Transformers can optimally learn regression mixture models.

Reese Pathak, Rajat Sen, Weihao Kong, Abhimanyu Das

Browse the full ICLR paper archive.