# What do Reward Models Memorize?

arXiv: https://arxiv.org/abs/2607.24484 · HTML: https://mlsanity.com/paper/2607.24484/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Reward modeling and RLHF, one of 10 directions in LLM alignment / RLHF / preference optimisation.
- Builds on 42 of its 71 references that are in the map.
- First posted on arXiv 27 Jul 2026.

## Placed in

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

## Builds on (most cited first, 30 of 42)

- [Adam: A Method for Stochastic Optimization](https://mlsanity.com/paper/1412.6980/) (2014)
- [Training language models to follow instructions with human feedback](https://mlsanity.com/paper/2203.02155/) (2022)
- [Proximal Policy Optimization Algorithms](https://mlsanity.com/paper/1707.06347/) (2017)
- [LoRA: Low-Rank Adaptation of Large Language Models](https://mlsanity.com/paper/2106.09685/) (2021)
- [Direct Preference Optimization: Your Language Model is Secretly a Reward Model](https://mlsanity.com/paper/2305.18290/) (2023)
- [A Unified Approach to Interpreting Model Predictions](https://mlsanity.com/paper/1705.07874/) (2017)
- [Fine-Tuning Language Models from Human Preferences](https://mlsanity.com/paper/1909.08593/) (2019)
- [Scaling Laws for Reward Model Overoptimization](https://mlsanity.com/paper/2210.10760/) (2022)
- [A General Language Assistant as a Laboratory for Alignment](https://mlsanity.com/paper/2112.00861/) (2021)
- [Towards Understanding Sycophancy in Language Models](https://mlsanity.com/paper/2310.13548/) (2023)
- [GoEmotions: A Dataset of Fine-Grained Emotions](https://mlsanity.com/paper/2005.00547/) (2020)
- [Does Learning Require Memorization? A Short Tale about a Long Tail](https://mlsanity.com/paper/1906.05271/) (2019)
- [Interpretable Preferences via Multi-Objective Reward Modeling and Mixture-of-Experts](https://mlsanity.com/paper/2406.12845/) (2024)
- [A Long Way to Go: Investigating Length Correlations in RLHF](https://mlsanity.com/paper/2310.03716/) (2023)
- [Helping or Herding? Reward Model Ensembles Mitigate but do not Eliminate Reward Hacking](https://mlsanity.com/paper/2312.09244/) (2023)
- [RewardBench 2: Advancing Reward Model Evaluation](https://mlsanity.com/paper/2506.01937/) (2025)
- [Learning Multi-Level Features with Matryoshka Sparse Autoencoders](https://mlsanity.com/paper/2503.17547/) (2025)
- [Rule Based Rewards for Language Model Safety](https://mlsanity.com/paper/2411.01111/) (2024)
- [A Survey on Sparse Autoencoders: Interpreting the Internal Mechanisms of Large Language Models](https://mlsanity.com/paper/2503.05613/) (2025)
- [From Lists to Emojis: How Format Bias Affects Model Alignment](https://mlsanity.com/paper/2409.11704/) (2024)
- [Explore Spurious Correlations at the Concept Level in Language Models for Text Classification](https://mlsanity.com/paper/2311.08648/) (2023)
- [Curry-DPO: Enhancing Alignment using Curriculum Learning & Ranked Preferences](https://mlsanity.com/paper/2403.07230/) (2024)
- [Cultivating Pluralism In Algorithmic Monoculture: The Community Alignment Dataset](https://mlsanity.com/paper/2507.09650/) (2025)
- [Curriculum-RLAIF: Curriculum Alignment with Reinforcement Learning from AI Feedback](https://mlsanity.com/paper/2505.20075/) (2025)
- [Post-hoc Reward Calibration: A Case Study on Length Bias](https://mlsanity.com/paper/2409.17407/) (2024)
- [AnthroScore: A Computational Linguistic Measure of Anthropomorphism](https://mlsanity.com/paper/2402.02056/) (2024)
- [What's In My Human Feedback? Learning Interpretable Descriptions of Preference Data](https://mlsanity.com/paper/2510.26202/) (2025)
- [An Empirical Study of Memorization in NLP](https://mlsanity.com/paper/2203.12171/) (2022)
- [MIDAS: A Dialog Act Annotation Scheme for Open Domain Human Machine Spoken Conversations](https://mlsanity.com/paper/1908.10023/) (2019)
- [The Pitfalls of Memorization: When Memorization Hurts Generalization](https://mlsanity.com/paper/2412.07684/) (2024)

## Abstract (arXiv)

This paper studies what discriminatively trained reward models (RMs) memorize by measuring counterfactual memorization on two human preference datasets. We show that RMs 1) misallocate memorization to easy, high margin preference pairs, 2) memorize dataset-specific shortcuts (e.g., model identity, user sampling strategy), and 3) overgeneralize simple heuristic correlates of human preference (e.g., length, compliance) when confronted with unseen preference pairs. Overall, our findings indicate that discriminative training of RMs from human preference data results in biased RMs not yet capable of judging response quality in context-dependent scenarios.