Narrative generation is an open-ended NLP task in which a model generates a story given a prompt. The task is similar to neural response generation for chatbots; however, innovations in response generation are often not applied to narrative generation, despite the similarity between these tasks. We aim to bridge this gap by applying and evaluating advances in decoding methods for neural response generation to neural narrative generation. In particular, we employ GPT-2 and perform ablations across nucleus sampling thresholds and diverse decoding hyperparameters-specifically, maximum mutual information-analyzing results over multiple criteria with automatic and human evaluation. We find that (1) nucleus sampling is generally best with thresholds between 0.7 and 0.9; (2) a maximum mutual information objective can improve the quality of generated stories; and (3) established automatic metrics do not correlate well with human judgments of narrative quality on any qualitative metric.
Service learning is a community engagement pedagogy often used in the context of the undergraduate classroom to synergize course-learning objectives with community needs. We find that an effective way to catalyze student engagement in service learning is for student participation to occur outside the context of a graded course, driven by students’ own interests and initiative. In this paper, we describe the creation and implementation of a self-driven service learning program and discuss its benefits from the community, student, and faculty points of view. This experience allows students to explore careers in the sciences as well as identify skill strengths and weaknesses in an environment where mentoring is available but where student initiative and self-motivation are the driving forces behind the project’s success. Self-driven service learning introduces young scientists to the idea that their careers serve a larger community that benefits not only from their discoveries but also from effective communication about how these discoveries are relevant to everyday life.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.