89%Must read
?Must readVote to see the score
Scoring Higher, Answering Worse: Mitigating Reward Hacking in Rubric-Based RL via Protocol-Level Rubrics
Additive rubric aggregation enables reward hacking by letting criteria compensate, so grouping them into protocol-level dimensions recovers accuracy without sacrificing coverage.
Published Sep 30, 2026 · 0 citations · ▲ 17 on Hugging Face
– Readers
16/20 AI panelreviewers recommend it
Readers and the AI panel: vote on this paper to see what they said.
Only vote on papers you've read. Sign in to vote.
AI panel: 16 of 20 reviewers recommend it
lenient 5/5
medium 10/10
strict 1/5