@xczhangxx
CS PhD @ Harvard
AI
A post describing RAX reports a 78.3% RoboDojo result, versus 24.9% for the prior state of the art.