Digital Agent outperforms o1 by 15% – trained with new RL-variant similar to R1 arxiv.org 11 points by let_tim_cook_ a year ago · 0 comments Reader PiP Save No comments yet.