We propose a project that uses the T5 and GPT-3 models (both pre-trained, and tuned) to explore auto-generating abstracts using the National Science Foundation's dataset of award winning Research Grants. We hypothesize that using this dataset will generate abstracts that would also score highly and increase the probability of one's Grant being accepted by the NSF.
ejdunk / w266_final_project Goto Github PK
View Code? Open in Web Editor NEWW266_Final_Project