Hacker News new | ask | show | jobs
by echelon 3057 days ago
Seems like you could add two additional training parameters to the model to compensate.