concept devised by Carol Gray in 1991 to improve the social skills of people with autism spectrum disorders (ASD). The objective is to share information, which is often through a description of the events occurring around the subject and also why.
I'm wondering if this kind of "common sense" interactions could be leveraged to train models?
This results in improved performance on the question-answering task.
This is fascinating, although in hindsight, not entirely surprising: inducing a machine to learn to model human explanations helps the machine perform better in testing.
A natural question follows:
Can we find ways to induce much larger models to learn to generate human explanations about a growing number of subjects of increasing complexity?