No article text was recorded for this version.
reward hacking (proxy objective exploitation)
Version 2 of 2704 papers879 conceptsfrozen 2026-10-04
Papers in this version
- Blog: Scaling Laws for Reward Model Overoptimization
- Blog: How to Evaluate Reward Models for RLHF
- Blog: Reward Under Attack: Analyzing the Robustness and Hackability of Process Reward Models
- Blog: Rethinking Reward Model Evaluation: Are We Barking up the Wrong Tree?
- Blog: The Accuracy Paradox in RLHF: When Better Reward Models Don't Yield Better Language Models
- Blog: Scaling Laws for Reward Model Overoptimization in Direct Alignment Algorithms
- Blog: Elephant in the Room: Unveiling the Impact of Reward Model Quality in Alignment
- Blog: Reward Model Ensembles Help Mitigate Overoptimization
- Blog: Correcting the Mythos of KL-Regularization: Direct Alignment without Overoptimization via Chi-Squared Preference Optimization
- Blog: The Accuracy Paradox in RLHF: When Better Reward Models Don't Yield Better Language Models
- Blog: Inference-Time Reward Hacking in Large Language Models
- Blog: Reward Shaping to Mitigate Reward Hacking in RLHF
- Blog: Style Outweighs Substance: Failure Modes of LLM Judges in Alignment Benchmarking
- Blog: More is Less: The Pitfalls of Multi-Model Synthetic Preference Data in DPO Safety Alignment
- Blog: Information Theoretic Guarantees For Policy Alignment In Large Language Models
- Blog: The Alignment Ceiling: Objective Mismatch in Reinforcement Learning from Human Feedback
- Blog: Iterative Data Smoothing: Mitigating Reward Overfitting and Overoptimization in RLHF
- Blog: RRM: Robust Reward Model Training Mitigates Reward Hacking
- Blog: Learning to summarize from human feedback
- Blog: Directional Alignment Mitigates Reward Hacking in Reinforcement Learning for Language Models
- Blog: Reassessing the Role of Supervised Fine-Tuning: An Empirical Study in VLM Reasoning
- Blog: LongWriter-Zero: Mastering Ultra-Long Text Generation via Reinforcement Learning
- Blog: Mitigating Reward Hacking in LLM-based Recommendation: A Preference Optimization Approach
- Blog: Adversarial Training of Reward Models
- Blog: Countering Reward Over-optimization in LLM with Demonstration-Guided Reinforcement Learning
- Blog: Scaling Laws for Reward Model Overoptimization in Direct Alignment Algorithms
- Blog: Mitigating Preference Hacking in Policy Optimization with Pessimism
- Blog: Transforming and Combining Rewards for Aligning Large Language Models
- Blog: Jailbreaking as a Reward Misspecification Problem
- Blog: Qwen2.5 Technical Report
- Blog: Rethinking Reward Model Evaluation: Are We Barking up the Wrong Tree?
- Blog: Countering Reward Over-Optimization in LLM with Demonstration-Guided Reinforcement Learning
- Blog: Helping or Herding? Reward Model Ensembles Mitigate but do not Eliminate Reward Hacking
- Blog: Catastrophic Goodhart: regularizing RLHF with KL divergence does not mitigate heavy-tailed reward misspecification
- Blog: P1: Mastering Physics Olympiads with Reinforcement Learning
- Blog: Revisiting the (Sub)Optimality of Best-of-N for Inference-Time Alignment
- Blog: Reinforcement Learning Meets Large Language Models: A Survey of Advancements and Applications Across the LLM Lifecycle
- Blog: Understanding Reinforcement Learning-Based Fine-Tuning of Diffusion Models: A Tutorial and Review
- Blog: Jailbreaking as a Reward Misspecification Problem
- Blog: A Technical Survey of Reinforcement Learning Techniques for Large Language Models
- Blog: Exploring Question-Specific Rewards for Generating Deep Questions
- Blog: Bridging the Gap: A Survey on Integrating (Human) Feedback for Natural Language Generation
- Blog: Interpreting Language Reward Models via Contrastive Explanations
- Blog: The Anatomy of Alignment: Decomposing Preference Optimization by Steering Sparse Features
- Blog: Compositional Preference Models for Aligning LMs
- Blog: Secrets of RLHF in Large Language Models Part I: PPO
- Blog: Principled Reinforcement Learning with Human Feedback from Pairwise or K-wise Comparisons
- Blog: Mitigating Reward Over-Optimization in RLHF via Behavior-Supported Regularization
- Blog: Exploring Data Scaling Trends and Effects in Reinforcement Learning from Human Feedback
- Blog: SALMON: Self-Alignment with Instructable Reward Models
- Blog: WARM: On the Benefits of Weight Averaged Reward Models
- Blog: Adversarial Training of Reward Models
- Blog: A Comprehensive Survey of Reward Models: Taxonomy, Applications, Challenges, and Future
- Blog: Best-of-N through the Smoothing Lens: KL Divergence and Regret Analysis
- Blog: Reasoning Under 1 Billion: Memory-Augmented Reinforcement Learning for Large Language Models
- Blog: Reinforcement Learning from Human Feedback
- Blog: Reward Model Ensembles Help Mitigate Overoptimization
- Blog: Rewarding Curse: Analyze and Mitigate Reward Modeling Issues for LLM Reasoning
- Blog: PushGen: Push Notifications Generation with LLM
- Blog: Confronting Reward Model Overoptimization with Constrained RLHF
Showing 60 of 704 papers.
History & citation
Current · Permanent link
- The Reward-Utility Gap in Reinforcement Learning Alignment merged into this direction (50% overlap)
Cite this direction
Latest version (a living page):
@misc{lacuna_reward_hacking_proxy_objective_exploitat_37fa73bd,
title = {reward hacking (proxy objective exploitation)},
author = {{Lacuna}},
howpublished = {Lacuna research direction},
year = {2026},
url = {https://lacuna.tiptreesystems.com/direction/reward-hacking-proxy-objective-exploitation/rdr_37fa73bdcd3a41e0999ff626aae2074d},
note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-22},
}This exact version (version 2, frozen text):
@misc{lacuna_reward_hacking_proxy_objective_exploitat_37fa73bd_v2,
title = {reward hacking (proxy objective exploitation)},
author = {{Lacuna}},
howpublished = {Lacuna research direction},
year = {2026},
url = {https://lacuna.tiptreesystems.com/direction/reward-hacking-proxy-objective-exploitation/rdr_37fa73bdcd3a41e0999ff626aae2074d/v/2},
note = {Version 2, frozen 2026-10-04. Accessed 2026-11-22},
}