Definition. Group-relative advantage [shao2024deepseekmath, Section 4.1.2] [ftip-004D]
✍️source

    When \(S_R>0\), the group-relative advantage of rollout member \(i\) is

    \[ \widehat A_i=\frac {R_i-\bar R}{S_R}. \]

    The estimator compares responses sampled for the same prompt and does not require a learned value function. Its members are statistically coupled through \(\bar R\) and \(S_R\); it is not an independent estimate of the population advantage of each response.