Definition. Group mean and standard deviation [shao2024deepseekmath, Section 4.1.2] [ftip-004C]
✍️source

    For rollout-group rewards \(R_1,\ldots ,R_g\), define the group mean and group standard deviation by

    \[ \bar R=\frac 1g\sum _{i=1}^g R_i, \qquad S_R=\operatorname {std}(R_1,\ldots ,R_g). \]

    Both statistics are prompt-local and depend on every member of the sampled group. The source leaves the standard-deviation divisor symbolic and does not add a numerical stabilizer. An implementation must state those conventions and its treatment of zero-variance groups.