English
Related papers

Related papers: Soft Best-of-n Sampling for Model Alignment

200 papers

Inference-time alignment effectively steers large language models (LLMs) by generating multiple candidates from a reference model and selecting among them with an imperfect reward model. However, current strategies face a fundamental…

Artificial Intelligence · Computer Science 2026-03-10 Hsiang Hsu , Eric Lei , Chun-Fu Chen

We present an approach called Q-probing to adapt a pre-trained language model to maximize a task-specific reward function. At a high level, Q-probing sits between heavier approaches such as finetuning and lighter approaches such as few shot…

Machine Learning · Computer Science 2024-06-04 Kenneth Li , Samy Jelassi , Hugh Zhang , Sham Kakade , Martin Wattenberg , David Brandfonbrener

Existing studies on preference optimization (PO) have centered on constructing pairwise preference data following simple heuristics, such as maximizing the margin between preferred and dispreferred completions based on human (or AI) ranked…

Artificial Intelligence · Computer Science 2025-02-10 Zhuotong Chen , Fang Liu , Xuan Zhu , Yanjun Qi , Mohammad Ghavamzadeh

Language model (LM) post-training (or alignment) involves maximizing a reward function that is derived from preference annotations. Direct Preference Optimization (DPO) is a popular offline alignment method that trains a policy directly on…

Machine Learning · Computer Science 2025-03-04 Adam Fisch , Jacob Eisenstein , Vicky Zayats , Alekh Agarwal , Ahmad Beirami , Chirag Nagpal , Pete Shaw , Jonathan Berant

Language model alignment is a critical step in training modern generative language models. Alignment targets to improve win rate of a sample from the aligned model against the base model. Today, we are increasingly using inference-time…

We consider the problem of choosing the best of $n$ samples, out of a large random pool, when the sampling of each member is associated with a certain cost. The quality (worth) of the best sample clearly increases with $n$, but so do the…

Statistics Theory · Mathematics 2015-06-16 Joseph D. Skufca , Daniel ben-Avraham

We study the out-of-sample properties of robust empirical optimization problems with smooth $\phi$-divergence penalties and smooth concave objective functions, and develop a theory for data-driven calibration of the non-negative "robustness…

Machine Learning · Statistics 2020-05-20 Jun-Ya Gotoh , Michael Jong Kim , Andrew E. B. Lim

We propose Guided Speculative Inference (GSI), a novel algorithm for efficient reward-guided decoding in large language models. GSI combines soft best-of-$n$ test-time scaling with a reward model $r(x,y)$ and speculative samples from a…

Machine Learning · Computer Science 2026-04-28 Jonathan Geuter , Youssef Mroueh , David Alvarez-Melis

The relationship between the quality of a string, as judged by a human reader, and its probability, $p(\boldsymbol{y})$ under a language model undergirds the development of better language models. For example, many popular algorithms for…

Computation and Language · Computer Science 2024-10-29 Naaman Tan , Josef Valvoda , Tianyu Liu , Anej Svete , Yanxia Qin , Kan Min-Yen , Ryan Cotterell

Despite some empirical success at correcting exposure bias in machine translation, scheduled sampling algorithms suffer from a major drawback: they incorrectly assume that words in the reference translations and in sampled sequences are…

Computation and Language · Computer Science 2019-05-07 Weijia Xu , Xing Niu , Marine Carpuat

Language model (LM) alignment improves model outputs to reflect human preferences while preserving the capabilities of the base model. The most common alignment approaches are (i) reinforcement learning, which maximizes the expected reward…

Machine Learning · Computer Science 2026-05-11 Lucas Monteiro Paes , Natalie Mackraz , Barry-John Theobald , Federico Danieli

Aligning large language models (LLMs) with human preferences has become a critical step in their development. Recent research has increasingly focused on test-time alignment, where additional compute is allocated during inference to enhance…

Computation and Language · Computer Science 2025-09-24 Bolian Li , Yanran Wu , Xinyu Luo , Ruqi Zhang

Aligning language models (LMs) to human preferences has emerged as a critical pursuit, enabling these models to better serve diverse user needs. Existing methods primarily focus on optimizing LMs for a single reward function, limiting their…

Machine Learning · Computer Science 2024-10-29 Ruizhe Shi , Yifang Chen , Yushi Hu , Alisa Liu , Hannaneh Hajishirzi , Noah A. Smith , Simon S. Du

We introduce Robust Multi-Objective Decoding (RMOD), a novel inference-time algorithm that robustly aligns Large Language Models (LLMs) to multiple human objectives (e.g., instruction-following, helpfulness, safety) by maximizing the…

Machine Learning · Computer Science 2026-02-17 Seongho Son , William Bankes , Sangwoong Yoon , Shyam Sundhar Ramesh , Xiaohang Tang , Ilija Bogunovic

Best-of-N sampling is a powerful method for improving Large Language Model (LLM) performance, but it is often limited by its dependence on massive, text-based reward models. These models are not only computationally expensive but also…

Machine Learning · Computer Science 2026-01-09 Jizhou Guo , Zhaomin Wu , Hanchen Yang , Philip S. Yu

Aligning diffusion models to downstream tasks often requires finetuning new models or gradient-based guidance at inference time to enable sampling from the reward-tilted posterior. In this work, we explore a simple inference-time…

Computer Vision and Pattern Recognition · Computer Science 2025-05-06 Anuj Singh , Sayak Mukherjee , Ahmad Beirami , Hadi Jamali-Rad

We introduce a principled probabilistic framework for reward-guided decoding in large language models, addressing the limitations of standard decoding methods that optimize token-level likelihood rather than sequence-level quality. Our…

Machine Learning · Computer Science 2026-04-21 Jelena Markovic-Voronov , Wenhui Zhu , Bo Long , Zhipeng Wang , Suyash Gupta , Kayhan Behdin , Bee-Chung Chen , Deepak Agarwal

Despite the remarkable advances in language modeling, current mainstream decoding methods still struggle to generate texts that align with human texts across different aspects. In particular, sampling-based methods produce less-repetitive…

Computation and Language · Computer Science 2024-06-06 Haozhe Ji , Pei Ke , Hongning Wang , Minlie Huang

Reward learning enables the application of reinforcement learning (RL) to tasks where reward is defined by human judgment, building a model of reward by asking humans questions. Most work on reward learning has used simulated environments,…

Computation and Language · Computer Science 2020-01-10 Daniel M. Ziegler , Nisan Stiennon , Jeffrey Wu , Tom B. Brown , Alec Radford , Dario Amodei , Paul Christiano , Geoffrey Irving

Diffusion models excel in generative tasks, but aligning them with specific objectives while maintaining their versatility remains challenging. Existing fine-tuning methods often suffer from reward over-optimization, while approximate…

Machine Learning · Computer Science 2025-04-18 Sunwoo Kim , Minkyu Kim , Dongmin Park