Can Bayesian Optimization Efficiently Find a Strong Single Expert in Neural Thickets?

• URL: http://arxiv.org/abs/2608.10867v1
• Date: Tue, 11 Aug 2026 12:41:49 GMT
• Title: Can Bayesian Optimization Efficiently Find a Strong Single Expert in Neural Thickets?
• Abstract summary: We show that a surrogate-guided search can substantially reduce the evaluation cost of gradient-free post-training.<n>Our method requires no backpropagation and uses a Gaussian process surrogate to guide candidate evaluations efficiently.<n>These results show that surrogate-guided search can substantially reduce the evaluation cost of gradient-free post-training while producing stronger deployable single experts.
• Score: 44.39621309549252
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Gradient-free post-training has emerged as a compelling alternative to gradient-based optimization for large language models (LLMs), but existing approaches remain costly. We ask whether structured search can identify a strong single expert under a modest evaluation budget. Motivated by evidence that useful weight updates lie in low-dimensional subspaces, we apply Bayesian optimization within a random linear embedding of weight space. Our method requires no backpropagation and uses a Gaussian process surrogate to guide candidate evaluations efficiently. Across several reasoning benchmarks with Qwen2.5-Instruct models from 0.5B to 3B parameters, Bayesian optimization using five times less candidate evaluations matches or exceeds RandOpt. These results show that surrogate-guided search can substantially reduce the evaluation cost of gradient-free post-training while producing stronger deployable single experts.

Related papers

• Rethinking Learnability in Offline Data-driven Optimization [20.146021332129703]
  Black-Box Optimization (BBO) faces efficiency challenges as real-world problems grow increasingly complex.<n>Data-driven optimization has been the most popular paradigm to improve the efficiency of BBO, by learning from data.<n>Many offline optimization methods have been proposed, but a fundamental question remains unanswered: what learnability is sufficient for offline optimization?
  arXiv  Detail & Related papers  (2026-09-01T16:22:28Z)
• Generative Refinement for Low-Budget Black-Box Optimization [33.559218731885046]
  Black-box optimization is a fundamental science and engineering tool that makes it possible to optimize objectives without information.<n>Existing methods leveraging generative models to navigate subspaces are built to sample from reward-aligned distributions.<n>We propose SPARROW, an algorithm that completely decouples the generative prior from the reward signal.
  arXiv  Detail & Related papers  (2026-07-01T09:41:41Z)
• Large-scale empirical tuning and comparison of default optimizers for variational inference [16.671882609909396]
  Black-box variational inference (BBVI) is a methodology for posterior approximation that relies on optimization.<n>Over the past decade, many new adaptive optimization algorithms have been developed that reduce or remove entirely the need for tuning.<n>We present a large-scale empirical evaluation of 56-based optimization algorithms applied to 1092 inference problems involving over 550,000 individual optimization runs and 15 core-years of compute.
  arXiv  Detail & Related papers  (2026-06-05T21:04:12Z)
• SparseEval: Efficient Evaluation of Large Language Models by Sparse Optimization [64.95852289011385]
  Large language models (LLMs) continue to scale up, their performance on various downstream tasks has significantly improved.<n> evaluating their capabilities has become increasingly expensive, as performing inference on a large number of benchmark samples incurs high computational costs.<n>We propose SparseEval, a method that, for the first time, adopts gradient descent to optimize anchor weights and employs an iterative refinement strategy for anchor selection.
  arXiv  Detail & Related papers  (2026-02-08T11:12:45Z)
• Bayesian Optimization in Language Space: An Eval-Efficient AI Self-Improvement Framework [0.0]
  Large Language Models (LLMs) have recently enabled self-improving AI, i.e., AI that iteratively generates, evaluates, and refines its own outcomes.<n>In many societal applications, the primary limitation is not generating new solutions but evaluating them.<n>This paper overcomes this challenge by proving that the combination of the simple and widely used Best-of-N selection strategy and simple textual gradients statistically emulates the behavior of the gradients on the canonical UCB acquisition function.
  arXiv  Detail & Related papers  (2025-11-15T07:04:44Z)
• On the Optimal Construction of Unbiased Gradient Estimators for Zeroth-Order Optimization [57.179679246370114]
  A potential limitation of existing methods is the bias inherent in most perturbation estimators unless a stepsize is proposed.<n>We propose a novel family of unbiased gradient scaling estimators that eliminate bias while maintaining favorable construction.
  arXiv  Detail & Related papers  (2025-10-22T18:25:43Z)
• Rethinking LLM Evaluation: Can We Evaluate LLMs with 200x Less Data? [82.09573568241724]
  EssenceBench is a coarse-to-fine framework utilizing an iterative Genetic Algorithm (GA)<n>Our approach yields superior compression results with lower reconstruction error and markedly higher efficiency.<n>On the HellaSwag benchmark (10K samples), our method preserves the ranking of all models shifting within 5% using 25x fewer samples, and achieves 95% ranking preservation shifting within 5% using only 200x fewer samples.
  arXiv  Detail & Related papers  (2025-10-12T05:38:10Z)
• BAPE: Learning an Explicit Bayes Classifier for Long-tailed Visual Recognition [78.70453964041718]
  Current deep learning algorithms usually solve for the optimal classifier by emphimplicitly estimating the posterior probabilities.<n>This simple methodology has been proven effective for meticulously balanced academic benchmark datasets.<n>However, it is not applicable to the long-tailed data distributions in the real world.<n>This paper presents a novel approach (BAPE) that provides a more precise theoretical estimation of the data distributions.
  arXiv  Detail & Related papers  (2025-06-29T15:12:50Z)
• Voronoi Candidates for Bayesian Optimization [2.499907423888049]
  Many practical BO methods, particularly in high dimension, eschew a formal, continuous optimization of the acquisition function.<n>We propose to use candidates which lie on the boundary of the Voronoi tessellation of the current design points, so they are equidistant to two or more of them.<n>We discuss strategies for efficient implementation by directly sampling the Voronoi boundary without explicitly generating the tessellation.
  arXiv  Detail & Related papers  (2024-02-07T14:47:13Z)
• Learning Regions of Interest for Bayesian Optimization with Adaptive Level-Set Estimation [84.0621253654014]
  We propose a framework, called BALLET, which adaptively filters for a high-confidence region of interest. We show theoretically that BALLET can efficiently shrink the search space, and can exhibit a tighter regret bound than standard BO.
  arXiv  Detail & Related papers  (2023-07-25T09:45:47Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.