paper-with-me

Papers

Distributionally Robust Token Optimization in RLHF

2026-03-27 · Yeping Jin, Jiaming Hu, Ioannis Ch. Paschalidis arxiv

Large Language Models (LLMs) tend to respond correctly to prompts that align well with the data they were trained and fine-tuned on. Yet, small shifts in wording, format, or language can trigger surprisingly large failures, especially on multi-step reasoning problems. To address this problem, we propose a Distributionally Robust Token Optimization (DRTO) approach, which combines token-level Reinforcement Learning from Human Feedback (RLHF) with Distributionally Robust Optimization (DRO). DRTO constructs f-divergence ambiguity sets over span-level actor losses, providing a principled way to emphasize difficult response segments during policy optimization. Empirically, DRTO enhances consistency under distribution shifts in multiple reasoning benchmarks among different tasks, achieving $+4.4$ percentage points on MATH-500 and $+2.7$ percentage points on LiveCodeBench over standard RTO.

📄 PDF Abstract BibTeX arXiv:2604.08577

Code (0)

등록된 구현이 없습니다.

Tasks

Reinforcement Learning

Similar Papers 제목 키워드 기반

Wasserstein Distributionally Robust Regret Optimization for Reinforcement Learning from Human Feedback

2026-04-30 · Yikai Wang, Shang Liu, Jose Blanchet arxiv

Reinforcement learning from human feedback (RLHF) is a central post-training tool for aligning large language models, but its training reward is only a learned proxy for true human utility. This creates a decision proble…

Reinforcement Learning

A Unifying Lens on Reward Uncertainty in RLHF

2026-06-08 · Ely Hahami, Yoel Zimmermann, Ray Zhou, Jack Benarroch Jedlicki arxiv

Reinforcement learning from human feedback (RLHF) is bottlenecked by reward hacking, where the policy exploits errors in a proxy reward model (RM) and produces high RM scores without genuine quality gains. A natural miti…

Reinforcement LearningBayesian Inference

Online Distributionally Robust LLM Alignment via Regression to Relative Reward

2025-09-23 · Sharan Sahu, Martin T. Wells arxiv

Reinforcement Learning with Human Feedback (RLHF) has become crucial for aligning Large Language Models (LLMs) with human intent. However, existing offline RLHF approaches suffer from overoptimization, where language mod…

Reinforcement Learning

Enhancing RLHF with Human Gaze Modeling

2025-07-11 · Karim Galliamov, Ivan Titov, Ilya Pershin arxiv

Reinforcement Learning from Human Feedback (RLHF) aligns language models with human preferences but is computationally expensive. We explore two approaches that leverage human gaze modeling to enhance RLHF: (1) gaze-awar…

Reinforcement Learning

AlignDistil: Token-Level Language Model Alignment as Adaptive Policy Distillation

2025-03-04 · Songming Zhang, Xue Zhang, Tong Zhang, Bojie Hu 외

In modern large language models (LLMs), LLM alignment is of crucial importance and is typically achieved through methods such as reinforcement learning from human feedback (RLHF) and direct preference optimization (DPO).…

Language ModelingLanguage Modelling