A new technique enables users to compare several large models and choose the one that works best for their task. Publication Date: July 16, 2024 Press Inquiries 2 dog photos on left, plus 2 spheres. Arrows from the dog photos point to areas on the spheres. Text says, “Dog?” and then “Dog!” Caption: To estimate the reliability of massive deep-learning models called foundation models, MIT researchers developed a technique to assess the consistency of representations an ensemble of similar models learn about the same test data point. Credits: MIT News; Courtesy of the researchers Foundation models are massive deep-learning models that have been pretrained on an enormous amount of general-purpose, unlabeled data. They can be applied to a variety of tasks, like generating images or answering customer questions. But […]
Original web page at news.mit.edu
