# Rubric Dropout: A Simple Way to Mitigate Reward Hacking in Rubric-as-Reward RL

arXiv: https://arxiv.org/abs/2608.11669 · HTML: https://mlsanity.com/paper/2608.11669/

## Key facts

- Cited by 9 papers in the map (data through 14 Oct 2026).
- Placed in Reward hacking and misspecification, one of 10 directions in LLM alignment / RLHF / preference optimisation.
- Builds on 26 of its 34 references that are in the map.
- First posted on arXiv 12 Aug 2026.

## Placed in

- LLM alignment / RLHF / preference optimisation / [Reward hacking and misspecification](https://mlsanity.com/alignment/reward-hacking-and-misspecification/)

## Builds on (most cited first, 26 of 26)

- [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://mlsanity.com/paper/2402.03300/) (2024)
- [Qwen3 Technical Report](https://mlsanity.com/paper/2505.09388/) (2025)
- [Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena](https://mlsanity.com/paper/2306.05685/) (2023)
- [DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning](https://mlsanity.com/paper/2501.12948/) (2025)
- [Concrete Problems in AI Safety](https://mlsanity.com/paper/1606.06565/) (2016)
- [Scaling Laws for Reward Model Overoptimization](https://mlsanity.com/paper/2210.10760/) (2022)
- [Tulu 3: Pushing Frontiers in Open Language Model Post-Training](https://mlsanity.com/paper/2411.15124/) (2024)
- [HealthBench: Evaluating Large Language Models Towards Improved Human Health](https://mlsanity.com/paper/2505.08775/) (2025)
- [Rubrics as Rewards: Reinforcement Learning Beyond Verifiable Domains](https://mlsanity.com/paper/2507.17746/) (2025)
- [The Effects of Reward Misspecification: Mapping and Mitigating Misaligned Models](https://mlsanity.com/paper/2201.03544/) (2022)
- [Reward Model Ensembles Help Mitigate Overoptimization](https://mlsanity.com/paper/2310.02743/) (2023)
- [Helping or Herding? Reward Model Ensembles Mitigate but do not Eliminate Reward Hacking](https://mlsanity.com/paper/2312.09244/) (2023)
- [GDPO: Group reward-Decoupled Normalization Policy Optimization for Multi-reward RL Optimization](https://mlsanity.com/paper/2601.05242/) (2026)
- [WARM: On the Benefits of Weight Averaged Reward Models](https://mlsanity.com/paper/2401.12187/) (2024)
- [ODIN: Disentangled Reward Mitigates Hacking in RLHF](https://mlsanity.com/paper/2402.07319/) (2024)
- [OpenRubrics: Towards Scalable Synthetic Rubric Generation for Reward Modeling and LLM Alignment](https://mlsanity.com/paper/2510.07743/) (2025)
- [Checklists Are Better Than Reward Models For Aligning Language Models](https://mlsanity.com/paper/2507.18624/) (2025)
- [Reinforcement Learning with Rubric Anchors](https://mlsanity.com/paper/2508.12790/) (2025)
- [Breaking the Exploration Bottleneck: Rubric-Scaffolded Reinforcement Learning for General LLM Reasoning](https://mlsanity.com/paper/2508.16949/) (2025)
- [AdvancedIF: Rubric-Based Benchmarking and Reinforcement Learning for Advancing LLM Instruction Following](https://mlsanity.com/paper/2511.10507/) (2025)
- [ResearchQA: Evaluating Scholarly Question Answering at Scale Across 75 Fields with Survey-Mined Questions and Rubrics](https://mlsanity.com/paper/2509.00496/) (2025)
- [Online Rubrics Elicitation from Pairwise Comparisons](https://mlsanity.com/paper/2510.07284/) (2025)
- [LLMEval-Med: A Real-world Clinical Benchmark for Medical LLMs with Physician Validation](https://mlsanity.com/paper/2506.04078/) (2025)
- [Reward Hacking in Rubric-Based Reinforcement Learning](https://mlsanity.com/paper/2605.12474/) (2026)
- [Reproducing, Analyzing, and Detecting Reward Hacking in Rubric-Based Reinforcement Learning](https://mlsanity.com/paper/2606.04923/) (2026)
- [Not Every Rubric Teaches Equally: Policy-Aware Rubric Rewards for RLVR](https://mlsanity.com/paper/2605.20164/) (2026)

## Built on by (most cited first, 9 of 9)

- [ImpossibleRubrics: Stress-Testing Generated Rubrics as Reward Signals](https://mlsanity.com/paper/2609.16816/) (2026)
- [RubricArmor: Adversarial Evolution Improves LLM-Based Rubric Generation](https://mlsanity.com/paper/2610.05308/) (2026)
- [Binarization Flattens the Score Space](https://mlsanity.com/paper/2609.35797/) (2026)
- [Scoring Higher, Answering Worse: Mitigating Reward Hacking in Rubric-Based RL via Protocol-Level Rubrics](https://mlsanity.com/paper/2609.38847/) (2026)
- [MetaRubric: Learning to Reward for Rubric-Based Reinforcement Learning](https://mlsanity.com/paper/2610.02824/) (2026)
- [IdeaScientist: Orchestrating Agents for Grounded Scientific Ideation](https://mlsanity.com/paper/2610.04074/) (2026)
- [Self-Evaluating Recursive Agents](https://mlsanity.com/paper/2610.04902/) (2026)
- [Reflective Prompt Optimization of a Deep-Research Agent: A Topic-Disjoint, Multi-Seed Protocol for Measuring Transferable Gains and Prompt-Level Overfitting](https://mlsanity.com/paper/2610.12601/) (2026)
- [Understanding and Reducing Misalignment in Persuasion Optimization](https://mlsanity.com/paper/2610.16070/) (2026)

## Abstract (arXiv)

Rubric-based reinforcement learning, in which an LLM judge scores responses against prompt-specific criteria, has become a standard way to post-train language models on tasks without verifiable answers. The rubric, however, is only a proxy for response quality, and a policy trained against it long enough learns to exploit the difference. We train models with GRPO on medical and science rubrics and grade their responses on out-of-distribution (OOD) benchmarks throughout training with two judges: the training judge and a stronger OOD judge. The two scores first rise together and then diverge. The policy is learning to satisfy the training judge without writing better responses, which is a form of reward hacking. To mitigate it, we propose Rubric Dropout, which applies dropout to rubric criteria: at each training step, a random subset of each rubric's criteria is removed from the reward, so the rewarded subset of criteria is resampled at each step. All responses sampled for the same prompt are scored on the same subset, so their rewards remain comparable, and evaluation always uses the full rubric. We also introduce Block Dropout, which drops semantically similar criteria together so that a near duplicate of a dropped criterion cannot keep rewarding the same behavior. Across both benchmarks and models from 1.7B to 8B, dropout improves the gold judge's score on OOD benchmarks by up to 3 percent on HealthBench-Hard and 7 percent on ResearchQA, and it reduces both the gap between the two judges and over-crediting. It keeps in-domain reward close to the baseline, and dropout rates of 30-50 percent work best. Block Dropout improves the gold judge's OOD score further at the same dropout rate. Changing which criteria are rewarded at each step is a simple, inexpensive way to make rubric rewards harder to exploit.