Concept
Rollout group
A rollout group is the set of completions sampled from the policy for a single prompt during on-policy RL training. If you know GRPO (Group Relative Policy Optimization), this is the "group" in the name: for each prompt you draw N responses, score them, and compute each…
The rest of “Rollout group” is a premium feature: every concept in the library gets a precise, practitioner-focused write-up like this one, cross-linked straight from the paper summaries that use it.
Log in to unlock→