GPT-3 and its Few-shot Learning Capabilities – Talk by Prof. Yoav Goldberg

Title: GPT-3 and its Few-shot Learning Capabilities
Abstract: A recent paper by OpenAI, “Language Models are Few-Shot Learners”, introduced GPT-3, a huge pre-trained language model, and a claim that it acts as a very effective few-shot learner. While I was skeptic at first, I got access to the GPT-3 model, and after playing with it for a while, I am indeed very impressed with its few-shot pattern recognition abilities. In this highly-informal talk, I will describe some of its impressive observed capabilities.
Zoom link: Coming soon.