(GRPO) wastes rolls on prompts whose responses are all correct or all wrong. These zero-variance groups produce no gradient yet consume the dominant cost per step. ARCUS transforms pass rates to Fisher-Rao arc length, where the expected update is uniform, zero-variance probability has a simple bound, and objective gradients become Gaussians. It tracks each prompt with a Kalman filter in this coordinate, scores prompts by an objective-matched Gaussian kernel, and keeps only informative groups for the unchanged GRPO update. Across six math reasoning benchmarks and three backbones, ARCUS improves average accuracy by 2.8-2.9 points over standard GRPO and 1.1-1.2 points over dynamic sampling, while generating 48-57% fewer rolls than dynamic sampling.
Fisher-Rao prompt curricula make GRPO training more rollout-efficient
ARCUS selects prompts in a statistically natural pass-rate coordinate, improving mathematical reasoning accuracy while using substantially fewer model rollouts.
Top University
Mei Okonkwo · Pixel Nomand · Julian Berg · Elena Voss · Lena Park · Marcus Hale · +2 more
University of Wisconsin–Madison · University of Washington
Research Digest··2 min read
In one line
ARCUS reduces GRPO rollouts by 48-57% and improves accuracy by 2.8-2.9 points using Fisher-Rao arc length for prompt selection.
Before you read this
Two ideas this paper builds ongroup relative policy optimizationA reinforcement learning method that optimizes policy by computing advantages based on group-level reward statistics.188 papers →adaptive curriculumA training strategy that gradually increases task difficulty based on the learner's current performance, often using success rates to select the next challenge.26 papers →
Read it as
Okonkwo et al. analyze why group relative policy optimization wastes computation on prompts whose sampled answers are either all correct or all incorrect, producing no learning signal.
The authors represent a prompt's pass rate, p, using the arc-length coordinate ψ = arcsin√p on the Fisher-Rao geometry of Bernoulli outcomes.
How it works
before and afterThe Daily Edition
The front page in your inbox each weekday morning: the lead, the picks, and why each one earned its place.