Join RISELab’s Joseph Gonzalez on the TWIML (This Week in Machine Learning) Podcast for a discussion of his paper “Train Large, Then Compress: Rethinking Model Size for Efficient Training and Inference of Transformers”.