Aran Komatsuzaki   @arankomatsuzaki   6/10/2021       

CoAtNet: Marrying Convolution and Attention for All Data Sizes Achieves SotA perf under different resource constraints across various datasets. Achieves 88.56% top-1 ImageNe accuracy, matching ViT-huge while using 23x less data. https://t.co/EeOrxOpgx1

 Reply  0     Retweet   27      Like   131





Posted by Aran Komatsuzaki