HaoxiangYou

Hi, Hope you figure out already. From my understanding, $\text{KL}(\pi_\text{new} \| \pi_\text{old}) = \frac{1}{2} (\pi_\text{new} - \pi_\text{old})^T A (\pi_\text{new} - \pi_\text{old}) + o(\|\pi_\text{new} - \pi_\text{old}\|)$. The initial step size $\beta$...