arXiv:2512. 19905v3 Announce Type: replace-cross Abstract: Recent developments in large language models have shown advantages in reallocating a notable share of computational resource from training time to inference time.
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:2602.13551v3 Announce Type: replace Abstract: Reward models (RMs) play a central role throughout the language model (LM) pipeline, particularly in non-verifiable domains. However, the dominant...
The paper addresses reward hacking that arises when large language models (LLMs) are post‑trained using cheap surrogate judges instead of expensive human feedback. It introduces envelope sampling, a theoretically grounded recalibration technique that uses a small set of ground‑truth annotated outputs to adjust the surrogate judge before further training. Experiments on clinical note generation and a sycophancy task demonstrate that envelope sampling reduces reward hacking compared to traditional base‑model recalibration methods.
arXiv:2605. 17609v2 Announce Type: replace Abstract: Many inference-time language-model pipelines combine a cheap reward signal with an expensive verifier, such as exact answer checking in mathematical reasoning or hidden-test execution in code generation.
arXiv:2608. 22915v1 Announce Type: cross Abstract: Inference-time pipelines often sample multiple outputs, filter them with a learned safety model, and return the proxy-feasible output with the highest learned reward.
arXiv:2609.38672v1 Announce Type: cross Abstract: Beam-search-based test-time methods provide an effective way to improve large language model (LLM) performance on long-horizon generation by pruning...
arXiv:2610. 21363v1 Announce Type: cross Abstract: Test-time scaling can improve the performance of large language models by generating multiple answers to a question and aggregating them.
arXiv:2605.20854v3 Announce Type: replace Abstract: We provide the first regret analysis of ReMax in stochastic multi-armed bandits. Originally introduced for reinforcement learning, ReMax is motivat...
arXiv:2604. 10727v2 Announce Type: replace-cross Abstract: Classical information-theoretic learning bounds typically rely on KL mutual information and moment-generating-function (MGF) arguments, which are well matched to bounded or sub-Gaussian losses but can be ineffective when losses or rewards are heavy-tailed.
The paper introduces SELECT, an algorithmic framework for satisficing regret minimization in bandit problems, achieving constant expected satisficing regret when a satisficing arm exists. A variant, SELECT‑LITE, further ensures a light‑tailed satisficing regret distribution while maintaining constant expected regret in the realizable case and sub‑linear standard regret otherwise. Experiments on synthetic data and a real‑world dynamic pricing scenario demonstrate the practical effectiveness of both algorithms.
arXiv:2608. 08002v1 Announce Type: new Abstract: Language-model judges and reward models enable scalable supervision, but finite optimization can exploit evaluator errors rather than improve response quality.