Patterning in Practice: Debiasing Reward Models with Susceptibilities
cs.LG
Submitted: 2026-09-01
Updated: 2026-09-01
Terminology
Sources
- The Loss Kernel: A Geometric Probe for Deep Learning Interpretability
- Concrete Problems in AI Safety
- ODIN: Disentangled Reward Mitigates Hacking in RLHF
- Deep reinforcement learning from human preferences
- Linear Response Estimators for Singular Statistical Models
- Susceptibilities and Patterning: A Primer on Linear Response in Bayesian Learning
- Influence Functions for Preference Dataset Pruning
- Gemma 2: Improving Open Language Models at a Practical Size
- Towards Spectroscopy: Susceptibility Clusters in Language Models
- Studying Large Language Model Generalization with Influence Functions
- WildGuard: Open One-Stop Moderation Tools for Safety Risks, Jailbreaks, and Refusals of LLMs
- From Global to Local: A Scalable Benchmark for Local Posterior Sampling
- Bayesian Influence Functions for Hessian-Free Data Attribution
- RewardBench: Evaluating Reward Models for Language Modeling
- You Are What You Eat -- AI Alignment Requires Understanding How Data Shapes Structure and Generalisation
- Skywork-Reward: Bag of Tricks for Reward Modeling in LLMs
- RM-Bench: Benchmarking Reward Models of Language Models with Subtlety and Style
- Spurious Correlation Learning in Preference Optimization: Mechanisms, Consequences, and Mitigation via Tie Training
- OffsetBias: Leveraging Debiased Data for Tuning Evaluators
- Disentangling Length from Quality in Direct Preference Optimization
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: Local-Consistency Optimisation for Bayesian Neural Networks
- Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks