Hacker News new | past | comments | ask | show | jobs | submit
Seems like it just overfitting the problems?
allot of the models are just constantly eval maxxing