# Stabilizing RLHF through Advantage Model and Selective Rehearsal

arXiv: https://arxiv.org/abs/2309.10202 · HTML: https://mlsanity.com/paper/2309.10202/

## Key facts

- Cited by 14 papers in the map (data through 14 Oct 2026).
- Placed in Reward modeling and RLHF, one of 10 directions in LLM alignment / RLHF / preference optimisation.
- Builds on 30 of its 35 references that are in the map.
- First posted on arXiv 18 Sep 2023.

## 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 30)

- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [GPT-4 Technical Report](https://mlsanity.com/paper/2303.08774/) (2023)
- [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)
- [Llama 2: Open Foundation and Fine-Tuned Chat Models](https://mlsanity.com/paper/2307.09288/) (2023)
- [Training Verifiers to Solve Math Word Problems](https://mlsanity.com/paper/2110.14168/) (2021)
- [Direct Preference Optimization: Your Language Model is Secretly a Reward Model](https://mlsanity.com/paper/2305.18290/) (2023)
- [Deep reinforcement learning from human preferences](https://mlsanity.com/paper/1706.03741/) (2017)
- [Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback](https://mlsanity.com/paper/2204.05862/) (2022)
- [Constitutional AI: Harmlessness from AI Feedback](https://mlsanity.com/paper/2212.08073/) (2022)
- [Self-Instruct: Aligning Language Models with Self-Generated Instructions](https://mlsanity.com/paper/2212.10560/) (2022)
- [Learning to summarize from human feedback](https://mlsanity.com/paper/2009.01325/) (2020)
- [DeBERTa: Decoding-enhanced BERT with Disentangled Attention](https://mlsanity.com/paper/2006.03654/) (2020)
- [SimCSE: Simple Contrastive Learning of Sentence Embeddings](https://mlsanity.com/paper/2104.08821/) (2021)
- [LIMA: Less Is More for Alignment](https://mlsanity.com/paper/2305.11206/) (2023)
- [A General Language Assistant as a Laboratory for Alignment](https://mlsanity.com/paper/2112.00861/) (2021)
- [Red Teaming Language Models to Reduce Harms: Methods, Scaling Behaviors, and Lessons Learned](https://mlsanity.com/paper/2209.07858/) (2022)
- [OpenAssistant Conversations -- Democratizing Large Language Model Alignment](https://mlsanity.com/paper/2304.07327/) (2023)
- [Instruction Tuning with GPT-4](https://mlsanity.com/paper/2304.03277/) (2023)
- [RAFT: Reward rAnked FineTuning for Generative Foundation Model Alignment](https://mlsanity.com/paper/2304.06767/) (2023)
- [Improving alignment of dialogue agents via targeted human judgements](https://mlsanity.com/paper/2209.14375/) (2022)
- [Reinforced Self-Training (ReST) for Language Modeling](https://mlsanity.com/paper/2308.08998/) (2023)
- [RRHF: Rank Responses to Align Language Models with Human Feedback without tears](https://mlsanity.com/paper/2304.05302/) (2023)
- [Scalable agent alignment via reward modeling: a research direction](https://mlsanity.com/paper/1811.07871/) (2018)
- [How Far Can Camels Go? Exploring the State of Instruction Tuning on Open Resources](https://mlsanity.com/paper/2306.04751/) (2023)
- [Preference Ranking Optimization for Human Alignment](https://mlsanity.com/paper/2306.17492/) (2023)
- [Towards Continual Reinforcement Learning: A Review and Perspectives](https://mlsanity.com/paper/2012.13490/) (2020)
- [Continual Pre-Training of Large Language Models: How to (re)warm your model?](https://mlsanity.com/paper/2308.04014/) (2023)
- [MM-LIMA: Less Is More for Alignment in Multi-Modal Datasets](https://mlsanity.com/paper/2308.12067/) (2023)
- [ValueNet: A New Dataset for Human Value Driven Dialogue System](https://mlsanity.com/paper/2112.06346/) (2021)

## Built on by (most cited first, 14 of 14)

- [Dense Reward for Free in Reinforcement Learning from Human Feedback](https://mlsanity.com/paper/2402.00782/) (2024)
- [Systematic Evaluation of LLM-as-a-Judge in LLM Alignment Tasks: Explainable Metrics and Diverse Prompt Templates](https://mlsanity.com/paper/2408.13006/) (2024)
- [Iterative Nash Policy Optimization: Aligning LLMs with General Preferences via No-Regret Learning](https://mlsanity.com/paper/2407.00617/) (2024)
- [The Alignment Ceiling: Objective Mismatch in Reinforcement Learning from Human Feedback](https://mlsanity.com/paper/2311.00168/) (2023)
- [Modality-Fair Preference Optimization for Trustworthy MLLM Alignment](https://mlsanity.com/paper/2410.15334/) (2024)
- [Improving LLM General Preference Alignment via Optimistic Online Mirror Descent](https://mlsanity.com/paper/2502.16852/) (2025)
- [HyperCLOVA X Technical Report](https://mlsanity.com/paper/2404.01954/) (2024)
- [Reasons to Reject? Aligning Language Models with Judgments](https://mlsanity.com/paper/2312.14591/) (2023)
- [Sail into the Headwind: Alignment via Robust Rewards and Dynamic Labels against Reward Hacking](https://mlsanity.com/paper/2412.09544/) (2024)
- [Mapping Post-Training Forgetting in Language Models at Scale](https://mlsanity.com/paper/2510.17776/) (2025)
- [Rethinking the Sampling Criteria in Reinforcement Learning for LLM Reasoning: A Competence-Difficulty Alignment Perspective](https://mlsanity.com/paper/2505.17652/) (2025)
- [Enabling Language Models to Implicitly Learn Self-Improvement](https://mlsanity.com/paper/2310.00898/) (2023)
- [COPR: Continual Learning Human Preference through Optimal Policy Regularization](https://mlsanity.com/paper/2310.15694/) (2023)
- [HSCR: Hierarchical Self-Contrastive Rewarding for Aligning Medical Vision Language Models](https://mlsanity.com/paper/2506.00805/) (2025)

## Abstract (arXiv)

Large Language Models (LLMs) have revolutionized natural language processing, yet aligning these models with human values and preferences using RLHF remains a significant challenge. This challenge is characterized by various instabilities, such as reward hacking and catastrophic forgetting. In this technical report, we propose two innovations to stabilize RLHF training: 1) Advantage Model, which directly models advantage score i.e., extra reward compared to the expected rewards and regulates score distributions across tasks to prevent reward hacking. 2) Selective Rehearsal, which mitigates catastrophic forgetting by strategically selecting data for PPO training and knowledge rehearsing. Our experimental analysis on public and proprietary datasets reveals that the proposed methods not only increase stability in RLHF training but also achieve higher reward scores and win rates.