karpathy's autoresearch has been making rounds as it can optimize the loss through self-learning.

are we in epock 5.5 already?