pass-at-k-inversion

Tag

Cards List
#pass-at-k-inversion

When RLVR Shrinks the Reasoning Boundary: Diagnosing Pass@k Inversion

arXiv cs.LG ↗ · 2026-07-24 Cached

This paper diagnoses 'pass@k inversion' in reinforcement learning with verifiable rewards (RLVR), where training improves one-shot accuracy but reduces performance under repeated sampling, particularly on boundary prompts where correct trajectories are rare. It proposes a per-problem base anchoring (PBA) method to mitigate this issue.

0 favorites 0 likes
← Back to home

Submit Feedback