Papers
Topics
Authors
Recent
Detailed Answer
Quick Answer
Concise responses based on abstracts only
Detailed Answer
Well-researched responses based on abstracts and relevant paper content.
Custom Instructions Pro
Preferences or requirements that you'd like Emergent Mind to consider when generating responses
Gemini 2.5 Flash
Gemini 2.5 Flash 60 tok/s
Gemini 2.5 Pro 51 tok/s Pro
GPT-5 Medium 18 tok/s Pro
GPT-5 High 14 tok/s Pro
GPT-4o 77 tok/s Pro
Kimi K2 159 tok/s Pro
GPT OSS 120B 456 tok/s Pro
Claude Sonnet 4 38 tok/s Pro
2000 character limit reached

One-Shot Safety Alignment for Large Language Models via Optimal Dualization (2405.19544v3)

Published 29 May 2024 in cs.AI, cs.LG, math.OC, stat.ML, and cs.CL

Abstract: The growing safety concerns surrounding LLMs raise an urgent need to align them with diverse human preferences to simultaneously enhance their helpfulness and safety. A promising approach is to enforce safety constraints through Reinforcement Learning from Human Feedback (RLHF). For such constrained RLHF, typical Lagrangian-based primal-dual policy optimization methods are computationally expensive and often unstable. This paper presents a perspective of dualization that reduces constrained alignment to an equivalent unconstrained alignment problem. We do so by pre-optimizing a smooth and convex dual function that has a closed form. This shortcut eliminates the need for cumbersome primal-dual policy iterations, greatly reducing the computational burden and improving training stability. Our strategy leads to two practical algorithms in model-based and preference-based settings (MoCAN and PeCAN, respectively). A broad range of experiments demonstrate the effectiveness and merits of our algorithms.

Citations (2)

Summary

We haven't generated a summary for this paper yet.

List To Do Tasks Checklist Streamline Icon: https://streamlinehq.com

Collections

Sign up for free to add this paper to one or more collections.

Lightbulb On Streamline Icon: https://streamlinehq.com

Continue Learning

We haven't generated follow-up questions for this paper yet.