An OpenAI research team fine-tunes the GPT-3 pretrained language model to enable it to answer long-form questions by searching and navigating a text-based web browsing environment, achieving retrieval and synthesis improvements and reaching human-level long-form question-answering performance.
Comments are closed.