Aligning Language Models with General Preferences: An Empirical Study of Direct Nash Optimization
cs.LG, cs.AI, cs.CL
Submitted: 2026-11-03
Updated: 2026-11-03
Code: https://github.com/tatsu-lab/alpaca_eval
Terminology
Sources
- Concrete Problems in AI Safety
- A General Theoretical Paradigm to Understand Learning from Human Preferences
- Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback
- Language Models are Few-Shot Learners
- Human Alignment of Large Language Models through Online Preference Optimisation
- Do LLM Evaluators Prefer Themselves for a Reason?
- Self-Play Fine-Tuning Converts Weak Language Models to Strong Language Models
- Adversarial Preference Optimization: Enhancing Your Alignment via RM-LLM Game
- Chatbot Arena: An Open Platform for Evaluating LLMs by Human Preference
- Deep reinforcement learning from human preferences
- UltraFeedback: Boosting Language Models with Scaled AI Feedback
- RLHF Workflow: From Reward Modeling to Online RLHF
- Length-Controlled AlpacaEval: A Simple Way to Debias Automatic Evaluators
- REBEL: Reinforcement Learning via Regressing Relative Rewards
- The Llama 3 Herd of Models
- Direct Language Model Alignment from Online AI Feedback
- The Curious Case of Neural Text Degeneration
- Unpacking DPO and PPO: Disentangling Best Practices for Learning from Preference Feedback
- Mistral 7B
- Mixtral of Experts
Related papers
- A Physics-Regulated Neural Framework for Learning 3D Grain Growth Dynamics
- FlexMoE: Learning One-for-All Multi-Budget Expert Pruning for Flexible MoE Deployment
- OptiMind: Improving Small Language Models for Optimization Modeling through Formulation Guidance and Multi-Turn Learning
- Staying in the Sweet Spot: Boost Reasoning Evolution via Responsive Rollout Difficulty Manipulation
- Examining the Impact of Data and Training Methods on Emergent and Subliminal Misalignment
- Rethinking Black-Box On-Policy Distillation: A Rubric-based Analysis