Adaptive Blockwise Search: Inference-Time Alignment for Large Language Models

• URL: http://arxiv.org/abs/2510.23334v1
• Date: Mon, 27 Oct 2025 13:48:59 GMT
• Title: Adaptive Blockwise Search: Inference-Time Alignment for Large Language Models
• Authors: Mohammad Atif Quamar, Mohammad Areeb, Nishant Sharma, Ananth Shreekumar, Jonathan Rosenthal, Muslum Ozgur Ozmen, Mikhail Kuznetsov, Z. Berkay Celik,
• Abstract summary: In-time methods provide a flexible alternative to fine-tuning, but their uniform computational effort often yields suboptimal alignment.<n>We introduce AdaSearch, a novel blockwise search strategy.<n>It adaptively allocates a fixed computational budget using a sampling schedule, focusing search effort on critical tokens.
• Score: 13.368340836611075
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: LLM alignment remains a critical challenge. Inference-time methods provide a flexible alternative to fine-tuning, but their uniform computational effort often yields suboptimal alignment. We hypothesize that for many alignment tasks, the initial tokens of a response are disproportionately more critical. To leverage this principle, we introduce AdaSearch, a novel blockwise search strategy. It adaptively allocates a fixed computational budget using a sampling schedule, focusing search effort on these critical tokens. We apply AdaSearch to sequential decoding and introduce its tree-search counterpart, AdaBeam. Our comprehensive evaluation across eight LLMs demonstrates that AdaSearch outperforms strong Best-of-N and fine-tuning baselines. Specifically, win-rates improve by over 10% for harmlessness generation, controlled sentiment generation, and for mathematical reasoning tasks relative to Best-of-N.

Related papers

• $\ abla$-Reasoner: LLM Reasoning via Test-Time Gradient Descent in Latent Space [71.23672814629448]
  $nabla$-Reasoner is an iterative generation framework that integrates differentiable optimization over token logits into the decoding loop.<n>$nabla$-Reasoner achieves over 20% accuracy improvement on a challenging mathematical reasoning benchmark.
  arXiv  Detail & Related papers  (2026-03-05T08:42:54Z)
• READER: Retrieval-Assisted Drafter for Efficient LLM Inference [0.0386965802948046]
  Autoregressive Language Models instantiate a factorized likelihood over token sequences, yet their strictly sequential decoding process imposes an intrinsic lower bound on latency inference.<n>This bottleneck has emerged as a central obstacle to the scalable deployment of large-scale generative models.<n>We present READER, a speculative decoding framework that bypasses the training of the auxiliary draft model.
  arXiv  Detail & Related papers  (2025-08-12T16:47:48Z)
• LLM-First Search: Self-Guided Exploration of the Solution Space [29.780554400938335]
  Large Language Models (LLMs) have demonstrated remarkable improvements in reasoning and planning through increased test-time compute.<n>We propose textbfLLM-First Search (LFS), a novel textitLLM Self-Guided Search method.
  arXiv  Detail & Related papers  (2025-06-05T16:27:49Z)
• In-context Demonstration Matters: On Prompt Optimization for Pseudo-Supervision Refinement [71.60563181678323]
  Large language models (LLMs) have achieved great success across diverse tasks, and fine-tuning is sometimes needed to further enhance generation quality.<n>To handle these challenges, a direct solution is to generate high-confidence'' data from unsupervised downstream tasks.<n>We propose a novel approach, pseudo-supervised demonstrations aligned prompt optimization (PAPO) algorithm, which jointly refines both the prompt and the overall pseudo-supervision.
  arXiv  Detail & Related papers  (2024-10-04T03:39:28Z)
• Training Greedy Policy for Proposal Batch Selection in Expensive Multi-Objective Combinatorial Optimization [52.80408805368928]
  We introduce a novel greedy-style subset selection algorithm for batch acquisition. Our experiments on the red fluorescent proteins show that our proposed method achieves the baseline performance in 1.69x fewer queries.
  arXiv  Detail & Related papers  (2024-06-21T05:57:08Z)
• e-COP : Episodic Constrained Optimization of Policies [12.854752753529151]
  We present the first policy optimization algorithm for constrained Reinforcement Learning (RL) in episodic (finite horizon) settings.<n>We show that our algorithm has similar or better performance than SoTA (non-episodic) algorithms adapted for the episodic setting.
  arXiv  Detail & Related papers  (2024-06-13T20:12:09Z)
• Stop Relying on No-Choice and Do not Repeat the Moves: Optimal, Efficient and Practical Algorithms for Assortment Optimization [38.57171985309975]
  We develop efficient algorithms for the problem of regret in assortment selection with emphPlackett Luce (PL) based user choices. Our methods are practical, provably optimal, and devoid of the aforementioned limitations of the existing methods.
  arXiv  Detail & Related papers  (2024-02-29T07:17:04Z)
• Quality-Diversity Algorithms Can Provably Be Helpful for Optimization [24.694984679399315]
  Quality-Diversity (QD) algorithms aim to find a set of high-performing, yet diverse solutions. This paper tries to shed some light on the optimization ability of QD algorithms via rigorous running time analysis.
  arXiv  Detail & Related papers  (2024-01-19T07:40:24Z)
• Don't Search for a Search Method -- Simple Heuristics Suffice for Adversarial Text Attacks [11.196974000738729]
  We implement an algorithm inspired by zeroth order optimization-based attacks and compare with the benchmark results in the TextAttack framework. Surprisingly, we find that optimization-based methods do not yield any improvement in a constrained setup. We conclude from these results that current TextAttack benchmark tasks are too easy and constraints are too strict, preventing meaningful research on black-box adversarial text attacks.
  arXiv  Detail & Related papers  (2021-09-16T12:22:17Z)
• Machine Learning for Online Algorithm Selection under Censored Feedback [71.6879432974126]
  In online algorithm selection (OAS), instances of an algorithmic problem class are presented to an agent one after another, and the agent has to quickly select a presumably best algorithm from a fixed set of candidate algorithms. For decision problems such as satisfiability (SAT), quality typically refers to the algorithm's runtime. In this work, we revisit multi-armed bandit algorithms for OAS and discuss their capability of dealing with the problem. We adapt them towards runtime-oriented losses, allowing for partially censored data while keeping a space- and time-complexity independent of the time horizon.
  arXiv  Detail & Related papers  (2021-09-13T18:10:52Z)
• Towards Optimally Efficient Tree Search with Deep Learning [76.64632985696237]
  This paper investigates the classical integer least-squares problem which estimates signals integer from linear models. The problem is NP-hard and often arises in diverse applications such as signal processing, bioinformatics, communications and machine learning. We propose a general hyper-accelerated tree search (HATS) algorithm by employing a deep neural network to estimate the optimal estimation for the underlying simplified memory-bounded A* algorithm.
  arXiv  Detail & Related papers  (2021-01-07T08:00:02Z)
• Adaptive Sampling for Best Policy Identification in Markov Decision Processes [79.4957965474334]
  We investigate the problem of best-policy identification in discounted Markov Decision (MDPs) when the learner has access to a generative model. The advantages of state-of-the-art algorithms are discussed and illustrated.
  arXiv  Detail & Related papers  (2020-09-28T15:22:24Z)
• Best-First Beam Search [78.71330480725668]
  We show that the standard implementation of beam search can be made up to 10x faster in practice. We propose a memory-reduced variant of Best-First Beam Search, which has a similar beneficial search bias in terms of downstream performance.
  arXiv  Detail & Related papers  (2020-07-08T05:56:01Z)

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.