Aggregate Feedback
A feedback form in which the target data owner returns only statistics, such as held-out set accuracy or average reward, rather than item-level fine-grained information. It carries privacy-preserving significance, and its granularity and frequency constitute a key trade-off between communication cost and optimization effectiveness.
Timeline (last 90 days)
The core idea of the L2RR method is that the target data owner can provide limited aggregate feedback (such as accuracy) from a small held-out target set without exposing target prompts or rollout trajectories.
Aggregate feedback means that the target data owner returns only a statistic (such as accuracy or average reward on the held-out set) rather than fine-grained information for each prompt or each trajectory.
All Facts (2)
The core idea of the L2RR method is that the target data owner can provide limited aggregate feedback (such as accuracy) from a small held-out target set without exposing target prompts or rollout trajectories.
50%UnverifiedAggregate feedback means that the target data owner returns only a statistic (such as accuracy or average reward on the held-out set) rather than fine-grained information for each prompt or each trajectory.
50%