arXiv:2512. 21917v3 Announce Type: replace-cross Abstract: Policy alignment to preference data typically assumes a known link function between observed preferences and latent rewards (e.
arXiv:2606. 01382v1 Announce Type: cross Abstract: Preference alignment is central to improving large language models, but standard reward-based formulations can be restrictive when human preferences are cyclic, non-transitive, or otherwise not representable by a scalar reward.
The paper investigates preference-based bandits where a learner selects pairs of arms and receives binary preference feedback modeled by Bradley–Terry. It introduces the locally sensitive eluder dimension, a new complexity measure for logistic preference feedback, and proposes the GINOP algorithm that uses log-loss confidence sets to balance optimism and exploration. The authors prove a first-order regret bound showing that learning with preference feedback can be as statistically efficient as learning from direct rewards, and they validate their theory with empirical experiments.
arXiv:2607. 26358v1 Announce Type: new Abstract: Reinforcement learning (RL) fine-tuning is widely used in language model training to improve model performance on a target task while limiting drift from a reference policy.
Reinforcement learning (RL) fine-tuning is widely used in language model training to improve model performance on a target task while limiting drift from a reference policy. A standard way to balance this trade-off is via a KL-regularized RL objective, although this formulation does not by itself provide a principled way to set the regularization coefficient.
The paper presents a new scaling law for reward optimization in AI alignment, showing that performance scales as Θ(√min{log(M), K}), where M is the number of preference comparisons used to train a proxy reward model and K is the KL‑divergence budget relative to a reference policy. The authors derive this law using an information‑theoretic model, prove its tightness, and validate it with extensive experiments involving a 70B gold reward model and smaller proxy models (0.6B–4B). The empirical results demonstrate a strong fit (R² 97–99 %) across different model sizes, noise levels, and optimization methods, suggesting that reward optimization behaves like a simple selection task over IID Gaussian variables with noisy feedback.
arXiv:2607. 26094v1 Announce Type: new Abstract: Reinforcement Learning from Human Feedback (RLHF) is the standard approach for aligning large language models with human preferences, but its quality is limited by static, task-agnostic reward models.
arXiv:2606. 18111v1 Announce Type: cross Abstract: Fairness is an important aspect of decision-making in multi-objective reinforcement learning (MORL), where policies must ensure both optimality and equity across multiple, potentially conflicting objectives.
arXiv:2609.37209v1 Announce Type: new Abstract: Pairwise losses are increasingly used for reward learning even when pointwise rewards are observed, with mixed empirical results. When and why do pairw...
arXiv:2403.05006v2 Announce Type: replace-cross Abstract: Pluralistic alignment requires learning from feedback that reflects persistent and potentially conflicting stakeholder preferences while ulti...
arXiv:2501.06926v5 Announce Type: replace Abstract: Double reinforcement learning (DRL) provides efficient off-policy inference for policy values in nonparametric Markov decision processes (MDPs), bu...
arXiv:2606. 19117v1 Announce Type: cross Abstract: Offline policy learning has received growing attention in causal inference.
arXiv:2607. 11432v1 Announce Type: new Abstract: In this work, we study the reinforcement learning (RL) problem from pairwise trajectory comparisons provided by a human expert.