Understanding Helpfulness and Harmless Tension in Reward Models

arXiv:2606.13209 · cs.LG, cs.CL · Submitted 2026-06-11 · Read on arXiv

cs.LG, cs.CL

Submitted: 2026-06-11

Updated: 2026-10-13

Code: https://github.com/EshaanT/RM-alignment_tension

Terminology

• Alignment — 5× in this paper · explained in 1 episode
• Ablation — 1× in this paper · explained in 1 episode
• Alignment Methods — 1× in this paper · explained in 1 episode
• Reinforcement Learning — 1× in this paper · explained in 4 episodes

Sources

• SmolLM2: When Smol Goes Big -- Data-Centric Training of a Small Language Model
• Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback
• Constitutional AI: Harmlessness from AI Feedback
• BeaverTails: Towards Improved Safety Alignment of LLM via a Human-Preference Dataset
• RewardBench: Evaluating Reward Models for Language Modeling
• SAFER: Probing Safety in Reward Models with Sparse Autoencoder
• RewardBench 2: Advancing Reward Model Evaluation
• Training language models to follow instructions with human feedback
• Proximal Policy Optimization Algorithms
• DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models
• Secrets of RLHF in Large Language Models Part II: Reward Modeling

Related papers

• Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximation
• AIRL-S: Unifying Reinforcement Learning and Search-Based Test-Time Scaling via Adversarial Inverse Reinforcement Learning
• Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation
• Convergence issues in Relational Concept Analysis based on AOC-posets
• Beliefs Beyond Posteriors: The Predictive Cost of ELBO Binding
• Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks