Predicting Alignment Generalization with Value Representations
cs.CL, cs.AI, cs.LG
Submitted: 2026-10-08
Updated: 2026-10-08
Code: https://github.com/andyjliu/value-generalization
Terminology
Sources
- Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback
- Constitutional AI: Harmlessness from AI Feedback
- Subliminal Learning Is Steering Vector Distillation
- Persona Vectors: Monitoring and Controlling Character Traits in Language Models
- Constitutional Midtraining: Content Presence Drives Alignment Gains
- Sycophancy to Subterfuge: Investigating Reward-Tampering in Large Language Models
- The Llama 3 Herd of Models
- Reinforcement Learning Towards Broadly and Persistently Beneficial Models
- What are human values, and how do we align AI to them?
- Model Spec Midtraining: Improving How Alignment Training Generalizes
- Natural Emergent Misalignment from Reward Hacking in Production RL
- Open Character Training: Shaping the Persona of AI Assistants through Constitutional AI
- OpenAI GPT-5 System Card
- When Role-playing, Do Models Believe What They Say?
- Olmo 3
- Qwen3 Technical Report
- Can Pre-training Indicators Reliably Predict Fine-tuning Outcomes of LLMs?
Related papers
- Exploring Solution Divergence and Its Effect on Large Language Model Problem Solving
- Ishigaki-IDS-Bench: A Benchmark for Generating Information Delivery Specification from BIM Information Requirements
- Subliminal Steering: Stronger Encoding of Hidden Signals
- MedStruct-S: A Benchmark for Key Discovery, Key-Conditioned QA and Semi-Structured Extraction from OCR Clinical Reports
- The End of Transformers? On Challenging Attention and the Rise of Sub-Quadratic Architectures
- Untangling the Mechanisms of Misleading Context in Medical Question Answering