Twitter Debate Leads to New Academic Discovery! Xie Saining's Team Releases iREPA with Just 3 Lines of Code!

Xie Saining's team accidentally created a new paper, iREPA, as a result of a four-month Twitter debate. The debate focused on whether self-supervised learning models should focus on dense tasks that rely on spatial local information. It ended with Xie Saining yielding, but it led to this important paper showcasing innovative ideas.